Time Series Prediction & Dialog Systems

Notes:

“Time Series Prediction” and “Time Series Forecasting” are often used interchangeably to refer to the process of using past data to make predictions about future events. Time series prediction is a common task in many fields, including finance, economics, and meteorology, and can be used to make predictions about a wide range of phenomena, including stock prices, weather patterns, and economic indicators.

Time series prediction typically involves analyzing past data points, such as prices or temperature readings, over a series of time intervals, and using this data to build a model that can be used to make predictions about future events. There are a variety of different techniques and approaches that can be used for time series prediction, including statistical models, machine learning algorithms, and artificial neural networks.

Time series prediction can be a useful tool for a variety of different applications, including financial forecasting, weather forecasting, and demand forecasting. By analyzing past data and making informed predictions about future events, it is possible to make more informed decisions and better prepare for future challenges and opportunities.

Time series prediction can be used in dialog systems, also known as conversational agents or chatbots, to make predictions about future events and provide relevant and timely responses to user queries.

For example, a dialog system might use time series prediction to forecast future stock prices, weather patterns, or other types of data, and provide this information to users in real-time. The system might also use time series prediction to anticipate user needs and provide relevant recommendations or suggestions based on this analysis.

Time series prediction can be a useful tool for improving the accuracy and relevance of responses provided by a dialog system, and can help the system provide more valuable and engaging experiences for users. By analyzing past data and making informed predictions about future events, a dialog system can provide more timely and relevant responses, and can help users make better informed decisions.

Wikipedia:

Efficiently trainable text-to-speech system based on deep convolutional networks with guided attention
H Tachibana, K Uenoyama… – 2018 IEEE International …, 2018 – ieeexplore.ieee.org
… Most of the existing methods above use RNN, a natural tech- nique of time series prediction … a standard technique for mapping a sequence to another sequence, especially in the field of natural language processing, eg machine transla- tion [17, 18], dialogue system [19, 20], etc …

Prediction of turn-taking using multitask learning with prediction of backchannels and fillers
K Hara, K Inoue, K Takanashi, T Kawahara – Listener, 2018 – isca-speech.org
… separated models for the subjects and for the robot, but we focus on the behavior of the robot (dialogue system) for evaluation … Prediction models are trained for each inter- locutor role in each dialog corpus (eg interviewer (robot) and interviewee (subject) in the … time series data …

A Neural Network Architecture Combining Gated Recurrent Unit (GRU) and Support Vector Machine (SVM) for Intrusion Detection in Network Traffic Data
AFM Agarap – Proceedings of the 2018 10th International Conference …, 2018 – dl.acm.org
… This amendment was seen as viable for the fast prediction time of SVM … Approach Combining Recurrent Neural Network and Support Vector Machines for Time Series Classification … Semantically conditioned lstm-based natural language generation for spoken dialogue systems …

Predicting engagement breakdown in hri using thin-slices of facial expressions
T Liu, A Kappas – Workshops at the Thirty-Second AAAI Conference on …, 2018 – aaai.org
… IEEE. Bohus, D., and Horvitz, E. 2009. Learning to predict en- gagement with a spoken dialog system in open-world set- tings … Li, D.; Han, M.; and Wang, J. 2012. Chaotic time series prediction based on a novel robust echo state network …

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
… the time series is long, it cannot reflect past information well. To overcome this issue, the LSTMs algorithm was applied to control the use of cell state information by employing a forget gate [31]. Yuan et al. [32] proposed a fault diagnosis and residual useful life prediction method …

Multimodal local-global ranking fusion for emotion recognition
PP Liang, A Zadeh, LP Morency – Proceedings of the 2018 on …, 2018 – dl.acm.org
… It has im- mense applications towards robotics [2, 15], dialog systems [20, 21], intelligent tutoring … Fig- ure 2) aims to integrate both direct and relative emotion prediction approaches to … usion allows us to capture the temporal dependencies across multimodal time series data and …

Wind Speed Prediction Model Using LSTM and 1D-CNN
R Fukuoka, H Suzuki, T Kitajima… – Journal of Signal …, 2018 – jstage.jst.go.jp
… How- ever, there is a time-delay prediction error[1]. In this paper, we propose a new wind speed prediction system with long short-term memory … It can learn the long-term dependence of time series data, and it is often used for sen- tence generation and dialogue system …

