Autoencoder & Dialog Systems

Notes:

An autoencoder is a type of artificial neural network that is used to learn a compressed representation of a set of data. Autoencoders are often used for feature extraction, which is the process of identifying and extracting the most relevant and informative features of a dataset.

Autoencoders are composed of two main components: an encoder and a decoder. The encoder is responsible for learning a compressed representation of the data, while the decoder is responsible for reconstructing the original data from the compressed representation.

To learn the compressed representation, the autoencoder is trained on a dataset by minimizing the difference between the original data and the reconstructed data. This process forces the autoencoder to identify and preserve the most important features of the data, while discarding less relevant or noisy features.

Autoencoders are often used for tasks such as dimensionality reduction, data denoising, and anomaly detection, as they can effectively identify and extract the most important features of a dataset. They are also widely used in a variety of other applications, including image and video processing, natural language processing, and machine learning.

Autoencoders can be used in dialog systems to learn a compressed representation of the dialog data, which can be useful for tasks such as feature extraction and dimensionality reduction. By learning a compressed representation of the data, autoencoders can identify and extract the most important features of the dialog, which can be useful for understanding and responding to user input.

In a dialog system, autoencoders might be used to learn a compressed representation of the conversation history, user preferences, or other relevant information. This compressed representation can then be used by the dialog system to generate more accurate and relevant responses to user input.

For example, an autoencoder might be used to learn a compressed representation of a user’s conversation history, which can include information about the user’s preferences, interests, and past interactions with the system. The autoencoder could then be used to extract the most important features of this conversation history, which could be used by the dialog system to generate more personalized and relevant responses to the user.

Between 2016 and 2018 there was a huge increase in the co-occurrence of autoencoder and dialog systems in academic papers, doubling each year.

Resources:

dblp.org/db/conf/nldb .. annual conference on applications of natural language to data bases
kueri.me .. search-box interface to allow users to navigate, explore, and present data

Wikipedia:

An auto-encoder matching model for learning utterance-level semantic dependency in dialogue generation
L Luo, J Xu, J Lin, Q Zeng, X Sun – arXiv preprint arXiv:1808.08795, 2018 – arxiv.org
… In this work, we propose an Auto-Encoder Match- ing model to learn the utterance-level semantic de … Autoencoders, unsupervised learn- ing, and deep architectures … Autoencoder as assistant supervisor: Improving text representation for chinese social me- dia text summarization …

Probabilistic natural language generation with wasserstein autoencoders
H Bahuleyan, L Mou, K Vamaraju, H Zhou… – arXiv preprint arXiv …, 2018 – arxiv.org
… This section describes the autoencoding experi- ments carried out with the three models … In this paper, we propose the use of Wasserstein autoencoders (WAEs) for probabilistic … WAE is com- pared to existing generation models, specifically the variational autoencoder (VAE) and …

Variational cross-domain natural language generation for spoken dialogue systems
BH Tseng, F Kreyssig, P Budzianowski… – arXiv preprint arXiv …, 2018 – arxiv.org
… autoencoder ar- chitecture. We demonstrate that our model outperforms the original RNN-based gen- erator, while yielding highly diverse sen- tences. In addition, our model performs better when the training data is limited. 1 Introduction Conventional spoken dialogue systems …

Dialogue generation with GAN
H Su, X Shen, P Hu, W Li, Y Chen – Thirty-Second AAAI Conference on …, 2018 – aaai.org
… Makhzani, A.; Shlens, J.; Jaitly, N.; and Goodfellow, IJ 2015. Adversarial autoencoders. CoRR. Serban, IV; Sordoni, A.; Bengio, Y.; Courville, AC; and Pineau, J. 2016. Building end-to-end dialogue systems using generative hierarchical neural network models …

A knowledge-grounded neural conversation model
M Ghazvininejad, C Brockett, MW Chang… – Thirty-Second AAAI …, 2018 – aaai.org
… 2017) employ a knowl- edge graph to embed side information into dialog systems … Finally, one interesting option is to replace the response in task (2) with one of the facts (R = fi), which makes task (2) similar to an autoencoder and helps produce responses that are even more …

Two can play this game: visual dialog with discriminative question generation and answering
U Jain, S Lazebnik, AG Schwing – Proceedings of the IEEE …, 2018 – openaccess.thecvf.com
… In contrast to the two aforementioned techniques, Jain et al. [15] argued for more diverse predictions and employed a variational auto-encoder approach. Work by Li et al … Hence, we focus on discriminative visual dialog systems …

Dual Latent Variable Model for Low-Resource Natural Language Generation in Dialogue Systems
VK Tran, LM Nguyen – arXiv preprint arXiv:1811.04164, 2018 – arxiv.org
… Recently, the RNN-based generators have shown improving results in tackling the NLG problems in task oriented-dialogue systems with varied pro- posed methods, such as HLSTM (Wen et al., 2015a), SCLSTM … (2015) presented a variational autoencoder for unsupervised …

Generating Classical Chinese Poems via Conditional Variational Autoencoder and Adversarial Training
J Li, Y Song, H Zhang, D Chen, S Shi, D Zhao… – Proceedings of the 2018 …, 2018 – aclweb.org
… More- over, boosting autoencoder with variational infer- ence (Kingma and Welling, 2014), known as vari- ational autoencoder (VAE), can generate not … Similar to that proposed for dialogue systems (Li et al., 2016), this evaluation is employed to measure character diversity by …

Improving Variational Encoder-Decoders in Dialogue Generation
X Shen, H Su, S Niu, V Demberg – arXiv preprint arXiv:1802.02032, 2018 – arxiv.org
… Vari- ational autoencoders (VAEs) (Kingma and Welling 2014; Rezende, Mohamed, and Wierstra 2014) bring … The training alternates between the autoencoder (AE) phase to optimize ?Eq?(z|c,x)p? … In the AE phase, they should autoencode utterenaces to make the real posterior …

Adversarial Domain Adaptation for Variational Neural Language Generation in Dialogue Systems
VK Tran, LM Nguyen – arXiv preprint arXiv:1808.02586, 2018 – arxiv.org
… More recently, the development of the variational autoencoder (VAE) framework (Kingma and Welling, 2013; Rezende and Mohamed, 2015) has paved the way … Bowman et al., 2015; Miao et al., 2016; Purushotham et al., 2017; Mnih and Gregor, 2014), dialogue system (Wen et …

Style transfer in text: Exploration and evaluation
Z Fu, X Tan, N Peng, D Zhao, R Yan – Thirty-Second AAAI Conference on …, 2018 – aaai.org
… models are able to generate sentences with similar content preservation score but higher style transfer strength comparing to auto- encoder … and Le 2014) have demonstrated great success in many generation tasks, such as machine transla- tion, dialog system and image …

Deep Learning in Natural Language Processing
L Deng, Y Liu – 2018 – books.google.com
… analysis Point-wise mutual information Part of speech Paragraph vector Question answering Recursive autoencoder Restricted Boltzmann … for gisting evaluation Referenced metric and unreferenced metric blended evaluation routine Spoken dialog system Spoken language …

Speech recognition in a dialog system: from conventional to deep processing
A Becerra, JI de la Rosa, E González – Multimedia Tools and Applications, 2018 – Springer
… process can be performed by some strategies like discriminative pre-training (DPT) [73], DBN pre-training (deep belief networks) [9, 10, 31], denoising autoencoder [4, 93], hybrid pre-training [68] and dropout [33, 93, 97]. An essential element in a spoken dialog system is the …

Importance of a Search Strategy in Neural Dialogue Modelling
I Kulikov, AH Miller, K Cho, J Weston – arXiv preprint arXiv:1811.00907, 2018 – arxiv.org
… Serban et al. (2017) introduce la- tent variables to their earlier hierarchical model and train it to maximize the variational lower bound, similar to Zhao et al. (2017) who propose to build a neural dialogue model as a conditional variational autoencoder. Xu et al …

Hierarchical Variational Memory Network for Dialogue Generation
H Chen, Z Ren, J Tang, YE Zhao, D Yin – … of the 2018 World Wide Web …, 2018 – dl.acm.org
… exist in current neural mod- els for dialogue generation [6]: (1) Meaningless responses: Given a wide range of contexts, dialogue systems trained via … It combines the spirits of variational autoencoder [18] and memory networks [39, 40] in a hierarchical recurrent neural network set …

Variational memory encoder-decoder
H Le, T Tran, T Nguyen, S Venkatesh – Advances in Neural …, 2018 – papers.nips.cc
… Standard neural encoder-decoder models and their extensions using conditional variational autoencoder often result in either trivial or digressive responses … 2.2 Conditional Variational Autoencoder (CVAE) for Conversation Generation …

Elastic responding machine for dialog generation with dynamically mechanism selecting
G Zhou, P Luo, Y Xiao, F Lin, B Chen… – AAAI Conference on …, 2018 – researchgate.net
Page 1. Elastic Responding Machine for Dialog Generation with Dynamically Mechanism Selecting Ganbin Zhou1,2, Ping Luo1,2, Yijun Xiao3, Fen Lin4, Bo Chen4, Qing He1,2 1Key Lab of Intelligent Information Processing …

Aiming to Know You Better Perhaps Makes Me a More Engaging Dialogue Partner
Y Zemlyanskiy, F Sha – arXiv preprint arXiv:1808.07104, 2018 – arxiv.org
… Similarly, (Zhao et al., 2017; Shen et al., 2018; Gu et al., 2018) use variants of variational autoencoder … for automatic evaluation (so it can be used to optimize a dialogue model to be more engaging, at least in principle) is as hard as cre- ating human-like dialogue system itself …

Chitty-Chitty-Chat Bot: Deep Learning for Conversational AI.
R Yan – IJCAI, 2018 – ijcai.org
… Conditional Variational Auto-Encoder (CVAE) can capture the discourse-level diversity during en- coding [Zhao et al., 2017] … A hierarchical neural autoencoder for paragraphs and documents … End-to-end task- completion neural dialogue systems …

Texar: A modularized, versatile, and extensible toolkit for text generation
Z Hu, H Shi, Z Yang, B Tan, T Zhao, J He… – arXiv preprint arXiv …, 2018 – arxiv.org
… Such tasks include machine translation [2, 8], dialog systems [49, 39], text summarization [16, 37], article writing [50, 27], text paraphrasing … We deploy the self-attention Transformer decoder on two tasks, namely, variational autoencoder (VAE) based language modeling [5] and …

Unsupervised Multilingual Natural Language Generation with Denoising Autoencoders
M Freitag, S Roy – 2018 – ai.google
… Unsupervised Multilingual Natural Language Generation with Denoising Autoencoders … structured data is important for various tasks such as question answering and dialog systems … a corrupt representation of the desired output and use a denoising auto-encoder to reconstruct …

Improving Robustness of Neural Dialog Systems in a Data-Efficient Way with Turn Dropout
I Shalyminov, S Lee – arXiv preprint arXiv:1811.12148, 2018 – arxiv.org
… our knowledge, is presented here for the first time — uses a Variational Autoencoder as the … differs from the previous two in that it learns dialog control and autoencoding jointly … Learning discourse-level diversity for neural dialog models using conditional variational autoencoders …

Attention-Based Dialog State Tracking for Conversational Interview Coaching
MH Su, CH Wu, KY Huang… – 2018 IEEE International …, 2018 – ieeexplore.ieee.org
… Finally, an LSTM- based autoencoder is adopted to model the transition and … CH Wu, MH Su, and WB Liang, “Miscommunication handling in spoken dialog systems based on … Unit-Selection Synthesizer using Sequence-To-Sequence LSTM-based Autoencoders,” in Interspeech …

Unsupervised Natural Language Generation with Denoising Autoencoders
M Freitag, S Roy – arXiv preprint arXiv:1804.07899, 2018 – arxiv.org
Unsupervised Natural Language Generation with Denoising Autoencoders … from structured data is impor- tant for various tasks such as question answer- ing and dialog systems … as a corrupt represen- tation of the desired output and use a denois- ing auto-encoder to reconstruct …

DialogWAE: Multimodal Response Generation with Conditional Wasserstein Auto-Encoder
X Gu, K Cho, J Ha, S Kim – arXiv preprint arXiv:1805.12352, 2018 – arxiv.org
… VAE conversation models The variational autoencoder (VAE) [Kingma and Welling, 2014] is among the most popular frameworks for … Inspired by GAN and the adversarial auto-encoder (AAE) [Makhzani et al., 2015; Tolstikhin et al., 2017], we model the … Adversarial autoencoders …

Auto-Dialabel: Labeling Dialogue Data with Unsupervised Learning
C Shi, Q Chen, L Sha, S Li, X Sun, H Wang… – Proceedings of the 2018 …, 2018 – aclweb.org
… In this framework, we collect a set of context features, leverage an autoencoder for feature assembly, and adapt a dynamic hierarchical clustering method for intent and slot labeling … 1 Introduction Building a task-oriented dialogue system is chal- lenging …

Semi-supervised learning for information extraction from dialogue
A Kannan, K Chen, D Jaunzeikare… – Proc. Interspeech …, 2018 – isca-speech.org
… pro- pose a variation that predicts nearby turns and observe that it outperforms the autoencoder … prediction to train “turn embeddings” as part of an end-to-end dialogue system, which is … 2) we analyze our method under various conditions and compare it directly to autoencoders …

Generating lyrics with variational autoencoder and multi-modal artist embeddings
O Vechtomova, H Bahuleyan, A Ghabussi… – arXiv preprint arXiv …, 2018 – arxiv.org
… We use the variational autoencoder (VAE) [1] with Long Short Term Memory networks … text conditioned on sentiment [5, 6] and persona-conditioned responses in dialogue systems [7]. To … other models for pre-training of artist embeddings, for example spectrogram autoencoders …

Unsupervised Discrete Sentence Representation Learning for Interpretable Neural Dialog Generation
T Zhao, K Lee, M Eskenazi – arXiv preprint arXiv:1804.08069, 2018 – arxiv.org
… Discrete latent variables have also been used for task-oriented dialog systems (Wen et al., 2017), where … 4 is similar to the objec- tives used in adversarial autoencoders (Makhzani et al., 2015; Kim … term from DI-VAE and DI-VST leads to a simple discrete autoencoder (DAE) and …