Audio-Visual Prediction of Head-Nod and Turn-Taking Events in Dyadic Interactions
BB Türker, E Erzin, Y Yemez, M Sezgin – Proc. Interspeech 2018, 2018 – iui.ku.edu.tr
… Since the model is sequential, frame based time-series features (fA and fB) are used … Head-nod prediction experiments showed that speech activity features have potential to improve the … R. Rzepka, and K. Araki, “Activating humans with humor–a dialogue system that users want …

First Insights on a Passive Major Depressive Disorder Prediction System with Incorporated Conversational Chatbot
F Delahunty, ID Wood – ceur-ws.org
… We have trained a dialogue system, powered by sequence-to-sequence neural networks that can have a real-time … Passive Major Depressive Disorder Prediction System 5 … upon the work of [30] who identified users with suicidal tendencies by applying a time series approach to …

Mongolian Text-To-Speech System Based on Deep Convolutional Network with Guided Attention
O Jargalsaikhan, Z Byambadorj, T Erdene-Ochir – researchgate.net
… Most of the existing methods above use RNN, a natural technique for time series prediction … been a standard technique to map a sequence into another sequence, especially in the field of natural language pro- cessing, eg machine translation [17, 18], dialogue system [19, 20 …

First insights on a passive major depressive disorder prediction system with incorporated conversational chatbot
M Arcan, ID Wood, F Delahunty – Irish Conference on Artificial …, 2018 – library.nuigalway.ie
… We have trained a dialogue system, powered by sequence-to-sequence neural networks that can have a real-time … Passive Major Depressive Disorder Prediction System 5 … upon the work of [37] who iden- tified users with suicidal tendencies by applying a time series approach to …

Textual Summarization of Time Series using Case-based Reasoning: A Case Study
N Dubey, S Chakraborti, D Khemani – ICCBR 2018 – cse.iitm.ac.in
… Language Engineering 14(04), 431–455 (2008) 4. Langner, B., Black, AW: MOUNTAIN: A Translation-based Approach to Natural Language Generation for Dialog Systems 5. Miura, N … Sripada, S., Reiter, E., Hunter, J., Yu, J.: Segmenting time series for weather forecasting …

Spiking Neural Networks for Early Prediction in Human Robot Collaboration
T Zhou, JP Wachs – arXiv preprint arXiv:1807.11096, 2018 – arxiv.org
… Spiking Neural Networks for Early Prediction in Human Robot Collaboration … In spoken dialogue systems, turn-taking is detected by finding short pauses (usually between 0.5 to 1 second (Ferrer et al., 2002)) and they indicate the current speaker’s intent to yield the turn …

RITS: Real-Time Interactive Text Steganography Based on Automatic Dialogue Model
Z Yang, P Zhang, M Jiang, Y Huang… – … Conference on Cloud …, 2018 – Springer
… of negotiating and ultimately reaching a transaction, they design an automatic dialogue system that can … overcome the problem of long distance dependence, and realize the modeling of long time series … To be more specific, we define the Prediction Weight (PW) as matrix \(W_P …

Advanced Data Analytics Using Python: With Machine Learning, Deep Learning and NLP Examples
S Mukhopadhyay – 2018 – dl.acm.org
… in domains such as high-frequency algorithmic trading and goal-oriented dialog systems … covers important traditional data analysis techniques such as time series and principal … analysis techniques such as classification, clustering, regression, and forecasting Handle structured …

Artificial Intelligence for Conversational Robo-Advisor
MY Day, JT Lin, YC Chen – 2018 IEEE/ACM International …, 2018 – ieeexplore.ieee.org
… Therefore, this study focuses on using a time series DL model for stock market forecasting and integrates several theories of asset allocation to provide portfolio … Dialogue systems have always been an important topic in the field of human–computer interaction research …

Human Capacity—Biopsychosocial Perspective
B Xing, T Marwala – Smart Maintenance for Human–Robot Interaction, 2018 – Springer
… scenario concerned by Ferreira and Lefèvre (2015) is MaRDi dialogue system which was … Usually, the time series prediction is capable of handling errors-in variable problem with … Network Architectures: The most widely used ANN structure in time series forecasting has been the …

INISTA’2018 Conference Detailed Program
V Piuri, L Heller, M Tsikerdekis, SE Constructor… – inista.org
… Marios Krestenitis, Zoe Doulgeri Real-Time Event Detection in Time-Series Classification Based on … full) Nikolaos Kolokas, Thanassis Vafeiadis, Dimosthenis Ioannidis, Dimitrios Tzovaras Forecasting Faults of … the Analysis of a Dialogue Corpus to a Dialogue System (full) Tu?ba …

Residual Recurrent Highway Networks for Learning Deep Sequence Prediction Models
T Zia, S Razzaq – Journal of Grid Computing, 2018 – Springer
… Page 7. Residual Recurrent Highway Networks for Learning Deep Sequence Prediction Models Fig … Serban, IV, Sordoni, A., Bengio, Y., Courville, AC, Pineau, J.: Building end-to-end dialogue systems using gen- erative hierarchical neural network models. In: AAAI, pp …

Deep Learning and Neural Networks
S Mukhopadhyay – Advanced Data Analytics Using Python, 2018 – Springer
… prediction in the airline domain, passenger load in month t is heavily dependent on t-12 months of data rather on t-1 or t-2 data. Hence, the neural network normally produces a better result than the time-series model or even image classification. In a chatbot dialogue system, the …

Conversational Temporal Coherence
J Sedoc – seas.upenn.edu
… This builds on my work on time series multiscale models [12, 6]. This model attempts to capture … How not to evaluate your dialogue system: An empirical study of unsupervised evaluation metrics for dialogue response … Multiscale hidden markov models for covariance prediction …

Exploration and Mining Learning Robot of Autonomous Marine Resources Based on Adaptive Neural Network Controller
L Pan – Polish Maritime Research, 2018 – content.sciendo.com
… Neural network prediction of time series is usually based on existing sample data to train the … commonly used to construct one-step prediction model and multi-step prediction model are … 83 6. IV Serban, and A. Sordoni, Building End-To-End Dialogue Systems Using Generative …

Modeling multiple time series annotations as noisy distortions of the ground truth: An Expectation-Maximization approach
R Gupta, K Audhkhasi, Z Jacokes, A Rozga… – IEEE transactions on …, 2018 – ncbi.nlm.nih.gov
… 5], [6]. However, often times the variable of interest may not be directly observable (such as in behavioral time series of psychological states … We demonstrate the effectiveness of the proposed algorithm in a study involving prediction of time continuous confidence ratings of smile …

Neuro-symbolic reasoning system for modelling complex behaviours
N Balse, J Carter – researchgate.net
… This chapter describes the application of a hybrid artificial intelligence approach to prediction in the domain of oceanography … However, time series forecasting, based on neural network or statistical analysis, may not provide sufficiently accurate forecasting capability in chaotic …

Prediction of a hotspot pattern in keyword search results
J Gao, A Radeva, C Shen, S Wang, Q Wang… – Computer Speech & …, 2018 – Elsevier
… Prediction of keyword hotspots could potentially support better prediction of speech recognition quality for conversational speech or dialog systems (Goldwater et al … learning due to their wide applicability, and the increasing availability of large time series datasets (Xu et …

KNADIA: Enterprise KNowledge Assisted DIAlogue Systems Using Deep Learning
M Singh, P Agarwal, A Chaudhary… – 2018 IEEE 34th …, 2018 – ieeexplore.ieee.org
… The NADIA natural dialogue system [14] (note that this prior work has no relation to ours, despite the similarity of its name) makes the process of … the predictions made by the classifier are not right (ie, the bot throws an error), or the confidence of the prediction (calculated based …

Discourse Marker Detection for Hesitation Events on Mandarin Conversation
YW Wang, HH Huang, KY Chen… – Proc. Interspeech …, 2018 – nlg.csie.ntu.edu.tw
… For instance, dialogue systems can provide more explanation to users sounding more uncertain, and a reminding system can detect users’ difficulties in memory … For the text prediction model, the increase on recall is limited … [13] G. Gosztolya, “Optimized Time Series Filters for …