Autoencoder for Semisupervised Multiple Emotion Detection of Conversation Transcripts
DA Phan, Y Matsumoto… – IEEE Transactions on …, 2018 – ieeexplore.ieee.org
… representations. A deep-learning auto-encoder is then used to discover the underlying structure of the unsupervised data … annotators. Index Terms—Emotion recognition, semisupervised learning, multilabel, word2vec, autoencoder …

Another Diversity-Promoting Objective Function for Neural Dialogue Generation
R Nakamura, K Sudoh, K Yoshino… – arXiv preprint arXiv …, 2018 – arxiv.org
… Variational AutoEncoder (VAE) was also proposed in image genera- tion (Kingma and Welling … Semantically conditioned lstm-based natural language generation for spoken dialogue systems … level diversity for neural dialog models using con- ditional variational autoencoders …

Policy learning for task-oriented dialogue systems via reinforcement learning techniques
C Yin – 2018 – minerva-access.unimelb.edu.au
… To overcome this problem, Henderson et al. [28] also apply denoising autoencoder (DA) [83] to warm up the RNN model … In this way, their model dramatically outperforms baselines. DST is a very important component in task-oriented dialogue systems because subsequent …

Changing the Level of Directness in Dialogue using Dialogue Vector Models and Recurrent Neural Networks
L Pragst, S Ultes – Proceedings of the 19th Annual SIGdial Meeting on …, 2018 – aclweb.org
… Therefore, dialogue systems that can adjust the level of directness in their output to the user and … Shen et al., 2014; Kalchbrenner et al., 2014; Hu et al., 2014) and autoencoders (Socher … The mapping of ut- terances to their vector representations is trained akin to autoencoding …

NEXUS Network: Connecting the Preceding and the Following in Dialogue Generation
X Shen, H Su, W Li, D Klakow – Proceedings of the 2018 Conference on …, 2018 – aclweb.org
… Abstract Sequence-to-Sequence (seq2seq) models have become overwhelmingly popular in build- ing end-to-end trainable dialogue systems … Conditional Variational Autoencoder The idea of learning an appropriate prior distribution in Eq …

NEXUS Network: Connecting the Preceding and the Following in Dialogue Generation
H Su, X Shen, W Li, D Klakow – arXiv preprint arXiv:1810.00671, 2018 – arxiv.org
… Abstract Sequence-to-Sequence (seq2seq) models have become overwhelmingly popular in build- ing end-to-end trainable dialogue systems … Conditional Variational Autoencoder The idea of learning an appropriate prior distribution in Eq …

Modeling Non-Goal Oriented Dialog With Discrete Attributes
C Sankar, S Ravi – alborz-geramifard.com
… [43] B. Wei, S. Lu, L. Mou, H. Zhou, P. Poupart, G. Li, and Z. Jin. Why Do Neural Dialog Systems Generate Short and Meaningless Replies … Learning Discourse-level Diversity for Neural Dialog Models using Conditional Variational Autoencoders. ArXiv e-prints, March 2017 …

Out-of-domain Detection based on Generative Adversarial Network
S Ryu, S Koo, H Yu, GG Lee – Proceedings of the 2018 Conference on …, 2018 – aclweb.org
… Multi–domain dialog systems (Hakkani-Tur et al., 2016; Jiang et al., 2014; Lee et al., 2013; Ryu et al., 2015; Seon et al … This re- sult means that the reconstruction errors by the ideal autoencoder are not reliable evidence of OOD, so in OOD detection, autoencoders have lit- tle …

Question Rewrite Based Dialogue Response Generation
H Liu, W Rong, L Shi, Y Ouyang, Z Xiong – International Conference on …, 2018 – Springer
… popular generative models such as deep belief nets (DBN) [6], variational autoencoder (VAE) [10 … H., Liu, X., Yin, D., Tang, J.: A survey on dialogue systems: recent advances … T., Józefowicz, R., Chen, X., Sutskever, I., Welling, M.: Improving variational autoencoders with inverse …

Multi-task learning using variational auto-encoder for sentiment classification
G Lu, X Zhao, J Yin, W Yang, B Li – Pattern Recognition Letters, 2018 – Elsevier
… Pattern Recognition Letters. Multi-task learning using variational auto-encoder for sentiment classification. Author links open overlay panelGuangquanLu … Multi-Layer Perceptron and Multi-task learning. 2.1. Variational auto-encoder …

Learning Dialogue History for Spoken Language Understanding
X Zhang, D Ma, H Wang – … on Natural Language Processing and Chinese …, 2018 – Springer
… extended HRED by introducing Variational Autoencoder (VAE) and proposed VHRED model [20]. VHRED can generate more meaningful and diverse responses. HRED and VHRED are both used in conversational dialogue systems, while our propose HLSTM-SLU is designed …

Learning to Converse Emotionally Like Humans: A Conditional Variational Approach
R Zhang, Z Wang – CCF International Conference on Natural Language …, 2018 – Springer
… S., Severyn, A., Barth, E.: A hybrid convolutional variational autoencoder for text … The good, the bad and the neutral: affective profile in dialog system-user communication … M.: Learning discourse-level diversity for neural dialog models using conditional variational autoencoders …

Unsupervised Learning of Interpretable Dialog Models
D Madan, D Raghu, G Pandey, S Joshi – arXiv preprint arXiv:1811.01012, 2018 – arxiv.org
… Interpretable Dialog Models Traditional task oriented dialog systems were built using reinforcement learning (RL) approaches … Later (van den Oord, Vinyals, and others 2017) introduced the Vector Quantized-Variational AutoEncoder …

Zero-Shot Dialog Generation with Cross-Domain Latent Actions
T Zhao, M Eskenazi – arXiv preprint arXiv:1805.04803, 2018 – arxiv.org
… Abstract This paper introduces zero-shot dialog generation (ZSDG), as a step towards neu- ral dialog systems that can instantly gener- alize to new situations with minimal data … Finally, ZSL has been applied to individual components in the dialog system pipeline. Chen et al …

Two-Step Training and Mixed Encoding-Decoding for Implementing a Generative Chatbot with a Small Dialogue Corpus
J Kim, HG Lee, H Kim, Y Lee, YG Kim – Proceedings of the Workshop on …, 2018 – aclweb.org
… To realize this assumption, we adopt an autoencoder mechanism … How not to evalu- ate your dialogue system: An empirical study of unsupervised evaluation metrics for dialogue re- sponse generation. arXiv preprint arXiv:1603.08023 …

Adversarial Text Generation Without Reinforcement Learning
D Donahue, A Rumshisky – arXiv preprint arXiv:1810.06640, 2018 – arxiv.org
… Plotted embeddings indi- cate that LaTextGAN has properly modeled the la- tent space of a trained autoencoder, while sentence interpolations show that … Adversar- ial autoencoders … Building end-to-end dialogue systems using gener- ative hierarchical neural network models …

Towards Deep Conversational Recommendations
R Li, SE Kahou, H Schulz, V Michalski… – Advances in Neural …, 2018 – papers.nips.cc
… recommendations (ie in cold-start setting) we train this model as a denoising auto-encoder [24] … architectures, extending them with dynam- ically instantiated RNN models that drive an autoencoder-based recommendation … Autorec: Autoencoders meet collaborative filtering …

T2S: An Encoder-Decoder Model for Topic-Based Natural Language Generation
W Ou, C Chen, J Ren – International Conference on Applications of Natural …, 2018 – Springer
… critical role in various applications such as response generation in dialogue systems [17, 20 … Unit [4]. Specifically, general sentence generation usually employs sequence auto-encoder architecture [2 … Li, J., Luong, M., Jurafsky, D.: A hierarchical neural autoencoder for paragraphs …

Dialogue-act-driven Conversation Model: An Experimental Study
H Kumar, A Agarwal, S Joshi – … of the 27th International Conference on …, 2018 – aclweb.org
… 2016b. Building end-to-end dialogue systems using generative hierarchical neural network models. In AAAI … Tiancheng Zhao, Ran Zhao, and Maxine Eskenazi. 2017. Learning discourse-level diversity for neural dialog models using conditional variational autoencoders …

Deep Learning in Natural
L Deng, Y Liu – Springer
… mutual information POS Part of speech PV Paragraph vector QA Question answering RAE Recursive autoencoder RBM Restricted … for gisting evaluation RUBER Referenced metric and unreferenced metric blended evaluation routine SDS Spoken dialog system SLU Spoken …

Skeleton-to-Response: Dialogue Generation Guided by Retrieval Memory
D Cai, Y Wang, V Bi, Z Tu, X Liu, W Lam… – arXiv preprint arXiv …, 2018 – arxiv.org
… Introduction This paper focuses on tackling the challenges to develop a chit-chat style dialogue system (also known as chatbot). Chi- chat style dialogue system aims at giving meaningful and coherent responses given a dialogue query in open domain …

Interpreting Decision-Making in Interactive Visual Dialogue
U Sharma – 2018 – esc.fnwi.uva.nl
… Page 3. Abstract Dialogue systems that involve long-term planning can strongly benefit from a high-level notion of dialogue strategy and can avoid making poor decisions early in the game and opt for broadly successful strategies instead …

Prior Attention for Style-aware Sequence-to-Sequence Models
L Sterckx, J Deleu, C Develder, T Demeester – arXiv preprint arXiv …, 2018 – arxiv.org
… initial attention- based sequence-to-sequence model, we use a variational auto-encoder conditioned on … 2017), automated lyric annotation (Sterckx et al., 2017) and dialogue systems (Serban et al … attention) on a parallel text corpus, a conditional variational autoencoder (Sohn …

Generation of Company descriptions using concept-to-text and text-to-text deep models: dataset collection and systems evaluation
R Qader, K Jneid, F Portet, C Labbé – Proceedings of the 11th …, 2018 – aclweb.org
… cepts) using a dataset collected from Wikipedia. Similarly, in (Chisholm et al., 2017) a sequence- to-sequence autoencoder with attention was used to generate biographies from Wikipedia. How- ever, both of these work concentrate …

Context-Sensitive Generation of Open-Domain Conversational Responses
W Zhang, Y Cui, Y Wang, Q Zhu, L Li, L Zhou… – Proceedings of the 27th …, 2018 – aclweb.org
… As the advantages of generative adversarial net (GAN) and variational autoencoder (VAE), Yu et al … (2017) presented an end-to-end dialogue system for information accquisition, which is called KB-InfoBot from knowledge base (KB) by using reinforcement learning …

User Modeling for Task Oriented Dialogues
I Gur, D Hakkani-Tur, G Tur, P Shah – arXiv preprint arXiv:1811.04369, 2018 – arxiv.org
… Our user simulators aim to enable side-by-side comparisons of dialogue systems or their components across a given set of user … Variational approaches such as Vari- ational Autoencoders are used for unsupervised or semi-supervised model training [14, 15] that improves the …

Artificial Intelligence for Decision Support in Command and Control Systems
J Schubert, J Brynielsson, M Nilsson, P Svenmarck – researchgate.net
… of-the-art methods for anomaly detection are currently using deep autoencoders as a … 2 An autoencoder is a special artificial neural network architecture with an encoder part … all have products leveraging the latest deep learning technologies for speech-based dialog systems …

Dialogue act recognition via crf-attentive structured network
Z Chen, R Yang, Z Zhao, D Cai, X He – The 41st International ACM SIGIR …, 2018 – dl.acm.org
… task. Many applications have benefited from the use of au- tomatic dialogue act recognition such as dialogue systems, machine translation, automatic speech recognition, topic identification and talking avatars [20] [14]. One …

Deep Bayesian Learning and Understanding
JT Chien – Proceedings of the 27th International Conference on …, 2018 – aclweb.org
… network (CNN) – Dilated recurrent neural network – Generative adversarial network (GAN) – Variational auto-encoder (VAE … window to more practical tasks, eg reading comprehension, sentence generation, dialogue system, question answering … Ladder variational autoencoders …

S2spmn: a simple and effective framework for response generation with relevant information
J Pei, C Li – Proceedings of the 2018 Conference on Empirical …, 2018 – aclweb.org
… 2017. A network-based end-to-end trainable task-oriented dialogue system. In EACL … Tiancheng Zhao, Ran Zhao, and Maxine Eskenazi. Learning discourse-level diversity for neural dialog models using conditional variational autoencoders. In ACL, pages 654–664 …

Char2char generation with reranking for the E2E NLG Challenge
S Agarwal, M Dymetman, E Gaussier – arXiv preprint arXiv:1811.05826, 2018 – arxiv.org
… Traditionally, the Nat- ural Language Generation (NLG) component in Spoken Dialogue Systems has been rule-based, involving a two stage pipeline: ‘sentence … trained a reverse model which tries to reconstruct the MR given the target RF, similar to the autoencoder model by …

Scalable Sentiment for Sequence-to-sequence Chatbot Response with Performance Analysis
CW Lee, YS Wang, TY Hsu, KY Chen, HY Lee… – arXiv preprint arXiv …, 2018 – arxiv.org
… Terms— chatbot, dialogue, sequence-to-sequence, style- transfer, response generation 1. INTRODUCTION Unlike goal-oriented dialogue systems [1, 2], a … VRAE denotes variational recurrent auto-encoder We borrow the concept of plug and play previously used to gen- erate …

Overview of the NLPCC 2018 Shared Task: Multi-turn Human-Computer Conversations
J Li, R Yan – CCF International Conference on Natural Language …, 2018 – Springer
… Google Scholar. 5. Serban, IV, Sordoni, A., Bengio, Y., Courville, AC, Pineau, J.: Building end-to-end dialogue systems using generative … 9. Zhao, T., Zhao, R., Eskenazi, M.: Learning discourse-level diversity for neural dialog models using conditional variational autoencoders …

A Comparative Study of Statistical Conversion of Face to Voice Based on Their Subjective Impressions
Y Ohsugi, D Saito, N Minematsu – Proc. Interspeech 2018, 2018 – isca-speech.org
… 5] S. Miyake and A. Ito, “A spoken dialogue system using virtual … Yeh, Z. Liu, DB Goldman, and A. Agarwala, “Semantic Fa- cial Expression Editing using Autoencoded Flow,” ArXiv … 16] X. Hou, L. Shen, K. Sun, and G. Qiu, “Deep feature consistent variational autoencoder,” in 2017 …