Intention-Based Anticipatory Interactive Systems
A Wendemuth, R Boeck, A Nuernberger… – … on Systems, Man …, 2018 – ieeexplore.ieee.org
… of cognitive functions and on using these to realize the desired system characteristics, mainly adaptivity, prediction of user … based on a temporal integration of the image sequences and a fuzzy-based time-series analysis of … “Ten challenges in highly-interactive Dialog Systems” …

Exploring the use of artificial intelligence in price maximisation in the tourism sector: its application in the case of Airbnb in the Valencian Community
L Moreno-Izquierdo, G Egorova, A Peretó Rovira… – 2018 – rua.ua.es
… time series models», Economic Modelling, 36, 220-228. Claveria, O., Monte, E., and Torra, S. (2015): «Tourism demand forecasting with neural net- work models: different ways of … Limsombunchai, V., Gan, C., and Lee, M. (2004): «House Price Prediction: Hedonic Price Model …

ICBK 2018
G Li, X Fu, X Ren – computer.org
… Short-Attention Mechanism for Generative Dialogue System 268 … Prediction of Aluminum Electrolysis Superheat Based on Improved Relative Density Noise Filter SMO 376 … Matrix Profile XIII: Time Series Snippets: A New Primitive for Time Series Data Mining 382 …

A review on data fusion methods in multimodal human computer dialog
M YANG, J TAO – vr-ih.com
… Similarly, in recent human computer dialog system, different kind of channel information is converted into behavior recognition, and then the system gives … graph model can be used not only in uncertainty calculation, but also in decision-making reasoning for time series problems …

Coqa: A conversational question answering challenge
S Reddy, D Chen, CD Manning – arXiv preprint arXiv:1808.07042, 2018 – arxiv.org
… Following SQuAD, we use macro-average F1 score of word overlap as our main evaluation metric.7 In SQuAD, for computing a model’s performance, each individual prediction is compared against n human answers resulting in n F1 scores, the max- imum of which is chosen as …

A Joint Introduction to Natural Language Processing and to Deep Learning
L Deng, Y Liu – Deep Learning in Natural Language Processing, 2018 – Springer
… of the deep learning techniques popular for current spoken language understanding and dialogue systems as well as … codes, etc.) or from multiple cross-domain tasks (eg, point and structured prediction, ranking, recommendation, time-series forecasting, clustering, etc.) …

THU NGN at NAACL-2018 Metaphor Shared Task: Neural Metaphor Detecting with CNN-LSTM Model
C Wu, F Wu, Y Chen, S Wu, Z Yuan, Y Huang – researchgate.net
… the metaphors in texts are important to mine the semantic and sentiment information bet- ter, which is beneficial to many applications such as machine translation, dialog systems and senti … A neural network ensemble method with jittered training data for time series forecasting …

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 …
… TTS Backstory EVI Knowledge Base Data Scraper on EC2 animals movies news retrieval Figure 1: Social Bot Framework Figure 1 depicts the social bot dialog system framework … 3.3.6 Dialog Act Prediction Each segmented sentence from NLU is associated with a dialog act …

Counseling Robot Implementation and Evaluation
K Kurashige, S Tsuruta, E Sakurai… – … on Systems, Man …, 2018 – ieeexplore.ieee.org
… Further, it is important to consider both emotion and time series at once … prefix-based prediction with Recurrent Neural Networks (RNNs) for Japanese text prediction capacity based … 20 ‘ nodding timing after client’s utterance (when dumbness is detected) dialogue system CRECA …

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] … RGAN and RCGAN [13] can generate real-valued time-series data … Also, tableGAN explicitly optimizes the prediction accuracy on synthetic data by minimizing cross …

Web and Big Data: APWeb-WAIM 2018 International Workshops: MWDA, BAH, KGMA, DMMOOC, DS, Macau, China, July 23–25, 2018, Revised Selected …
H Xie – 2018 – books.google.com
… and Jiajie Xu Sequence-As-Feature Representation for Subspace Classification of Multivariate Time Series … Wu A Semantic Role Mining and Learning Performance Prediction Method in … Contents XIII MOOC Guider: An End-to-End Dialogue System for MOOC …

Speech emotion recognition: two decades in a nutshell, benchmarks, and ongoing trends
BW Schuller – Communications of the ACM, 2018 – dl.acm.org
… has become increasingly ‘chatty’ these days: Alexa, Cortana, Siri, and many more dialogue systems have hit … such as MFCC or mel-frequency bands as well as linear prediction coefficients.2 … one usually derives statistics by apply- ing functionals that map a time series of …

Web and Big Data
H Xie – Springer
… Jiajie Xu Sequence-As-Feature Representation for Subspace Classification of Multivariate Time Series … Wenjun Wu A Semantic Role Mining and Learning Performance Prediction Method in … MOOC Guider: An End-to-End Dialogue System for MOOC Users …

Long Short Term Memory Hyperparameter Optimization for a Neural Network Based Emotion Recognition Framework
B Nakisa, MN Rastgoo, A Rakotonirainy, F Maire… – IEEE …, 2018 – ieeexplore.ieee.org
… solutions in an acceptable time. It has been shown that PSO is efficient in finding the optimal number of input, hidden nodes and learning rate on time series prediction problems [48]. Although some studies have applied EC …

The Design and Implementation of XiaoIce, an Empathetic Social Chatbot
L Zhou, J Gao, D Li, HY Shum – arXiv preprint arXiv:1812.08989, 2018 – arxiv.org
… The development of social chatbots, or intelligent dialogue systems that are able to engage in empathetic conversations with humans … IQ capacities include knowledge and memory modeling, image and natural language understanding, reasoning, generation and prediction …

Enhancing Text Using Emotion Detected from EEG Signals
A Gupta, H Sahu, N Nanecha, P Kumar, PP Roy… – Journal of Grid …, 2018 – Springer
… Similarly, Long Short Term Memory (LSTM) networks have been used exten- sively in time series prediction [40], speech recogni- tion [15], grammar learning [14], protein homology detection [18], handwriting recognition [16], etc …

Learning Factorized Multimodal Representations
YHH Tsai, PP Liang, A Zadeh, LP Morency… – arXiv preprint arXiv …, 2018 – arxiv.org
… This interpretation method will help to visualize contributions from individual modalities that occur only in particular short segments of a time series. Both methods allow us to analyze how individual factors in MFM influence the dynamics of multimodal prediction and generation …

Experimental Analysis of Emotion Classification Techniques
T Dumitriu, C Cîmpanu, F Ungureanu… – 2018 IEEE 14th …, 2018 – ieeexplore.ieee.org
… Some entropy algorithms derived from Shannon’s definition were proposed to obtain suitable indices for quantifying various types of time series … Boosting represents an approach to machine learning based on the idea of creating a highly accurate prediction rule through the …

End-to-End Learning Artificial Intelligence
E Labintcev, H Jabbar, A Sieler, C Holland – di-lab.tum.de
… For instance, in dialog systems one could use reinforcement learning definitions to setup dialog [9] for two agents, where both of … Furthermore, there are some classical problems with time dependency such as time series prediction, which can be formulated as reinforcement …

Multi-Layer Ensembling Techniques for Multilingual Intent Classification
C Costello, R Lin, V Mruthyunjaya, B Bolla… – arXiv preprint arXiv …, 2018 – arxiv.org
… 2013. Easy contextual intent prediction and slot detection … 2017. Combin- ing cnns and pattern matching for question interpretation in a virtual patient dialogue system … 1995. Convolutional networks for images, speech, and time-series …

Incorporating Background Knowledge into Video Description Generation
S Whitehead, H Ji, M Bansal, SF Chang… – Proceedings of the 2018 …, 2018 – aclweb.org
Page 1. Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 3992–4001 Brussels, Belgium, October 31 – November 4, 2018. c 2018 Association for Computational Linguistics 3992 …

Recurrent Deep Multiagent Q-Learning for Autonomous Brokers in Smart Grid.
Y Yang, J Hao, M Sun, Z Wang, C Fan, G Strbac – IJCAI, 2018 – nos.netease.com
… to categorize the electricity consumers (eg C-vine mixture model clustering (CVMM) [Sun et al., 2017]), in time series analysis, DTW is … Our broker clusters consumers into 5 groups, this number if selected by obtaining the highest prediction accuracy for load forecasting …