When Less Is More: Using Less Context Information to Generate Better Utterances in Group Conversations
H Zhang, Z Chan, Y Song, D Zhao, R Yan – CCF International Conference …, 2018 – Springer
… 994–1003 (2016)Google Scholar. 6. Li, J., Luong, T., Jurafsky, D.: A hierarchical neural autoencoder for paragraphs and documents … 1106–1115 (2015)Google Scholar. 7. Liu, CW, Lowe, R., et al.: How not to evaluate your dialogue system: an empirical study of unsupervised …

Inferring User Emotive State Changes in Realistic Human-Computer Conversational Dialogs
R Li, Z Wu, J Jia, J Li, W Chen, H Meng – 2018 ACM Multimedia …, 2018 – dl.acm.org
… state changes prediction, multi-modal multi-task deep learning, convolution fusion auto- encoder, recurrent auto-encoder, structured output … 1 INTRODUCTION Automatic spoken dialog systems, known as interactive speech agents, receive speech as input and response via …

Photo-Realistic Blocksworld Dataset
M Asai – arXiv preprint arXiv:1812.01818, 2018 – arxiv.org
… It is still used as a testbed for evaluating various computer vision / dialog systems (Perera, Allen, Teng, & Galescu, 2018; Bisk, Shih … system learns the binary latent space of an arbitrary raw input (eg images) with a Gumbel-Softmax variational autoencoder (State AutoEncoder …

SocialNLP 2018 EmotionX Challenge Overview: Recognizing Emotions in Dialogues
CC Hsu, LW Ku – Proceedings of the Sixth International Workshop on …, 2018 – aclweb.org
… Still, textual emotion recognition needs further exploration in dialogue systems for many reasons … The joint training of the clas- sifer and the autoencoder improves generalizabil- ity, and linguistic features boost the performance on the minority class …

A Retrieve-and-Edit Framework for Predicting Structured Outputs
TB Hashimoto, K Guu, Y Oren… – Advances in Neural …, 2018 – papers.nips.cc
… Semantic hashing by autoencoders [16] is a related idea where an autoencoder’s latent representation is used to construct a hash function to identify similar images or texts [29, 6]. A related idea is cross-modal embeddings, which jointly embed and align items in different …

Towards a neural conversation model with diversity net using determinantal point processes
Y Song, R Yan, Y Feng, Y Zhang, D Zhao… – Thirty-Second AAAI …, 2018 – aaai.org
Page 1. Towards a Neural Conversation Model with Diversity Net Using Determinantal Point Processes Yiping Song, 1 Rui Yan, 2,3? Yansong Feng, 2 Yaoyuan Zhang, 2 Dongyan Zhao, 2,3 Ming Zhang 1? 1 Institute of Network …

Visual questioning agents
U Jain – 2018 – ideals.illinois.edu
… generative models. More concretely, we follow the variational autoencoder [9] paradigm rather … publications: 1. Unnat Jain*, Ziyu Zhang*, Alexander Schwing, “Creativity: Generating Diverse Questions using Variational Autoencoders”, CVPR 2017 [1] …

Variational Autoregressive Decoder for Neural Response Generation
J Du, W Li, Y He, R Xu, L Bing, X Wang – Proceedings of the 2018 …, 2018 – aclweb.org
… Abstract Combining the virtues of probability graphic models and neural networks, Conditional Variational Auto-encoder (CVAE) has shown promising performance in many applications such as response generation … 3.1 Conditional Variational Auto-Encoder …

Texar: A modularized, versatile, and extensible toolbox for text generation
Z Hu, Z Yang, T Zhao, H Shi, J He, D Wang… – … of Workshop for NLP …, 2018 – aclweb.org
… Such tasks include machine transla- tion (Bahdanau et al., 2014; Brown et al., 1990), dialog systems (Williams and Young, 2007; Ser- ban et al., 2016), text summarization (Hovy and Lin, 1998; See et al., 2017), article writing (Wise- man et al., 2017), text paraphrasing and …

Anomaly detection for short texts: Identifying whether your chatbot should switch from goal-oriented conversation to chit-chatting
A Bakarov, V Yadrintsev, I Sochenkov – International Conference on Digital …, 2018 – Springer
… Conversational agents (also called dialog systems) are systems that are able to converse with a human on a natural language, imitating dialogue with a … The idea is to train an autoencoder [20] on training data and for each new object compute a reconstruction error between …

Incorporating Memory into Deep Generative Dialogue Models using a Scalable Attention Mechanism
KA Selby – pdfs.semanticscholar.org
Page 1. Incorporating Memory into Deep Generative Dialogue Models using a Scalable Attention Mechanism by Kira A. Selby A thesis presented to the University of Waterloo in fulfillment of the thesis requirement for the degree …

Efficient purely convolutional text encoding
S Malik, A Lancucki, J Chorowski – arXiv preprint arXiv:1808.01160, 2018 – arxiv.org
… Apart from this straight- forward application in dialogue systems, sentence embed- dings are … Our model builds on the Byte-Level Recursive Convolutional Auto-Encoder [Zhang and … models is available: https://github.com/smalik169/ recursive-convolutional-autoencoder 2https …

Generating sentences by editing prototypes
K Guu, TB Hashimoto, Y Oren, P Liang – Transactions of the Association …, 2018 – MIT Press
… A thorough introduction to the ELBO is provided in Doersch (2016). Note that q(z | x ,x) and pedit(x | x ,z) combine to form a variational autoencoder (VAE) (Kingma and Welling, 2014), where q(z | x ,x) is the varia- tional encoder and pedit(x | x ,z) is the variational decoder …

Spoken language understanding without speech recognition
YP Chen, R Price, S Bangalore – 2018 IEEE International …, 2018 – ieeexplore.ieee.org
… We evaluate this approach in the context of a customer care dialog system and demonstrate its effec- tiveness in comparison to the conventional … based CNN-RNN lan- guage model is trained on a text corpus for encoding text queries and an RNN autoencoder is trained with …

Recent trends in deep learning based natural language processing
T Young, D Hazarika, S Poria… – ieee Computational …, 2018 – ieeexplore.ieee.org
… NLP enables computers to perform a wide range of natural language related tasks at all levels, ranging from parsing and part- of-speech (POS) tagging, to machine translation and dialogue systems … [13] used embeddings along with stacked denoising autoencoders for domain …

Teaching Machines to Ask Questions.
K Yao, L Zhang, T Luo, L Tao, Y Wu – IJCAI, 2018 – ijcai.org
… popular framework has been applied in text generation tasks such as dialogue system [Zhao et al … It is based on condi- tional variational autoencoders [Zhao et al., 2017] that cap- tures … At train- ing time, we follow the variational autoencoder framework [Kingma and Welling, 2013 …

LSTM based cross-corpus and cross-task acoustic emotion recognition
H Kaya, D Fedotov, A Ye?ilkanat… – Proc. Interspeech …, 2018 – researchgate.net
… multi-view approaches [4] to denoising auto-encoder based do- main adaptation schemes [5, 6 … [5] J. Deng, Z. Zhang, F. Eyben, and B. Schuller, “Autoencoder-based unsupervised … J. Deng, X. Xu, Z. Zhang, S. Frühholz, and B. Schuller, “Semisupervised autoencoders for speech …

Automatic Evaluation of Neural Personality-based Chatbots
Y Xing, R Fernández – arXiv preprint arXiv:1810.00472, 2018 – arxiv.org
… (2017) use multi-task learning to incorporate an autoencoder that learns the speaker’s language style from non- conversational data such … 2016. How not to evaluate your dialogue system: An em- pirical study of unsupervised evaluation metrics for dialogue response generation …

Language coverage and generalization in RNN-based continuous sentence embeddings for interacting agents
L Celotti, S Brodeur, J Rouat – nips2018vigil.github.io
… used for generative modeling of sentences [17] using sequence-to-sequence autoencoding (AE) [14 … fixed length con- tinuous spaces, such as sequence to sequence autoencoders [14] and … recurrent neural networks (RNNs), known as a sequence to sequence autoencoder [14] …

Multitask learning for neural generative question answering
Y Huang, T Zhong – Machine Vision and Applications, 2018 – Springer
Page 1. Machine Vision and Applications https://doi.org/10.1007/s00138-018-0908- 0 SPECIAL ISSUE PAPER Multitask learning for neural generative question answering Yanzhou Huang1 · Tao Zhong1 Received: 20 September …

CoChat: Enabling bot and human collaboration for task completion
X Luo, Z Lin, Y Wang, Z Nie – Thirty-Second AAAI Conference on Artificial …, 2018 – aaai.org
… Wen et al. (2016) propose a neural-network-based trainable dialog system along with a new way of collecting dialog data. Bor- des and Weston (2016) report an end-to-end dialog system based on Memory Networks that can achieve promising, yet imperfect, performance …

A char-based seq2seq submission to the E2E NLG Challenge
S Agarwal, M Dymetman, E Gaussier – macs.hw.ac.uk
… This is a component in Spo- ken Dialogue Systems, where recent advances in Deep Learning are stimulating interest towards us- ing end-to-end … In parallel, we trained a reverse model which tries to reconstruct the MR given the target RF, similar to the autoencoder model by …

Fighting offensive language on social media with unsupervised text style transfer
CN Santos, I Melnyk, I Padhi – arXiv preprint arXiv:1805.07685, 2018 – arxiv.org
… which can be thought to be similar to an auto- encoder loss in (1) but in the style domain. Page 3 … Hen- derson et al. (2018) have used Twitter and Reddit datasets to evaluate the impact of offensive lan- guage and hate speech in neural dialogue systems …

Generating Informative Responses with Controlled Sentence Function
P Ke, J Guan, M Huang, X Zhu – Proceedings of the 56th Annual Meeting …, 2018 – aclweb.org
… structure equipped with a latent variable in conditional vari- ational autoencoder (CVAE) (Sohn … func- tions to achieve different conversational pur- poses in open-domain dialogue systems … the information of controllable attributes as in the variational autoencoders (VAE) (Zhou …

Neural response generation with dynamic vocabularies
Y Wu, W Wu, D Yang, C Xu, Z Li – Thirty-Second AAAI Conference on …, 2018 – aaai.org
Page 1. Neural Response Generation with Dynamic Vocabularies Yu Wu, †? Wei Wu, ‡ Dejian Yang, † Can Xu, ‡ Zhoujun Li, †?? † State Key Lab of Software Development Environment, Beihang University, Beijing, China ‡ Microsoft …

A Hierarchical Latent Structure for Variational Conversation Modeling
Y Park, J Cho, G Kim – arXiv preprint arXiv:1804.03424, 2018 – arxiv.org
… Variational autoencoders (VAE) combined with hierarchical RNNs have emerged as a powerful framework for conversation model- ing … variable in the hierarchical RNNs by incorporat- ing the variational autoencoder (VAE) framework (Kingma and Welling, 2014; Rezende et al …

Generating informative and diverse conversational responses via adversarial information maximization
Y Zhang, M Galley, J Gao, Z Gan, X Li… – Advances in Neural …, 2018 – papers.nips.cc
… sampling at each local time step; while in our approach, ˜T is a deterministic function of S and Z, therefore, the randomness is global and separated out from the deterministic propagation, which resembles the reparameterization trick used in variational autoencoder [16] …

Syntactic manipulation for generating more diverse and interesting texts
JM Deriu, M Cieliebak – 11th International Conference on …, 2018 – digitalcollection.zhaw.ch
… and Schmidhu- ber, 1997) to control the semantic properties of an utterance, whereas (Hu et al., 2017) use varia- tional autoencoder (VAE) and … The dialogue system commu- nity has proposed most work on this topic, as the end-to-end trainable algorithms tend to pro- duce the …

SALSA-TEXT: self attentive latent space based adversarial text generation
J Gagnon-Marchand, H Sadeghi, M Haidar… – arXiv preprint arXiv …, 2018 – arxiv.org
… Text generation is of particular interest in many natural language processing (NLP) applications such as dialogue systems, machine translation … Adversarial autoencoder (AAE) (Makhzani et al., 2015b) proposes an adversarial setup to train probabilistic autoencoders …

Identifying Domain Adjacent Instances for Semantic Parsers
J Ferguson, J Christensen, E Li, E Gonzàlez – arXiv preprint arXiv …, 2018 – arxiv.org
… ORACLE 0.558 0.494 0.817 0.661 0.711 0.770 0.826 0.555 AUTOENCODER 0.413 0.268 0.581 0.447 0.463 0.530 0.543 0.417 … 2017. Neural sen- tence embedding using only in-domain sentences for out-of-domain sentence detection in dialog systems …

Multimodal Differential Network for Visual Question Generation
BN Patro, S Kumar, VK Kurmi… – arXiv preprint arXiv …, 2018 – arxiv.org
… Current dialog systems as evaluated in (Chattopadhyay et al., 2017) show that when trained between bots, AI-AI dialog sys- tems show … an encoder-decoder based framework whereas in the latter work, the authors extend it by using a varia- tional autoencoder based sequential …

Polite Dialogue Generation Without Parallel Data
T Niu, M Bansal – arXiv preprint arXiv:1805.03162, 2018 – arxiv.org
… Most current chatbots and conversational mod- els lack any such style, which can be a social issue because human users might learn biased styles from such interactions, eg, kids learning to be rude be- cause the dialogue system encourages short, curt re- sponses, and also …

MEMD: A Diversity-Promoting Learning Framework for Short-Text Conversation
M Zou, X Li, H Liu, Z Deng – … of the 27th International Conference on …, 2018 – aclweb.org
… ˆ hN as input, it constitutes an auto-encoder with the aux-encoder … 2016. Build- ing end-to-end dialogue systems using generative hierarchical neural network models … 2017. Learning discourse-level diversity for neural dialog models using conditional variational autoencoders …

Topic-based evaluation for conversational bots
F Guo, A Metallinou, C Khatri, A Raju… – arXiv preprint arXiv …, 2018 – arxiv.org
… systems. In Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, page 1711–1721, 2015. [24] T. Zhao, R. Zhao, and M. Eskenazi. Learning discourse-level diversity for neural dialog models using conditional variational autoencoders …