Attention-Based CNN-BLSTM Networks for Joint Intent Detection and Slot Filling
Y Wang, L Tang, T He – … and Natural Language Processing Based on …, 2018 – Springer
… two problems which are difficult to solve when propagating backwards in the time series [25] … the input sequence for intent detection and slot filling in task-oriented dialog systems in the … 3. Liu, B., Line, I.: Recurrent neural network structured output prediction for spoken language …

Engagement Recognition based on Multimodal Behaviors for Human-Robot Dialogue
K Inoue – 2018 – repository.kulib.kyoto-u.ac.jp
… such as social networking services [27–29]. Spoken dialogue systems have also been studied from the perspective of multimodal … It has also been found that eye-gaze and body movement features were informative for prediction of turn-taking behaviors [36– 39] …

Multimodal Speech Emotion Recognition Using Audio and Text
S Yoon, S Byun, K Jung – arXiv preprint arXiv:1810.04635, 2018 – arxiv.org
… its applications, as it is a crucial factor in optimal human- computer interactions, including dialog systems … to the formation of the network’s internal hidden state ht to model the time series patterns … this result, we note that textual data are informative in emotion prediction tasks, and …

Speech-Based Emotion Recognition: Linguistic and Saliency-Based Systems
KW Gamage – 2018 – unsworks.unsw.edu.au
… The main contributions of this thesis revolve around the use of verbal and non-verbal vocalisation cues for speech-based emotion recognition, which is complementary to popularly used acoustic features for both emotion classification and continuous emotion prediction tasks …

ICMLA 2017
VM Tavano – ieeexplore.ieee.org
… Anomaly Detection in Multivariate Non-stationary Time Series for Automatic DBMS Diagnosis 412 Doyup Lee (POSTECH) … Prediction of Power Grid Failure Using Neural Network Learning 505 … A Review of Deep Learning Methods Applied on Load Forecasting 511 …

A Generative Model for category text generation
Y Li, Q Pan, S Wang, T Yang, E Cambria – Information Sciences, 2018 – Elsevier
… 1). Most of them consider sentence generation as a process of character prediction and use RNN for feature extraction from time series data [40 … challenges, as well as in the business world, due to the remarkable benefits to be had from marketing and financial forecasting [47 …

A Conversation between Theory, Methods, and Data
SM Boker, M Martin – Multivariate behavioral research, 2018 – Taylor & Francis
… These three types of matrices are, respectively, the data organizations underlying cross-sectional, person-specific, and time series analyses … For a variety of reasons, a prediction for an individual cannot be inferred from population statistics (Molenaar, 2004 Molenaar, PCM (2004 …

Detecting Memory-Based Interaction Obstacles with a Recurrent Neural Model of User Behavior
F Putze, M Salous, T Schultz – 23rd International Conference on …, 2018 – dl.acm.org
… of detect- ing memory-based interaction obstacles without resorting to additional sensors by modeling time series of user … Learning to forget: Continual prediction with LSTM … In Proceedings of the Paralinguistic Information and its Integration in Spoken Dialogue Systems Workshop …

Towards a Novel Approach to High-accuracy Enterprise Search
S Guo, L Wang, Z Wang, N Shu… – … and Digital Content …, 2018 – ieeexplore.ieee.org
… prediction using process metrics elasticsearch engine case study … 2016:254-260 [3] Min Chang , Yuansheng Lou , Lei Qiu .An approach for time series similarity search … Hao Su ,et al.Improved TF-IDF Weight Approach based on SentenceSimilarity for SpoNen Dialogue System …

Institute of Communications Engineering Staff
M Bossert, R Fischer, W Minker, UC Fiebig… – Journal on Multimodal …, 2018 – uni-ulm.de
… Dialogue Systems … N. Rach, K. Weber, L. Pragst, E. André, W. Minker and S. Ultes EVA: A Multimodal Argumentative Dialogue System Accepted for presentation at the 20th ACM International Conference on Multimodal Interaction, Boulder, Colorado, October 2018 Bibtex …