Text Generation Based on Generative Adversarial Nets with Latent Variables
H Wang, Z Qin, T Wan – Pacific-Asia Conference on Knowledge Discovery …, 2018 – Springer
… poems. It is also essential to machine translation, text summarization, question answering and dialogue system [18]. One … variation. We combine recurrent neural network with variational autoencoder (VAE) [10] as generator G. In …

DP-GAN: Diversity-Promoting Generative Adversarial Network for Generating Informative and Diversified Text
J Xu, X Sun, X Ren, J Lin, B Wei, W Li – arXiv preprint arXiv:1802.01345, 2018 – arxiv.org
Page 1. DP-GAN: Diversity-Promoting Generative Adversarial Network for Generating Informative and Diversified Text Jingjing Xu?, Xu Sun?, Xuancheng Ren, Junyang Lin, Binzhen Wei, Wei Li School of Electronics Engineering …

Style Transfer Through Multilingual and Feedback-Based Back-Translation
S Prabhumoye, Y Tsvetkov, AW Black… – arXiv preprint arXiv …, 2018 – arxiv.org
… It is important for dialog systems such as personalized agents, customer service agents and smart home assistants to generate responses that are fluent and fit the social setting … The paper introduces cross-aligned auto- encoder with discriminators. Hu et al. (2017) …

Troubling trends in machine learning scholarship
ZC Lipton, J Steinhardt – arXiv preprint arXiv:1807.03341, 2018 – arxiv.org
… Take for instance [55], which characterizes an autoencoder as a “simulated brain … We note that recent interest in chatbot startups co- occurred with anthropomorphic descriptions of dialogue systems and reinforcement learners both in papers and in the media, although it may be …

Robust Spoken Language Understanding via Paraphrasing
A Ray, Y Shen, H Jin – arXiv preprint arXiv:1809.06444, 2018 – arxiv.org
… intents and slot labels from user utterances is a fundamental step in all spoken language un- derstanding (SLU) and dialog systems … to generate the exact same input utterance x, while the first decoder generates the paraphrase x . Such additional autoencoder constraint has …

Experimental research on encoder-decoder architectures with attention for chatbots
MR Costa-jussà, Á Nuez, C Segura – Computación y Sistemas, 2018 – cys.cic.ipn.mx
… An autoencoder is a type of neural network that aims at learning a representation of the input while allowing for a decoding of this representation by minimizing the recovering error … IRIS: a chat-oriented dialogue system based on the vector space model …

Neural Ideal Point Estimation Network
K Song, W Lee, IC Moon – Thirty-Second AAAI Conference on Artificial …, 2018 – aaai.org
… Further advances have made through casting this autoencoding mechanism to the variational inference approaches, and a variational autoencoder (VAE) (Kingma … learning the low dimensional representa- tions with multi-layered perceptron (MLP) autoencoders, ie SDAE …

Feature Selection and Nuisance Attribute Projection for Speech Emotion Recognition
MAK Man-Wai – eie.polyu.edu.hk
… A typical example of such scenario is spoken dialog systems for customer services … then, emotion features are extracted from the emotion hidden neurons of the autoencoder [14 … and PA Manzagol, “Extracting and composing robust features with denoising autoencoders,” in Proc …

Deep Learning in Natural Language Generation from Images
X He, L Deng – Deep Learning in Natural Language Processing, 2018 – Springer
… 3, mainly for the specific dialog system application … (2015) have investigated to generate dense image captions for individual regions in images, and a variational autoencoder was developed in Pu et al. (2016) for image captioning …

Lired: A light-weight real-time fault detection system for edge computing using lstm recurrent neural networks
D Park, S Kim, Y An, JY Jung – Sensors, 2018 – mdpi.com
… network and a compressor network. The compressor network uses models called recirculation networks in [23], which is identical to the neural network structure that is now known as an auto-encoder. The authors of [23] proposed …

Diversity-Promoting GAN: A Cross-Entropy Based Generative Adversarial Network for Diversified Text Generation
J Xu, X Ren, J Lin, X Sun – Proceedings of the 2018 Conference on …, 2018 – aclweb.org
Page 1. Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 3940–3949 Brussels, Belgium, October 31 – November 4, 2018. c 2018 Association for Computational Linguistics 3940 …

Speech emotion recognition: two decades in a nutshell, benchmarks, and ongoing trends
BW Schuller – Communications of the ACM, 2018 – dl.acm.org
… COMMUNICATION WITH COMPUTING machinery has become increasingly ‘chatty’ these days: Alexa, Cortana, Siri, and many more dialogue systems have hit the consumer market on a broader basis than ever, but do any of them truly notice our emotions and react to them like …

Noising and Denoising Natural Language: Diverse Backtranslation for Grammar Correction
Z Xie, G Genthial, S Xie, A Ng, D Jurafsky – Proceedings of the 2018 …, 2018 – aclweb.org
… text data. Sim- ilarly, while denoising autoencoders for images have been shown to help with representation learn- ing (Vincent et al., 2010), similar methods for learning representations are not well developed for text. Some …

A Joint Introduction to Natural Language Processing and to Deep Learning
L Deng, Y Liu – Deep Learning in Natural Language Processing, 2018 – Springer
… Typical applications in NLP include speech recognition, spoken language understanding, dialogue systems, lexical analysis, parsing, machine translation, knowledge graph, information retrieval, question answering, sentiment analysis, social computing, natural language …

A Generative Model for category text generation
Y Li, Q Pan, S Wang, T Yang, E Cambria – Information Sciences, 2018 – Elsevier
… Controllable text generation [17] applies the variable auto-encoder (VAE) together with controllable information to generate category sentences. Zhang et al. [49] and Semeniuta et al. [36] used GANs for text data generation and achieved state-of-the-art results. 2.3 …

Reusing Neural Speech Representations for Auditory Emotion Recognition
E Lakomkin, C Weber, S Magg, S Wermter – arXiv preprint arXiv …, 2018 – arxiv.org
… More- over, in many applications, such as dialog systems, we would need to transcribe spoken text and iden- tify its emotion jointly … An autoencoder network (Ghosh et al., 2016) was demonstrated to learn to com- press the speech frames before the emotion clas- sification …

From Emoji Usage to Categorical Emoji Prediction
G Guibon, M Ochs, P Bellot – … and Intelligent Text …, 2018 – hal-amu.archives-ouvertes.fr
… The methodology and resources can be used to recommend the emotion cat- egories to express by an embodied conversational agent or in general dialog system, such as trending … Li, J., Luong, MT, Jurafsky, D.: A hierarchical neural autoencoder for paragraphs and documents …

Better Conversations by Modeling, Filtering, and Optimizing for Coherence and Diversity
X Xu, O Dušek, I Konstas, V Rieser – arXiv preprint arXiv:1809.06873, 2018 – arxiv.org
… and the gener- ated response, (2) we filter our training cor- pora based on the measure of coherence to obtain topically coherent and lexically diverse context-response pairs, (3) we then train a re- sponse generator using a conditional varia- tional autoencoder model that …

Resolving Abstract Anaphora Implicitly in Conversational Assistants using a Hierarchically stacked RNN
P Khurana, P Agarwal, G Shroff, L Vig – Proceedings of the 24th ACM …, 2018 – dl.acm.org
… com ABSTRACT Recent proliferation of conversational systems has resulted in an increased demand for more natural dialogue systems, capable of more sophisticated interactions than merely pro- viding factual answers. This …

On the Generation of Medical Question-Answer Pairs
S Shen, Y Li, N Du, X Wu, Y Xie, S Ge, T Yang… – arXiv preprint arXiv …, 2018 – arxiv.org
… Moreover, to explore the diversity of medical QA pair generation with constraints of external knowledge in the cor- responding two levels, we introduce an idea based on Con- ditional Variational Autoencoder, which is constrained on the condition over the whole question and …

Understanding back-translation at scale
S Edunov, M Ott, M Auli, D Grangier – arXiv preprint arXiv:1808.09381, 2018 – arxiv.org
… (2018a) to beam search out- puts. Adding noise to input sentences has been very beneficial for the autoencoder se- tups of (Lample et al., 2018a; Hill et al., 2016) which is inspired by denoising autoencoders (Vincent et al., 2008) …

A Syntactically Constrained Bidirectional-Asynchronous Approach for Emotional Conversation Generation
J Li, X Sun – Proceedings of the 2018 Conference on Empirical …, 2018 – aclweb.org
… 2017. Multimodal autoencoder: A deep learning approach to filling in missing sen- sor data and enabling better mood prediction. In Proc … 2016. How not to evaluate your dialogue system: An em- pirical study of unsupervised evaluation metrics for dialogue response generation …

Unsupervised Text Style Transfer using Language Models as Discriminators
Z Yang, Z Hu, C Dyer, EP Xing… – arXiv preprint arXiv …, 2018 – arxiv.org
… 2 Page 3. to vy. Combining the content vector zx and the style label vy, we can generate a new sentence ˜x (the transferred sentences are denotes as ˜x and ˜y). One unsupervised approach is to use the auto-encoder model …

Unsupervised Controllable Text Formalization
P Jain, A Mishra, AP Azad… – arXiv preprint arXiv …, 2018 – arxiv.org
… For ex- ample, in the context of commercial dialog systems alone, there are several scenarios where a system’s answer (which … Denoyer, and Ranzato (2017) have proposed ar- chitectures for unsupervised language translation with un- supervised autoencoder based lexicon …

Improving Pointer-Generator Network with Keywords Information for Chinese Abstractive Summarization
X Jiang, P Hu, L Hou, X Wang – CCF International Conference on Natural …, 2018 – Springer
… 2015) and dialogue systems (Serban et al. 2016) … (2017) add historical dependencies on the latent variables of Variational Autoencoder (VAEs) and propose a deep recurrent generative decoder (DRGD) to distill the complex latent structures implied in the target summaries …

Zero-Shot Anticipation for Instructional Activities
F Sener, A Yao – arXiv preprint arXiv:1812.02501, 2018 – arxiv.org
… Sequence-to-sequence learning [55] has made it possi- ble to successfully generate continuous text [27] and build dialogue systems [13, 58] … These three RNNs are learned jointly as an auto-encoder in an initial training step …

Social Communicative Events in Human Computer Interactions
K El Haddad, H Cakmak, M Doumit, G Pironkov… – researchgate.net
… To explain this more, we first pre-train the autoencoder, using the neutral … we intend to use Stacked Denoising Autoencoders instead of simple autoencoders as was … using linguistic and nonlinguistic information and its application to spoken dialogue system.” in INTERSPEECH …

Decoupling Strategy and Generation in Negotiation Dialogues
H He, D Chen, A Balakrishnan, P Liang – arXiv preprint arXiv:1808.09637, 2018 – arxiv.org
… history. Our framework follows that of traditional goal- oriented dialogue systems (Young et al., 2013), with one important difference: coarse dialogue acts are not intended to and cannot capture the full meaning of an utterance …

An Affect-Rich Neural Conversational Model with Biased Attention and Weighted Cross-Entropy Loss
P Zhong, D Wang, C Miao – arXiv preprint arXiv:1811.07078, 2018 – arxiv.org
… degradation in lan- guage fluency. One note is that the out-domain test perplex- ity for all models is quite large as compared to in-domain perplexity, as well as other dialog systems, eg, (Vinyals and Le 2015). One possible reason is …

Chat More If You Like: Dynamic Cue Words Planning to Flow Longer Conversations
L Yao, R Xu, C Li, D Zhao, R Yan – arXiv preprint arXiv:1811.07631, 2018 – arxiv.org
… Many research efforts have been dedicated to building such dialogue systems, yet few shed light on model- ing the conversation flow in an ongoing dialogue … Previous efforts focus on task-oriented dialogue systems (Wen et al. 2017; Eric and Manning 2017; Liu et al …

Implementing ChatBots using Neural Machine Translation techniques
A Nuez Ezquerra – 2018 – upcommons.upc.edu
Page 1. Implementing ChatBots using Neural Machine Translation techniques Degree’s Thesis Telecommunications Sciences and Technologies Author: Alvaro Nuez Ezquerra Advisors: Marta R. Costa-Juss`a and Carlos Segura Perales …

A Hierarchical Structured Self-Attentive Model for Extractive Document Summarization (HSSAS)(April 2018)
K Al-Sabahi, Z Zuping, M Nadher – ieeexplore.ieee.org
… For example, a recursive autoencoder based model proposed by [22] to summarize documents on the Opinosis dataset [23 … For extractive query-oriented single-document summarization, [27] used a deep auto-encoder to compute a feature space from the term-frequency (tf) input …

Tree2Tree Learning with Memory Unit
N Miao, H Wang, R Le, C Tao, M Shang, R Yan… – 2018 – openreview.net
… Our model is compared with recursive Auto-Encoder (RAE) (Socher et al., 2011a) … that our tree2tree perfectly maintain almost all of the structured information when it is utilized as an autoencoder … Dy- namic pooling and unfolding recursive autoencoders for paraphrase detection …

ICBK 2018
G Li, X Fu, X Ren – computer.org
… 184 Michael Mayo (University of Waikato), Sarah Wakes (University of Otago), and Chris Anderson (University of Waikato) Discriminative Graph Autoencoder 192 … Short-Attention Mechanism for Generative Dialogue System 268 …

Automatic Acceptance Prediction for Answers in Online Healthcare Community
Q Liu, K Liao, Z Wei – 2018 IEEE International Conference on …, 2018 – ieeexplore.ieee.org
… [15] J. Li, M.-T. Luong, and D. Jurafsky, “A hierarchical neural autoencoder for paragraphs and documents,” arXiv preprint arXiv:1506.01057, 2015 … [19] Z. Wei, Q. Liu, B. Peng, H. Tou, T. Chen, X. Huang, K.-F. Wong, and X. Dai, “Task-oriented dialogue system for automatic …

Response Generation by Context-aware Prototype Editing
Y Wu, F Wei, S Huang, Z Li, M Zhou – arXiv preprint arXiv:1806.07042, 2018 – arxiv.org
… tion like variational auto-encoder. In the training phase, the parameter of the distribution is condi- tional on the context differences … CVAE: The conditional variational auto- encoder is a popular method of increasing the diversity of response generation (Zhao et al., 2017) …

Synthesizing Tabular Data using Generative Adversarial Networks
L Xu, K Veeramachaneni – arXiv preprint arXiv:1811.11264, 2018 – arxiv.org
… Other GAN applications include information retrieval [45], dialogue systems [27], and speech processing [32] … fully synthetic data, neural models are used to impute missing values in datasets; for example, Gondara and Wang [15] uses deep de-noising autoencoders, and Yoon …

Application of Emotion Recognition and Modification for Emotional Telugu Speech Recognition
VVR Vegesna, K Gurugubelli, AK Vuppala – Mobile Networks and …, 2018 – Springer
… Systems 4(4):301–306 3. Dybkjaer L, Bernsen NO, Minker W (2004) Evaluation and usability of multimodal spoken language dialogue systems … Deng J, Zhang Z, Marchi E, Schuller B (2013) Sparse autoencoder-based feature transfer learning for speech emotion recognition …

Generating Multiple Diverse Responses for Short-Text Conversation
J Gao, W Bi, X Liu, J Li, S Shi – arXiv preprint arXiv:1811.05696, 2018 – arxiv.org
… By choosing different responding mechanisms, multiple output responses can be generated. Zhao et al. (2017) adopted the conditional variational auto-encoder (CVAE) and generated multiple re- sponses by drawing samples from the prior distribution …

Word Representations for Emergent Communication and Natural Language Processing
M KÅGEBÄCK – research.chalmers.se
Page 1. THESIS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Word Representations for Emergent Communication and Natural Language Processing MIKAEL KÅGEBÄCK Department of Computer Science and Engineering …

Learning Factorized Multimodal Representations
YHH Tsai, PP Liang, A Zadeh, LP Morency… – arXiv preprint arXiv …, 2018 – arxiv.org
… 2.2 Minimizing Joint-Distribution Wasserstein Distance over Multimodal Data Two common choices for approximate inference in autoencoding structures are Variational Autoen- coders (VAEs) [24] and Wasserstein Autoencoders (WAEs) [57] …

QuaSE: Sequence Editing under Quantifiable Guidance
Y Liao, L Bing, P Li, S Shi, W Lam… – Proceedings of the 2018 …, 2018 – aclweb.org
… We propose a framework to address this task. The fundamental module of our framework is a Variational Autoencoder (VAE) (Kingma and Welling, 2013) to encode each input sentence into a latent content factor and a latent outcome fac …

Adversarial Training in Affective Computing and Sentiment Analysis: Recent Advances and Perspectives
J Han, Z Zhang, N Cummins, B Schuller – arXiv preprint arXiv:1809.08927, 2018 – arxiv.org
… on deep convolutional neural networks [19], the AEGAN based on autoencoders [71], the … For example, in [89], the authors proposed a conditional difference adversarial autoencoder, to learn … voice conversion in natural speech and proposed a varia- tional autoencoding WGAN …

Rich Short Text Conversation Using Semantic-Key-Controlled Sequence Generation
K Yu, Z Zhao, X Wu, H Lin, X Liu – IEEE/ACM Transactions on Audio …, 2018 – dl.acm.org
… in encoder side. r Auto-encoder: When post-comment pairs are not avail- able, it … richer external knowledge. Here, auto-encoder is trained on the non-parallel data set and convert the data into sentence embeddings. The sentence …

Artificial Intelligence in the Rising Wave of Deep Learning: The Historical Path and Future Outlook [Perspectives]
L Deng – IEEE Signal Processing Magazine, 2018 – saebogota.unal.edu.co
… Due to this strength, the first-generation AI systems are still in use today. Examples are nar- row-domain dialogue systems and chat- bots, chess-playing programs, traffic light controllers, optimization software for logistics of good deliveries, etc …

Building Context-aware Clause Representations for Situation Entity Type Classification
Z Dai, R Huang – arXiv preprint arXiv:1809.07483, 2018 – arxiv.org
Page 1. Building Context-aware Clause Representations for Situation Entity Type Classification Zeyu Dai, Ruihong Huang Department of Computer Science and Engineering Texas A&M University {jzdaizeyu, huangrh}@tamu.edu Abstract …

Incorporating Relevant Knowledge in Context Modeling and Response Generation
Y Li, W Li, Z Cao, C Chen – arXiv preprint arXiv:1811.03729, 2018 – arxiv.org
… 2.3 Knowledge-grounded Conversation Models In the line of combining conversational agents with knowledge bases, most work focus on developing task-oriented dialogue systems … (2017) that develops a dialogue system to talk about musics …

Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing
E Riloff, D Chiang, H Julia, T Jun’ichi – Proceedings of the 2018 …, 2018 – aclweb.org
… She works on spoken language processing and NLP, studying text-to-speech synthesis, spoken dialogue systems, entrainment in conversation … 105 Associative Multichannel Autoencoder for Multimodal Word Representation Shaonan Wang, Jiajun Zhang and Chengqing Zong …

Proceedings of the AMTA 2018 Workshop on Technologies for MT of Low Resource Languages (LoResMT 2018)
CH Liu – Proceedings of the AMTA 2018 Workshop on …, 2018 – aclweb.org
… One approach is to use a model called auto- encoder to train neural Language Models (LMs) for source and target languages … His research interests are dialogue systems, in particular robustness to noise in input and style transfer in dialogues …

Gunrock: Building A Human-Like Social Bot By Leveraging Large Scale Real User Data
CY Chen, D Yu, W Wen, YM Yang, J Zhang, M Zhou… – dex-microsites-prod.s3.amazonaws …
… 1 Introduction One considerable challenge in dialog system research is training and testing dialog systems with a large number of users … 2 Related Work Task-oriented and open domain dialog systems have been widely studied …

Language Style Transfer from Non-Parallel Text with Arbitrary Styles
Y Zhao, VW Bi, D Cai, X Liu, K Tu, S Shi – 2018 – openreview.net
… As described in Sec- tion 2 and 3, STB is built upon an auto-encoder framework. It focuses on transferring sentences from one style to the other … The ubuntu dialogue corpus: A large dataset for research in unstructured multi-turn dialogue systems …

Towards End-to-End Prosody Transfer for Expressive Speech Synthesis with Tacotron
RJ Skerry-Ryan, E Battenberg, Y Xiao, Y Wang… – arXiv preprint arXiv …, 2018 – arxiv.org
… Therefore, as with an autoencoder, care must be taken to choose an architecture that sufficiently bottlenecks the prosody embedding such that it is forced … This can be highly useful in building templated dialogue systems capable of synthesizing a template with a desired prosody …

On the Vector Representation of Utterances in Dialogue Context
L Pragst, N Rach, W Minker, S Ultes – Proceedings of the Eleventh …, 2018 – aclweb.org
… Rieser, V. and Lemon, O. (2011). Reinforcement learning for adaptive dialogue systems: a data-driven methodol- ogy for dialogue management and natural language gen- eration … Semi-supervised recursive autoencoders for predicting sentiment distributions …

Incorporating Pseudo-Parallel Data for Quantifiable Sequence Editing
Y Liao, L Bing, P Li, S Shi, W Lam, T Zhang – arXiv preprint arXiv …, 2018 – arxiv.org
Page 1. arXiv:1804.07007v1 [cs.CL] 19 Apr 2018 Incorporating Pseudo-Parallel Data for Quantifiable Sequence Editing Yi Liao? 1, Lidong Bing2, Piji Li1, Shuming Shi2, Wai Lam1, Tong Zhang2 1The Chinese University of …

Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
I Gurevych, Y Miyao – Proceedings of the 56th Annual Meeting of the …, 2018 – aclweb.org
Page 1. ACL 2018 The 56th Annual Meeting of the Association for Computational Linguistics Proceedings of the Conference, Vol. 1 (Long Papers) July 15 – 20, 2018 Melbourne, Australia Page 2. Diamond Sponsors: Platinum Sponsors: Gold Sponsors: ii Page 3. Silver Sponsors …

Deep learning for sentiment analysis: A survey
L Zhang, S Wang, B Liu – Wiley Interdisciplinary Reviews: Data …, 2018 – Wiley Online Library
… One often stacks autoencoders into layers. A higher level autoencoder uses the output of a lower one as its training data. The stacked autoencoders (Bengio, Lamblin, Popovici, & Larochelle, 2006) along with restricted Boltzmann …

Amanuensis: The Programmer’s Apprentice
T Dean, M Chiang, M Gomez, N Gruver, Y Hindy… – arXiv preprint arXiv …, 2018 – arxiv.org
Page 1. Amanuensis: The Programmer’s Apprentice Thomas Dean1,2 Maurice Chiang2 Marcus Gomez2 Nate Gruver2 Yousef Hindy2 Michelle Lam2 Peter Lu2 Sophia Sanchez2 Rohun Saxena2 Michael Smith2 Lucy Wang2 Catherine Wong2 Abstract …

Towards Text Generation with Adversarially Learned Neural Outlines
S Subramanian, SR Mudumba, A Sordoni… – Advances in Neural …, 2018 – papers.nips.cc
… AAEs and WAEs are similar to Variational Autoencoders (VAE) [28] in that they learn an encoder that produces an approximate latent posterior … ARAEs train, by means of a critic, a flexible prior distribution regularizing the sentence representations obtained by an auto-encoder …

Creating an Emotion Responsive Dialogue System
A Vadehra – 2018 – uwspace.uwaterloo.ca
Creating an Emotion Responsive Dialogue System by Ankit Vadehra … We present the different scoring models in Equation 2.4. In all our affect responsive dialogue system models we use the additive alignment scoring model proposed by Bahdanau et al. 2.2 Dialogue System …

Style transfer as unsupervised machine translation
Z Zhang, S Ren, S Liu, J Wang, P Chen, M Li… – arXiv preprint arXiv …, 2018 – arxiv.org
… 2018) leverages the auto- encoder framework to learn an encoder and a decoder, in which the encoder constructs a latent vector by removing the style information and extracting attribute-independent content from the input sentence, and the decoder generates the output …

Towards an Unsupervised Entrainment Distance in Conversational Speech using Deep Neural Networks
M Nasir, B Baucom, S Narayanan… – arXiv preprint arXiv …, 2018 – arxiv.org
… Signal Processing [8, 9]. Moreover, it also contributes to the modeling and development of ‘human-like’ spoken dialog systems or conversational … Even though this deep neural network resembles autoencoder architectures, it does not reconstruct itself but rather tries to encode …

Learning to Represent Bilingual Dictionaries
M Chen, Y Tian, H Chen, KW Chang, S Skiena… – arXiv preprint arXiv …, 2018 – arxiv.org
… (ii) Bilin- gual paraphrase identification asks whether two sentences in different languages essentially ex- press the same meaning, which is critical to ques- tion answering or dialogue systems that apprehend multilingual utterances (Bannard and Callison- Burch, 2005) …

Natural Language Generation with Neural Variational Models
H Bahuleyan – arXiv preprint arXiv:1808.09012, 2018 – arxiv.org
… 2.2.9 Dialog Systems … A class of models that combine deep learning and variational inference, namely Variational Autoencoders (VAE) (Kingma and Welling, 2013) have been success- fully … 1. A variational auto-encoder (VAE) is first designed following the work of Bowman et al …

Natural Language Generation with Neural Variational Models
H Pallikara Bahuleyan – 2018 – uwspace.uwaterloo.ca
… 2.2.9 Dialog Systems … A class of models that combine deep learning and variational inference, namely Variational Autoencoders (VAE) (Kingma and Welling, 2013) have been success- fully … 1. A variational auto-encoder (VAE) is first designed following the work of Bowman et al …

Database Systems for Advanced Applications
J Pei, Y Manolopoulos, S Sadiq, J Li – 2018 – books.google.com
Page 1. Jian Pei· Yannis Manolopoulos Shazia Sadiq · Jianxin Li (Eds.) Database Systems for Advanced Applications 23rd International Conference, DASFAA 2018 Gold Coast, QLD, Australia, May 21–24, 2018 Proceedings, Part II 123 Page 2 …

Weakly-supervised Neural Semantic Parsing with a Generative Ranker
J Cheng, M Lapata – arXiv preprint arXiv:1808.07625, 2018 – arxiv.org
Page 1. Weakly-supervised Neural Semantic Parsing with a Generative Ranker Jianpeng Cheng and Mirella Lapata Institute for Language, Cognition and Computation School of Informatics, University of Edinburgh 10 Crichton …

Latent Semantic Analysis Approach for Document Summarization Based on Word Embeddings
K Al-Sabahi, Z Zuping, Y Kang – arXiv preprint arXiv:1807.02748, 2018 – arxiv.org
… summarization [19]. For extractive summarization, [20] proposed extractive query-oriented single-document summarization using a deep auto- encoder to compute a feature space from the term-frequency (tf) input. They developed …

Adversarially-Trained Normalized Noisy-Feature Auto-Encoder for Text Generation
X Zhang, Y LeCun – arXiv preprint arXiv:1811.04201, 2018 – arxiv.org
… ADVERSARIALLY-TRAINED NORMALIZED NOISY- FEATURE AUTO-ENCODER FOR TEXT GENERATION … yann@cs.nyu.edu ABSTRACT This article proposes Adversarially-Trained Normalized Noisy-Feature Auto- Encoder (ATNNFAE) for byte-level text generation …

ICMLA 2017
VM Tavano – ieeexplore.ieee.org
Page 1. 2017 16th IEEE International Conference on Machine Learning and Applications ICMLA 2017 Table of Contents Preface xxvi Organizing Committee xxviii Program Commitee xxix Keynotes xxxii C1A: Deep Learning I …

Building a Neural Semantic Parser from a Domain Ontology
J Cheng, S Reddy, M Lapata – arXiv preprint arXiv:1812.10037, 2018 – arxiv.org
… approach. Subsequent work of Kociský et al. (2016) attempts to explore the meaning representation space with a generative autoencoder. They … sampled. These samples are used as semi-supervised training data to the autoencoder. Other …