Research on Optimization of Big Data Construction Engineering Quality Management Based on RNN-LSTM
D Wang, J Fan, H Fu, B Zhang – Complexity, 2018 – hindawi.com
… The dialogue system and the machine translation system also have time series features in the collection of architectural text information during the construction period. Therefore, the recurrent neural network adapts to the real-time forecasting tasks of construction project quality …

Continuous density hidden markov model for hindi speech recognition
S Sinha, SS Agrawal, A Jain – GSTF Journal on Computing (JoC), 2018 – dl6.globalstf.org
… expanded from simplest system of digit recognition to spontaneous dialogue systems.Such growth is … inclusion enhances the system performance[8]. • PLP-Perceptual linear prediction technique is … method of characterizing the observed data samples in the time series that can …

Generative Stock Question Answering
Z Tu, X Liu, L Shu, S Shi – arXiv preprint arXiv:1804.07942, 2018 – arxiv.org
… stock knowledge base KB. Most of the questions in StockQA are “what” and “how” questions, such as asking for stock trend prediction (ie, what- type) and action recommendation (ie, how-type). To generate reasonable answers …

Hierarchical RNN for Few-Shot Information Extraction Learning
S Liu, Y Li, B Fan – … of Pioneering Computer Scientists, Engineers and …, 2018 – Springer
… Language Processing (NLP) tasks (eg sentiment analysis, natural language translation and dialog systems) … fragment level, each fragment will be assigned an attribute label after prediction … neural networks as generative models – reconstructing gaps in time series, April 2015 …

Recurrent Neural Network for Predicting Transcription Factor Binding Sites
Z Shen, W Bao, DS Huang – Scientific reports, 2018 – nature.com
… a comprehensive application of ChIP-seq and DNase-seq 27,34,35,36,37,38,39,40,41 , which ensure the prediction accuracy of … Since recurrent neural networks (RNN) 66 can effectively extract feature information from time-series data, it has been widely used in the process of …

Information Sciences and Technologies Bulletin of the ACM Slovakia
M Blšták, P Laurinec, M Bystrický, R Šelmeci, D Bernát – 2018 – acmbulletin.fiit.stuba.sk
… for question answering task [10, 12], for build- ing databases of frequently-asked question section [16], for finding out if lectures are understandable [9] or for creat- ing a dialog system in role of … One of the main tasks of time series data mining is forecasting future values …

Induction of Emotional States in Educational Video Games through a Fuzzy Control System
CA Lara, H Mitre-Hernandez, J Flores… – IEEE Transactions on …, 2018 – ieeexplore.ieee.org
… valence, arousal, and dominance. Support Vector Machine for Regression with a polynomial kernel is adopted for learning. The out- put of this component is a prediction for each of the emotion primitives. 2.3 General Accuracy The …

Gender and Emotion Recognition Using Voice
P Rani, M Geeta – csjournals.com
… of a speaker gender is important for increasingly natural and personalized dialogue systems … variety of statistical (linear and nonlinear modeling, classical statistical tests, time-series analysis, classification … Fig:3 Gender and Emotion Prediction System using in data flow diagram …

Hierarchical recurrent highway networks
T Zia – Pattern Recognition Letters, 2018 – Elsevier
Skip to main content …

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
… Additionally, the loss function of MDN-RNN comprises two components: the prediction error in next state Ls and the prediction error in … Since the latent representations of observations in environment are in time series with the episode going, the correlation information as well as …

Research on Optimization of Big Data Quality Management Based on RNN-LSTM Construction Engineering
D Wang, J Fan, H Fu – downloads.hindawi.com
… Processing (NLP) The dialogue system and the machine translation system also have time-series features in the collection of architectural text information during the construction period. Therefore, the recurrent neural network adapts to the real-time forecasting tasks of …

The Short-term User Modeling for Predictive Applications
M Kompan, O Kassak, M Bielikova – Journal on Data Semantics, 2018 – Springer
… An average user session consists of only few actions, which brings several complications for the user modeling and also for subsequent prediction tasks … We evaluate our model by the task of session end intent prediction in the e-learning and news domain …