From Words to Emoticons: Deep Emotion Recognition in Text and Its Wider Implications
R Rzepka, M Takizawa, J Vallverd?u, M Ptaszynski… – researchgate.net
Page 1. From Words to Emoticons: Deep Emotion Recognition in Text and Its Wider Implications Rafal Rzepka 1 , Mitsuru Takizawa 1 , Jordi Vallverd?u 2 , Michal Ptaszynski 3 , Pawel Dybala 4 , Kenji Araki 1 1 Hokkaido University …

Structured Neural Models for Natural Language Processing
M Ma – 2018 – ir.library.oregonstate.edu
… 4.3. (b): the hidden activations h for the same input image from basic autoencoders. . . . . 39 … CNNs-multichannel [27] 81.1 47.4 92.2 86.0? Deep CNNs [26] – 48.5 93.0 – Recursive NNs Recursive Autoencoder [63] 77.7 43.2 – – Recursive Neural Tensor [64] – 45.7 …

Neural natural language generation with unstructured contextual information
H Gete Ugarte – 2018 – addi.ehu.es
… This new approach to NLG has been successful and has become an important component in multiple applications including dialogue systems, text summarisation, image captioning or video description, with considerable commercial interest in different NLG applications such …

Style Transfer Through Back-Translation
S Prabhumoye, Y Tsvetkov, R Salakhutdinov… – arXiv preprint arXiv …, 2018 – arxiv.org
… Baseline. We compare our model against the “cross-aligned” auto-encoder (Shen et al., 2017), which uses style-specific decoders to align the style of generated sentences to the actual distribu- tion of the style. We used the off-the-shelf senti- ment model released by Shen et al …

Speaker-independent speech separation with deep attractor network
Y Luo, Z Chen, N Mesgarani – IEEE/ACM Transactions on …, 2018 – ieeexplore.ieee.org
… For example, word embedding is currently one of the standard tools to extract the relation- ship and connection between different words, and serves as a front-end to more complex tasks such as machine translation and dialogue systems [45], [46] …

Speaker and language recognition and characterization: Introduction to the CSL special issue
E Lleida, LJ Rodriguez-Fuentes – 2018 – Elsevier
… On the other hand, spoken language recognition (SLR) has also witnessed a remarkable interest from the community as an auxiliary technology for speech recognition (Gonzalez-Dominguez et al., 2015b), dialogue systems (Lopez-Cozar and Araki, 2005) and multimedia …

Why are Sequence-to-Sequence Models So Dull? Understanding the Low-Diversity Problem of Chatbots
S Jiang, M de Rijke – arXiv preprint arXiv:1809.01941, 2018 – arxiv.org
… To increase variability, Serban et al. (2017); Zhao et al. (2017) propose to introduce variational autoencoders (VAEs) to Seq2Seq models … 2016. Building end-to-end dialogue systems using gener- ative hierarchical neural network models …

Why are Sequence-to-Sequence Models So Dull?
S Jiang, M de Rijke – EMNLP 2018, 2018 – aclweb.org
… To increase variability, Serban et al.(2017); Zhao et al.(2017) propose to introduce variational autoencoders (VAEs) to Seq2Seq models … 2016. Building end-to-end dialogue systems using gener- ative hierarchical neural network models. In AAAI, volume 16, pages 3776–3784 …

Redefining Concatenative Speech Synthesis for Use in Spontaneous Conversational Dialogues; A Study with the GBO Corpus
N Campbell – 2018 – real.mtak.hu
… For an interactive spoken dialogue system, there will be considerable context- ual information available for such a choice to be made … H., & Vit, J. (2017) Google’s Next-Generation Real-Time Unit-Selection Synthesizer using Sequence-To-Sequence LSTM-based Autoencoders …

A Clustering Based Adaptive Sequence-to-Sequence Model for Dialogue Systems
D Ren, Y Cai, WH Chan, Z Li – Big Data and Smart Computing …, 2018 – ieeexplore.ieee.org
… These responses will lead users don’t use dialogue systems any more … present a novel framework based on conditional variational autoencoders that captures the discourse-level diversity in the encoder [6]. However, they all use a single model to train all dialogue data …

VMAV-C: A Deep Attention-based Reinforcement Learning Algorithm for Model-based Control
X Liang, Q Wang, Y Feng, Z Liu, J Huang – arXiv preprint arXiv …, 2018 – arxiv.org
… We learn the environment model through Mixture Density Network Recurrent Network(MDN-RNN) for agents to interact, with combinations of variational auto-encoder(VAE) and attention incorporated in state value estimates during the process of learning policy …

Deep Learning Inference in Facebook Data Centers: Characterization, Performance Optimizations and Hardware Implications
J Park, M Naumov, P Basu, S Deng, A Kalaiah… – arXiv preprint arXiv …, 2018 – arxiv.org
… The latter has progressed from matrix and tensor-based factorizations [19, 37] to autoencoder and neural collaborative filtering [27, 41, 59] … 17, 46], speech recognition [23], syntactic and semantic parsing [11, 69] as well as question answering and dialog systems [9, 52] …

Multimodal Representation Learning for Visual Reasoning and Text-to-Image Translation
R Saha – 2018 – search.proquest.com
… As generative models saw a resurgence by the introduction of Generative Adver- sarial Networks (Goodfellow et al., 2014), Variational Autoencoders (Kingma and … 2017a), visual dialogue systems (Massiceti et al., 2018; Jain et al., 2018) etc. 10 Page 23. Chapter 3 …

An improved formula for Jacobi rotations
CF Borges, Z Majdisova, V Skala, A Monszpart… – arXiv preprint arXiv …, 2018 – arxiv.org
… Authors: Raja Asim. Comments: 8 pages, 6 images. Subjects: Computer Vision and Pattern Recognition (cs.CV). arXiv:1806.08079 [pdf, other] Title: GrCAN: Gradient Boost Convolutional Autoencoder with Neural Decision Forest …

Choice of a classifier, based on properties of a dataset: case study-speech emotion recognition
SG Koolagudi, YVS Murthy, SP Bhaskar – International Journal of Speech …, 2018 – Springer
… al. (2014). UNIVERSUM. 384. ABC GeWEC EMODB SUSAS. Autoencoder based Adaptation model. 63.30 (UAR). An auto encoder based unsupervised adaptation model called Universum has been proposed to recognize emotions. Deng et. al. (2017). DNN. 988. IEMOCAP. ELM …

CGMH: Constrained Sentence Generation by Metropolis-Hastings Sampling
N Miao, H Zhou, L Mou, R Yan, L Li – arXiv preprint arXiv:1811.10996, 2018 – arxiv.org
… 2016) and dialog systems (Mou et al. 2016) … However, we can- not find an existing dedicated model. We find it possible to train a variational autoencoder (VAE) with non-parallel corpus and sample sentences from the variational latent space (Bowman et al. 2016) …

A Hierarchical Structured Self-Attentive Model for Extractive Document Summarization (HSSAS)
K Al-Sabahi, Z Zuping, M Nadher – IEEE Access, 2018 – ieeexplore.ieee.org
… For extractive query-oriented single-document summariza- tion, Yousefi-Azar and Hamey [27] used a deep autoencoder to compute a feature space from … Then, a random noise is added to the word representation vector, affecting both the input and output of the auto-encoder …

Incorporating Discriminator in Sentence Generation: a Gibbs Sampling Method
J Su, J Xu, X Qiu, X Huang – arXiv preprint arXiv:1802.08970, 2018 – arxiv.org
… 2014), text summarization (Chopra, Auli, and Rush 2016), dialogue systems (Li et al … Hu et al. (2017) use the variational autoen- coder (VAE) (Kingma and Welling 2013), a regularized ver- sion of standard autoencoder, and employs the wake-sleep algorithm to train the whole …

Selecting and Generating Computational Meaning Representations for Short Texts
C Finegan-Dollak – 2018 – deepblue.lib.umich.edu
Page 1. Selecting and Generating Computational Meaning Representations for Short Texts by Catherine Finegan-Dollak A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy …

A Survey of Multi-View Representation Learning
Y Li, M Yang, ZM Zhang – IEEE Transactions on Knowledge …, 2018 – ieeexplore.ieee.org
… 16], to deep architecture- based methods including multi-modal deep Boltzmann machines [17], multi-modal deep autoencoders [18–20 … Representative examples are multi-modal autoencoder [18], multi-view convolutional neural network [38], and multi-modal recurrent neural …

Towards Faster Annotation Interfaces for Learning to Filter in Information Extraction and Search.
CA Aguirre, S Coen, F Maria, D La Torre, WH Hsu… – IUI …, 2018 – kdd.cs.ksu.edu
… classification task is to help identify low-level features and those that can be identified by modern feature extraction algorithms, such as deep learning autoencoders … Proceedings of the 2nd IJCAI Workshop on Knowledge and Reasoning in Practical Dialogue Systems, (pp …

An Attention-Based Long-Short-Term-Memory Model for Paraphrase Generation
K Nguyen-Ngoc, AC Le, VH Nguyen – International Symposium on …, 2018 – Springer
… tasks in NLP with effective results: language modeling [32], machine translation [4], speech recognition [15], and dialogue systems [26 … Socher, R., Huang, EH, Pennington, J., Ng, AY, Manning, CD: Dynamic pooling and unfolding recursive autoencoders for paraphrase detection …

Identify Shifts of Word Semantics through Bayesian Surprise
Z Wu, C Li, Z Zhao, F Wu, Q Mei – The 41st International ACM SIGIR …, 2018 – dl.acm.org
… of “deep?” “CNN?” “At?” How to detect and adapt to these changes is a critical challenge of natural language processing, which largely affects the performance of downstream tasks such as Web search, information filtering, personalized recommendation, and dialog system …

Symmetric Learning Data Augmentation Model for Underwater Target Noise Data Expansion
M He, H Wang, L Zhou, P Wang, A Ju – … -COMPUTERS MATERIALS & …, 2018 – tspsub.com
… (2014)] is proposed in which the Auto-Encoder or RBM convolution is used to perform unsupervised training tier by tier … 11, no. 3, pp. 625-660. Fischer, JV (2017): A case study on the relevance of the competence assumption for implicature calculation in dialogue systems …

Linguistic alignment classification and generation with deep learning in Spanish conversation
??? – 2018 – s-space.snu.ac.kr
… NLP research enables computers to carry out various tasks such as machine translation and dialogue systems. In particular, when a … main, and can be used to create a dialogue system that generates aligned responses based on user’s speech …

Synthetic speech detection using fundamental frequency variation and spectral features
M Pal, D Paul, G Saha – Computer Speech & Language, 2018 – Elsevier
Skip to main content …

A survey on deep neural network-based image captioning
X Liu, Q Xu, N Wang – The Visual Computer, 2018 – Springer
… ranking techniques based on deep neural net- works make use of deep convolutional neural networks and deep autoencoders to extract … to generate sentences is based on templates with linguistic constraints, which is widely used in summarization [28] and dialogue systems [29 …

A Hierarchy-to-Sequence Attentional Neural Machine Translation Model
J Su, J Zeng, D Xiong, Y Liu, M Wang… – IEEE/ACM Transactions on …, 2018 – dl.acm.org
Page 1. IEEE/ACM TRANSACTIONS ON AUDIO, SPEECH, AND LANGUAGE PROCESSING, VOL. 26, NO. 3, MARCH 2018 623 A Hierarchy-to-Sequence Attentional Neural Machine Translation Model Jinsong Su , Jiali Zeng …

Developing an Affect-Aware Rear-Projected Robotic Agent
A Mollahosseini – 2018 – digitalcommons.du.edu
… 2- Creating an Affect-Aware Robot: Finally, we integrated our automated FER sys- tem into the spoken dialog system of our robotic platform to extend and enrich the capabil- ities of ExpressionBot beyond spoken dialog and create an affect-aware robotic agent that …

Learning Supervised Feature Transformations on Zero Resources for Improved Acoustic Unit Discovery
M Heck, S Sakti, S Nakamura – IEICE TRANSACTIONS on …, 2018 – search.ieice.org
… where the objective is to construct a representa- tion of speech sounds that is robust to variation within and across speakers and that maximizes class discrimination [7]. This task was tackled by a spectrum of contributions: [8] ap- plies a correspondence auto-encoder to learn …

Learning personalized end-to-end goal-oriented dialog
L Luo, W Huang, Q Zeng, Z Nie, X Sun – arXiv preprint arXiv:1811.04604, 2018 – arxiv.org
… Abstract Most existing works on dialog systems only consider conver- sation content while neglecting the personality of the user the bot is interacting with, which begets several unsolved issues … 4 Personalized Dialog System We first propose two personalized models …

Get The Point of My Utterance! Learning Towards Effective Responses with Multi-Head Attention Mechanism.
C Tao, S Gao, M Shang, W Wu, D Zhao, R Yan – IJCAI, 2018 – ijcai.org
… However, previous research on neural gen- erative dialogue systems always generates univer- sal responses, and the attention distribution learned by the model always attends to the same seman- tic aspect … We call it Multi-head Attention Aware Dialog System (MHAM) …

Markov Chain Neural Networks
M Awiszus, B Rosenhahn – arXiv preprint arXiv:1805.00784, 2018 – openaccess.thecvf.com
… a foreseeable reaction given a specific game configuration or (b) always to the same answer for a given comment in a dialog system … In this paper we do not discuss the different variants of neural networks and their possibilities for optimization, autoencoders [17], in- cremental …

CSIndexbr: Exploring the Brazilian Scientific Production in Computer Science
MT Valente, K Paixão, D Hafner, D Tran, A Irpan… – arXiv preprint arXiv …, 2018 – arxiv.org

Reinforcement Learning Policy with Proportional-Integral Control
Y Huang, C Gu, K Wu, X Guan – International Conference on Neural …, 2018 – Springer
… Most of the methods we discuss above adopt standard neural networks like MLPs, single LSTMs or autoencoders as policy network for the non-vision … model-free RL-LSTM framework to solve non-Markovian tasks, and [29] uses LSTM to train an end-to-end dialog systems that is …

A Hierarchical Conditional Attention-Based Neural Networks for Paraphrase Generation
K Nguyen-Ngoc, AC Le, VH Nguyen – International Conference on Multi …, 2018 – Springer
… various tasks in NLP as below: language modeling [34], machine translation [4], speech recognition [15], and dialogue systems [30] … Socher, R., Huang, EH, Pennington, J., Ng, AY, Manning, CD: Dynamic pooling and unfolding recursive autoencoders for paraphrase detection …

BigComp 2018
H Li, X Liang, SH Park, SH Yoon, J Jeong, S Park… – computer.org
… ANE: Network Embedding via Adversarial Autoencoders 66 … 657 Hyeok-june Jung (Konkuk University), Kyeong-sik Park (Konkuk University), Cheol-Jin Kim (Konkuk University), and Young-Guk Ha (Konkuk University) The First International Workshop on Dialog Systems (IWDS) …

Deep Reinforcement Learning for Interactive Narrative Planning.
P Wang – 2018 – repository.lib.ncsu.edu
… Figure 2.1: Plot graph of the interactive fiction, Anchorhead …..15 Figure 2.2: Architecture of a spoken dialogue system with simulated users …..26 … some other fields, like spoken dialogue system, in which statistical user simulation serves an …

A Review on Emotion Recognition Algorithms Using Speech Analysis
TS Gunawan, MF Alghifari, MA Morshidi… – Indonesian Journal of …, 2018 – researchgate.net
… Total 10; 527 real-traffic Mandarin usable utterances from a Microsoft spoken dialogue system. Neutral, happy, sad, angry … [11] W. Fei, X. Ye, S. Zhaoyu, H. Yujia, Z. Xing, and S. Shengxing, “Research on speech emotion recognition based on deep auto-encoder,” in 2016 IEEE …

Fast and Accurate Text Classification: Skimming, Rereading and Early Stopping
K Yu, Y Liu, AG Schwing, J Peng – 2018 – openreview.net
… Iulian Vlad Serban, Alessandro Sordoni, Yoshua Bengio, Aaron C Courville, and Joelle Pineau. Building end-to-end dialogue systems using generative hierarchical neural network models … Semi-supervised recursive autoencoders for predicting sentiment distributions …

An empirical study about the use of Generative Adversarial Networks for text generation
IV Pérez, DA Penas-Padilla, DE Soria-Olivas – 2018 – uv.es
… 24 2.3.2 Variational autoencoders … They empirically proved that given a data set, it was much easier to train a deep autoencoder (a network with a bottleneck that tries to reproduce its input in the output producing a compressed representation) using the greedy layer-wise …

Social Behavior Learning with Realistic Reward Shaping
Y Gao, F Yang, M Frisk, D Hernandez, C Peters… – arXiv preprint arXiv …, 2018 – arxiv.org
… Prior works have utilized deep autoencoders (AEs) to learn a state representation, including Lange et al … To do this, we utilize an autoencoder (AE) [39], a neural net ? that maps inputs to itself, st ?(x) ? x … In the following sections, we refer it as Spatial Auto-encoder Variant (SAEV …

The History Began from AlexNet: A Comprehensive Survey on Deep Learning Approaches
MZ Alom, TM Taha, C Yakopcic, S Westberg… – arXiv preprint arXiv …, 2018 – arxiv.org
… presents a brief survey on development of DL approaches, including Deep Neural Network (DNN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) including Long Short Term Memory (LSTM) and Gated Recurrent Units (GRU), Auto-Encoder (AE), Deep …

Literature Survey and Datasets
S Poria, A Hussain, E Cambria – Multimodal Sentiment Analysis, 2018 – Springer
In this chapter we present the literature on unimodal and multimodal approaches to sentiment analysis and emotion recognition. As discussed in the Sect. 2.1, both of these topics can be brought…

TBCNN for Dependency Trees in Natural Language Processing
L Mou, Z Jin – Tree-Based Convolutional Neural Networks, 2018 – Springer
… models in NLP. It is mostly a consensus that RNNs are better than CNNs in most text generation applications, eg, machine translation, abstractive summarization, and human–computer dialog systems. Nevertheless, CNN might …

StaQC: A Systematically Mined Question-Code Dataset from Stack Overflow
Z Yao, DS Weld, WP Chen, H Sun – arXiv preprint arXiv:1803.09371, 2018 – arxiv.org
Page 1. StaQC: A Systematically Mined Question-Code Dataset from Stack Overflow Ziyu Yao † , Daniel S. Weld # , Wei-Peng Chen †† , Huan Sun † †The Ohio State University, #University of Washington, ††Fujitsu Labs of America …

Dissertations in Forestry and Natural Sciences
A SHOLOKHOV – epublications.uef.fi
… learning SVM Support vector machine TVAE Tied variational autoencoder UBM Universal background model VAE Variational autoencoder VQ Vector … control (eg controlling ac- cess to a bank account or a physical space) and personalization (eg personalized dialogue system) …

Manifold Learning and Nonlinear Recurrence Dynamics for Speech Emotion Recognition on Various Timescales
E Tzinis, ? ?????? – 2018 – researchgate.net
… modality is available [16]. In general, automatic dialogue systems which are empathetic to the inner state of the user are vital assets to a vast amount of applications including surveillance and tutoring agents [17]. Going a step …

Knowledge enhanced hybrid neural network for text matching
Y Wu, W Wu, C Xu, Z Li – Thirty-Second AAAI Conference on Artificial …, 2018 – aaai.org
Page 1. Knowledge Enhanced Hybrid Neural Network for Text Matching Yu Wu, †? Wei Wu, ‡ Can Xu, ‡ Zhoujun Li †?? † State Key Lab of Software Development Environment, Beihang University, Beijing, China ‡ Microsoft …

Reformulating level sets as deep recurrent neural network approach to semantic segmentation
THN Le, KG Quach, K Luu, CN Duong… – IEEE Transactions on …, 2018 – ieeexplore.ieee.org
Page 1. IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 27, NO. 5, MAY 2018 2393 Reformulating Level Sets as Deep Recurrent Neural Network Approach to Semantic Segmentation T. Hoang Ngan Le , Member, IEEE, Kha Gia Quach, Student Member, IEEE …

Towards a Better Metric for Evaluating Question Generation Systems
P Nema, MM Khapra – arXiv preprint arXiv:1808.10192, 2018 – arxiv.org
… More recently, there has been criticism (Liu et al., 2016) for using such metrics for evalu- ating dialog systems eventually resulting in a new metric (Lowe et al., 2017). This new metric while very important, came a bit late in the …

Variational Inference for Data-Efficient Model Learning in POMDPs
S Tschiatschek, K Arulkumaran, J Stühmer… – arXiv preprint arXiv …, 2018 – arxiv.org
… model. Instead of dealing with the prior p(s) directly, variational autoencoders (VAEs) infer p(s) using the posterior p(s|o) (Kingma and Welling, 2013; Rezende et al., 2014) … Reinforcement learning for spoken dialogue systems. In …

Sequence-to-Sequence Models for Emphasis Speech Translation
TQ Do, S Sakti, S Nakamura – IEEE/ACM Transactions on …, 2018 – ieeexplore.ieee.org
… to the encoder. The encoder is pre-trained by appending a linear neural- net layer on top of it with an output size of 1 to predict the emphasis level that is fed into the input layer, similar to an auto-encoder model [26] (Fig. 6 (a …

Automatic Generation of Review Content in Specific Domain of Social Network Based on RNN
Y Tai, H He, WZ Zhang, Y Jia – 2018 IEEE Third International …, 2018 – ieeexplore.ieee.org
… technology has the following problems in terms of generating review for specific fields in social network: 1. There are many mature theories, models in text generation research, which are mostly concentrated in fields of human-machine dialogue system, machine translation …

Multimodal architecture for video captioning with memory networks and an attention mechanism
W Li, D Guo, X Fang – Pattern Recognition Letters, 2018 – Elsevier
… with long sequences. Recently, memory networks [8–10], with the potential to capture long-term correlations in sequential problems, achieve great success in question answering [11] and dialog systems [12]. Among such investigations …

Automatic Generation of News Comments Based on Gated Attention Neural Networks
HT Zheng, W Wang, W Chen, AK Sangaiah – IEEE Access, 2018 – ieeexplore.ieee.org
… Bowman et al. [7] generated sentences from continuous semantic spaces with a variational auto-encoder. Mikolov and Zweig [9] and Wang and Cho [10] improved the performance of language modeling through the long dependency of RNN …

KOGWIS2018: Computational Approaches to Cognitive Science
C Rothkopf, D Balfanz, R Galuske, F Jäkel… – … Conference of the …, 2018 – tu-darmstadt.de
… 46 Zahra Moradi, Keyvan Yahya, Eckart Altenmüller, Boris Kleber and Nelson Mauro Moldanato Common Ground with Dialogue Systems for Fault Diagnosis: How Should They Ask for Clarification?…. 47 Romy Müller and Dennis Paul …

Natural language generation for commercial applications
A van de Griend, W OOSTERHEERT, T HESKES – 2018 – ru.nl
… This master thesis gives an overview on natural language generation with the focus of dialogue systems for commercial use … Fol- lowed by an overview of training methods for these dialogue systems (sec- tion 6.2) and evaluation methods (section 6.3) …

A Review of Computational Approaches for Human Behavior Detection
S Nigam, R Singh, AK Misra – Archives of Computational Methods in …, 2018 – Springer
Page 1. ORIGINAL PAPER A Review of Computational Approaches for Human Behavior Detection Swati Nigam1 • Rajiv Singh2 • AK Misra1 Received: 12 October 2017 / Accepted: 2 May 2018 © CIMNE, Barcelona, Spain 2018 …

Automatic Image Captioning with Style
AP Mathews – 2018 – openresearch-repository.anu.edu.au
Page 1. Automatic Image Captioning with Style Alexander Mathews A thesis submitted for the degree of Doctor of Philosophy The Australian National University November 2018 Page 2. c Alexander Mathews 2018 Page 3. Except …

Chat Discrimination for Intelligent Conversational Agents with a Hybrid CNN-LMTGRU Network
DS Moirangthem, M Lee – Proceedings of The Third Workshop on …, 2018 – aclweb.org
… Recently, intelligent dialog systems and smart assistants have attracted the atten- tion of many, and development of novel dialogue agents have become a research challenge … Dialogue systems can be classified as domain- specific task-oriented and open-domain chit-chat …

Visual coreference resolution in visual dialog using neural module networks
S Kottur, JMF Moura, D Parikh… – Proceedings of the …, 2018 – openaccess.thecvf.com
… believe such consistency in model outputs is a strongly desirable property as we move towards human-machine interaction in dialog systems … to solve visual coreferences [37], and more recently, a probabilistic treatment of dialogs using conditional variational autoencoders [30 …

Representation learning for natural language
O MOGREN – mogren.one
… this, while substantial progress has been made in recent years using dialog systems trained on … Unsupervised feature learning such as autoencoders are examples where this may be the case … both the performance (such as the recon- struction error for the autoencoder), as well …

Latent Topic Conversational Models
TH Wen, MT Luong – 2018 – openreview.net
… Pomdp-based statistical spoken dialog systems: A review. Proceedings of the IEEE, 2013. Tiancheng Zhao, Ran Zhao, and Maxine Eskenazi. Learning discourse-level diversity for neural dia- log models using conditional variational autoencoders. In ACL, pp …

Fast Node Embeddings: Learning Ego-Centric Representations
T Pimentel, A Veloso, N Ziviani – 2018 – openreview.net
… It uses autoencoders to learn a compact representation for nodes based on their adjacency matrix (second-order proximity), while forcing representations of connected nodes to be similar (first-order proximity) by using an hybrid cost function …

Navigating with graph representations for fast and scalable decoding of neural language models
M Zhang, W Wang, X Liu, J Gao, Y He – Advances in Neural …, 2018 – papers.nips.cc
… [14] Iulian Vlad Serban, Alessandro Sordoni, Yoshua Bengio, Aaron C Courville, and Joelle Pineau. Building End-To-End Dialogue Systems Using Generative Hierarchical Neural Network Models. volume 16 of AAAI ’16, pages 3776–3784, 2016 …

Speech Enhancement Based on Bayesian Low-Rank and Sparse Decomposition of Multichannel Magnitude Spectrograms
Y Bando, K Itoyama, M Konyo… – … on Audio, Speech …, 2018 – ieeexplore.ieee.org
Page 1. 2329-9290 (c) 2017 IEEE. Personal use is permitted, but republication/ redistribution requires IEEE permission. See http://www.ieee.org/ publications_standards/publications/rights/index.html for more information. This …

Reinforcement Learning and Robotics
A Vieira, B Ribeiro – Introduction to Deep Learning Business Applications …, 2018 – Springer
… Serban et al. [SSB + 15] recently proposed an end-to-end dialogue system using a generative hierarchical neural network model. The authors proposed a hierarchical recurrent auto-encoder and applied it to a data set (named MovieTriples) containing reviews and comments …

Personalized Prescription for Comorbidity
L Wang, W Zhang, X He, H Zha – International Conference on Database …, 2018 – Springer
… Recently, it is applied to image processing [23], dialog systems [24], machine translation [22] and popularity prediction [25] … propose an unsupervised method to learn the patients representations using a three-layer stack of denoising autoencoders [28] …

Training deep neural networks with non-uniform frame-level cost function for automatic speech recognition
A Becerra, JI de la Rosa, E González… – Multimedia Tools and …, 2018 – Springer
Page 1. Multimed Tools Appl https://doi.org/10.1007/s11042-018-5917-5 Training deep neural networks with non-uniform frame-level cost function for automatic speech recognition Aldonso Becerra1 ·J. Ismael de la Rosa1 · Efrén …

Emotional Conversation Generation Orientated Syntactically Constrained Bidirectional-asynchronous Framework
X Sun, J Li – arXiv preprint arXiv:1806.07000, 2018 – arxiv.org
Page 1. 1 Emotional Conversation Generation Orientated Syntactically Constrained Bidirectional-asynchronous Framework Xiao SUN Member, IEEE, Jingyuan LI Abstract—The field of open-domain conversation generation …

Sentence-State LSTM for Text Representation
Y Zhang, Q Liu, L Song – arXiv preprint arXiv:1805.02474, 2018 – arxiv.org
Page 1. Sentence-State LSTM for Text Representation Yue Zhang1, Qi Liu1 and Linfeng Song2 1Singapore University of Technology and Design 2Department of Computer Science, University of Rochester {yue zhang, qi liu}@sutd.edu.sg, lsong10@cs.rochester.edu Abstract …

Learning Activities From Human Demonstration Videos
J Lee – 2018 – scholarworks.iu.edu
… inputs to corresponding navigational actions for robot visual navigation [41]. The core idea of this paper is to learn how to move a robot (planning) jointly with a spatial memory that corresponds to an egocentric input image (mapping) based on an autoencoder neural network …

Confidence Modeling for Neural Semantic Parsing
L Dong, C Quirk, M Lapata – arXiv preprint arXiv:1805.04604, 2018 – arxiv.org
… errors and inconsis- tencies are difficult to trace. Moreover, from the perspective of application, semantic parsing is of- ten used to build natural language interfaces, such as dialogue systems. In this case it is important to know …

A Multiobjective Learning and Ensembling Approach to High-Performance Speech Enhancement With Compact Neural Network Architectures
Q Wang, J Du, LR Dai, CH Lee – IEEE/ACM Transactions on Audio …, 2018 – dl.acm.org
… unseen noise environments. A separate deep auto encoder (SDAE) to estimate the clean speech and noise spectra by min- imizing the total reconstruction error of noisy speech spectrum was proposed in [12]. Deep recurrent …

Multimodal Sentiment Analysis
S Poria, A Hussain, E Cambria – 2018 – Springer
… Examples of the second domain will include, but not limited to: computational and psychological models of emotions, bodily manifestations of affect (facial expressions, posture, behavior, physiology), and affective interfaces and applications (dialogue systems, games, learning …

Learning Criteria and Evaluation Metrics for Textual Transfer between Non-Parallel Corpora
Y Pang, K Gimpel – arXiv preprint arXiv:1810.11878, 2018 – arxiv.org
… (2017) transfer authorship so as to achieve anonymity. To address the lack of parallel data, Hu et al. (2017) use variational autoencoders to generate content representations devoid of style, which can be converted to sentences with a specific style …

Sentiment analysis method based on an improved modifying?matrix language model
Z Han, F Ren, D Miao – IEEJ Transactions on Electrical and …, 2018 – Wiley Online Library
… For our dialogue system, SCLM can fit the needs much better. 2.2. Supervised machine learning and neural net- works Supervised machine learning is a kind of machine learn- ing where instances are given with known labels or the correspond- ing correct outputs [37] …

Improved Models and Queries for Grounded Human-Robot Dialog
A Padmakumar – cs.utexas.edu
… 2.3 Reinforcement Learning . . . . . 7 2.4 Dialog Systems … 2.4. Dialog Systems Spoken Dialog Systems allow users to interact with information systems with speech as the primary form of communication (Young et al., 2013) …

Deep Learning as Feature Encoding for Emotion Recognition
B Nagarajan, V Oruganti – arXiv preprint arXiv:1810.12613, 2018 – arxiv.org
… Speech emotion recognition has been a growing area of research as more and more intelligent voice assistants and spoken dialogue systems are made … [12] Deng Jun, Xinzhou Xu, Zixing Zhang, Sascha Fruhholz and Bjorn Schuller, “Semisupervised autoencoders for speech …

Deep Learning Approaches to Feature Extraction, Modelling and Compensation for Short Duration Language Identification
S Fernando – 2018 – researchgate.net
Page 1. Deep Learning Approaches to Feature Extraction, Modelling and Compensation for Short Duration Language Identification Sarith Fernando A thesis submitted in fulfilment of the requirement for the degree of Doctor of Philosophy …

Challenges and Experiences in Building an Interactive Enterprise-Scale B2B Recommender System with Natural Language Support
VG Vassiliadis, M Vlachos, T Parnell, C Dünner… – Training – researchgate.net
Page 1. Challenges and Experiences in Building an Interactive Enterprise-Scale B2B Recommender System with Natural Language Support Vassilios G. Vassiliadis IBM Research – Zurich Michail Vlachos IBM Research – Zurich Thomas Parnell IBM Research – Zurich …

Daily Activity Recognition with Large-Scaled Real-Life Recording Datasets Based on Deep Neural Network Using Multi-Modal Signals
T Hayashi, M Nishida, N Kitaoka, T Toda… – IEICE Transactions on …, 2018 – jstage.jst.go.jp
… respectively. Third, we pre-train the DNN using greedy learning with a denoising auto encoder (DAE), in order to appropriately set the initial parameters of the DNN using the normalized, concatenated features. When training Page 5 …

A review on data fusion methods in multimodal human computer dialog
M YANG, J TAO – vr-ih.com
… After decades of development, the man-machine dialogue system has developed from the early telephone voice system, such as language learning, ticket and hotel booking, etc., [2-4] to the current inaccurate questions and answers, such as speech assistant: Apple Siri and …

Classification of Things in DBpedia using Deep Neural Networks
R Parundekar – arXiv preprint arXiv:1802.02528, 2018 – arxiv.org
… representation languages and services. Semantic Graphs are also used in other domains like Spoken Dialog Systems [3], Social Networks [4], Scene Understanding [3], Virtual & Augmented Reality [5], etc. Traditionally, software Agents …

Multimodal Sensing and Data Processing for Speaker and Emotion Recognition using Deep Learning Models with Audio, Video and Biomedical Sensors
F Abtahi – 2018 – academicworks.cuny.edu
Page 1. City University of New York (CUNY) CUNY Academic Works Dissertations, Theses, and Capstone Projects Graduate Center 2-2018 Multimodal Sensing and Data Processing for Speaker and Emotion Recognition using Deep Learning Models with Audio, Video and …

Recurrent Neural Networks
CC Aggarwal – Neural Networks and Deep Learning, 2018 – Springer
“Democracy is the recurrent suspicion that more than half the people are right more than half the time.”— The New Yorker, July 3, 1944.

Coupled context modeling for deep chit-chat: towards conversations between human and computer
R Yan, D Zhao – Proceedings of the 24th ACM SIGKDD International …, 2018 – dl.acm.org
Page 1. Coupled Context Modeling for Deep Chit-Chat: Towards Conversations between Human and Computer Rui Yan1,2 1Institute of Computer Science and Technology Peking University Beijing 100080, China ruiyan@pku.edu.cn …

Proceedings of the 27th International Conference on Computational Linguistics
EM Bender, L Derczynski, P Isabelle – Proceedings of the 27th …, 2018 – aclweb.org
Page 1. COLING 2018 The 27th International Conference on Computational Linguistics Proceedings of the Conference August 20-26, 2018 Santa Fe, New Mexico, USA Page 2. Copyright of each paper stays with the respective authors (or their employers) …

Towards Interpretable Chit-chat: Open Domain Dialogue Generation with Dialogue Acts
W Wu, C Xu, Y Wu, Z Li – 2018 – openreview.net
… customer service, etc. Traditional research on conversational agents focuses on task-oriented dialogue systems (Young et al., 2013) where task specific dialogue acts are handcrafted in a form of slot-value pairs. On the one …

DNN-HMM based Automatic Speech Recognition for HRI Scenarios
J Novoa, J Wuth, JP Escudero, J Fredes… – Proceedings of the …, 2018 – dl.acm.org
… unexpected distortions. As an example, in [52], it was investigated whether the open-source speech recognizer Sphinx can be tuned to outperform Google cloud-based speech recognition API in a spoken dialog system task. By …

Determining speaker attributes from stress-affected speech in emergency situations with hybrid SVM-DNN architecture
J Ahmad, M Sajjad, S Rho, S Kwon, MY Lee… – Multimedia Tools and …, 2018 – Springer
… and their emotions more efficiently [4]. Detecting speaker gender from a brief utterance is a challenging task with rapidly growing applications in communication, human-computer interaction (HCI), telephone speech forensic analysis, and natural language dialog systems …

Efficient Deep Reinforcement Learning via Planning, Generalization, and Improved Exploration
J Oh – 2018 – deepblue.lib.umich.edu
Page 1. Efficient Deep Reinforcement Learning via Planning, Generalization, and Improved Exploration by Junhyuk Oh A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy (Computer Science and Engineering) …

Development of Alexa Voice Services Software Development Kit For Speech Recognition Engine For Internet of Things
N Sharma – 2018 – tudr.thapar.edu
Page 1. DEVELOPMENT OF ALEXA VOICE SERVICES SOFTWARE DEVELOPMENT KIT FOR SPEECH RECOGNITION ENGINE FOR INTERNET OF THINGS A Thesis Submitted in partial Fulfillment of the Requirement for the Award of the Degree of …

Deep Learning for User Simulation in a Dialogue System
FL Kreyssig – mi.eng.cam.ac.uk
Deep Learning for User Simulation in a Dialogue System Florian Lennard Kreyssig Supervisor: Dr. Milica Gašic … Simulation in a Dialogue System Florian L. Kreyssig, Emmanuel College Task-Oriented Spoken Dialogue Systems (SDS) help users achieve goals such as finding …

Analysis, discovery and exploitation of open data for the creation of question-answering systems
G Molina Gallego – 2018 – rua.ua.es
… ALICANTE, 7 de septiembre de 2018 Page 4. Page 5. Abstract Text-based Dialogue Systems have become popular in recent years. More and more companies are using them because they can help in different processes of customer ser- vice … 6 2.3 Dialogue Systems …

KIT-Conferences
MIAR Roedder – 2018 – isl.anthropomatik.kit.edu
… An End-to-End Goal-Oriented Dialog System with a Generative Natural Language Response Generation, Stefan Constantin, Jan Niehues, and Alex Waibel. International Workshop on Spoken Dialogue Systems Technology – IWSDS 2018 …

Contextual Recurrent Level Set Networks and Recurrent Residual Networks for Semantic Labeling
NTH Le – 2018 – search.proquest.com
Contextual Recurrent Level Set Networks and Recurrent Residual Networks for Semantic Labeling. Abstract. Semantic labeling is becoming more and more popular among researchers in computer vision and machine learning …

Augmenting Natural Language Generation with external memory modules in Spoken Dialogue Systems
M Sun – mlsalt.eng.cam.ac.uk
… 2.2 State-of-the-art Natural Language Generation models in Spoken Dialogue Systems 9 … Fig. 2.4 Model structure of combining conditional VAE with SC-LSTM decoder (source: [23]) Variational auto-encoder [11] is an influential generative model which has been widely used …

Adversarial Gain
P Henderson, K Sinha, RN Ke, J Pineau – arXiv preprint arXiv:1811.01302, 2018 – arxiv.org
… 2016. How not to evaluate your dialogue system: An empirical study of unsupervised evaluation metrics for dialogue response generation. arXiv preprint arXiv:1603.08023 … Adversarially regularized autoencoders. Machine translation …

Applying Cooperative Machine Learning to Speed Up the Annotation of Social Signals in Large Multi-modal Corpora
J Wagner, T Baur, Y Zhang, MF Valstar… – arXiv preprint arXiv …, 2018 – arxiv.org
… A whole range of options to choose from exist, such as calculation of ‘meaningful’ con- fidence measures, detecting novelty (eg, by training auto-encoders and seeing for the deviation of input and output when new data runs through the auto- encoder), estimating the degree of …

Automatic quality estimation for ASR system combination
S Jalalvand, M Negri, D Falavigna, M Matassoni… – Computer Speech & …, 2018 – Elsevier
… Voice search engines, voice question answering, broadcast news transcriptions, video/TV programs subtitling, meeting transcriptions and spoken. dialog systems are just some of the many applications involving ASR technology …

Finding Good Representations of Emotions for Text Classification
JH Park – arXiv preprint arXiv:1808.07235, 2018 – arxiv.org
… human emotions. Popular NLP topics like task-oriented dialogue systems or question … learned through either (1) supervised learning (supervised neural networks [33] or semi- supervised methods [35], etc.) or (2) unsupervised learning (autoencoders [4], clustering [11], etc.) …

Tackling Sequence to Sequence Mapping Problems with Neural Networks
L Yu – arXiv preprint arXiv:1810.10802, 2018 – arxiv.org
Page 1. Tackling Sequence to Sequence Mapping Problems with Neural Networks Lei Yu Mansfield College University of Oxford A thesis submitted for the degree of Doctor of Philosophy Trinity 2017 arXiv:1810.10802v1 [cs.CL] 25 Oct 2018 Page 2 …

Explanation in artificial intelligence: Insights from the social sciences
T Miller – Artificial Intelligence, 2018 – Elsevier
Skip to main content …

Human Activity Recognition Based on Wearable Sensor Data: A Standardization of the State-of-the-Art
A Jordao, AC Nazare Jr, J Sena… – arXiv preprint arXiv …, 2018 – arxiv.org
… wearable sensors. In their work, the authors survey a series of deep learning based methods, which includes deep fully-connected networks, recurrent neural networks (RNN) and stacked autoencoders. Stisen et al. [26] investigated …

Improving and Scaling Mobile Learning via Emotion and Cognitive-state Aware Interfaces
P Pham – 2018 – d-scholarship.pitt.edu
Page 1. IMPROVING AND SCALING MOBILE LEARNING VIA EMOTION AND COGNITIVE-STATE AWARE INTERFACES by Phuong Ngoc Viet Pham Bachelor of Science, Vietnam National University, University of Science, 2004 …

Detecting New, Informative Propositions in Social Media
N Dewdney – 2018 – etheses.whiterose.ac.uk
Page 1. DOCTORAL THESIS Detecting New, Informative Propositions in Social Media Author: Nigel Dewdney Supervisor: Prof. Robert GAIZAUSKAS A thesis submitted in fulfillment of the requirements for the degree of Doctor of Philosophy …

Wearable affect and stress recognition: A review
P Schmidt, A Reiss, R Duerichen… – arXiv preprint arXiv …, 2018 – arxiv.org
Page 1. Wearable a ect and stress recognition: A review PHILIP SCHMIDT, ATTILA REISS, and ROBERT DUERICHEN, Robert Bosch GmbH KRISTOF VAN LAERHOVEN, University Siegen A ect recognition aims to detect a …