Early Turn-Taking Prediction for Human Robot Collaboration
T Zhou – 2018 – search.proquest.com
… 88 Page 13. xii Figure 4.15 Software architecture of the real-time turn-taking prediction. …. 90 … Page 16. xv ??? Exponentially Weighted Moving Average (EWMA) of ?? over time series ? weight of current measurement when calculating ??? …

Early Drought Plant Stress Detection with Bi-Directional Long-Term Memory Networks
H Li, Z Yin, P Manley, JG Burken… – … & Remote Sensing, 2018 – ingentaconnect.com
… First, we extract the time-series image patch sequences that contain the temporal variation information of the plant as patch sequences … The patch sequence classification is the BLSTM model’s prediction on the input patch sequence …

Sensornet: A scalable and low-power deep convolutional neural network for multimodal data classification
A Jafari, A Ganesan, CSK Thalisetty… – … on Circuits and …, 2018 – ieeexplore.ieee.org
… the raw sensor signals) as input and consists of 5 convolutional layers, 3 pooling layers, 1 dense layer and a softmax layer (for prediction) … In summary, most of the previous work do not propose a real-time hardware solution for multimodal time-series data classification or use …

Tree2Tree Learning with Memory Unit
N Miao, H Wang, R Le, C Tao, M Shang, R Yan… – 2018 – openreview.net
… It is widely used in various tasks, such as time series analysis(Lipton et al., 2015), speech recognition … Here the loss function in our model is defined as the cross entropy between prediction and target … Tree-based language model dedicated to natural spoken dialog systems …

On the Duration, Addressability, and Capacity of Memory-Augmented Recurrent Neural Networks
Z Quan, Z Gao, W Zeng, X Li, M Zhu – IEEE Access, 2018 – ieeexplore.ieee.org
Page 1. 2169-3536 (c) 2018 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/ redistribution requires IEEE permission. See http://www.ieee.org …

Towards Understanding User Requests in AI Bots
OT Tran, TC Luong – Pacific Rim International Conference on Artificial …, 2018 – Springer
… [21] presented a general solution towards building task-oriented dialogue systems for online … In the prediction phase, to predict tags of one layer, we must use trained models to predict tags of the … LeCun, Y., Bengio, Y.: Convolutional networks for images, speech, and time series …

Detecting Zero-day Controller Hijacking Attacks on the Power-Grid with Enhanced Deep Learning
Z He, A Raghavan, S Chai, R Lee – arXiv preprint arXiv:1806.06496, 2018 – arxiv.org
… The “accu- rate prediction” of the controller’s behavior is not necessary, since we are looking for the difference between the prediction error distri- butions … ? Collect the prediction error distribution (Reconstruction Error Distribution) D1 of the baseline behavior …

How the experts do it: Assessing and explaining agent behaviors in real-time strategy games
J Dodge, S Penney, C Hilderbrand… – Proceedings of the …, 2018 – dl.acm.org
… Army tab: Shows supply and resource value of currently held non-worker units. Present High 1 1 Income Advantage graph: Shows time series data comparing resource gain rate. Past High 1 1 Units Killed popup: Essentially the opposite of the Units Lost popup Past High 1 1 …

A critical review and analysis on techniques of speech recognition: The road ahead
AV Haridas, R Marimuthu… – International Journal of …, 2018 – content.iospress.com
… This enhanced phoneme prediction precision, when incorporated into standard large vocabulary incessant … the three novel language modeling techniques that employ semantic study for spoken dialog systems … of the SLDS scales exponentially with the length of the time series …

Prototypical recurrent unit
D Long, R Zhang, Y Mao – Neurocomputing, 2018 – Elsevier
… These networks are demonstrated as the state-of-the-art models for time series or sequence data [1], [10], [22] … Using these three kinds of recurrent unit, we not only experiment on constructing a standard language model for character prediction [19], but also test the recurrent …

Emotion extraction based on multi bio-signal using back-propagation neural network
G Yoo, S Seo, S Hong, H Kim – Multimedia Tools and Applications, 2018 – Springer
… We analyzed several emotions according to the time series analysis data acquired in LF … sensors: first step towards an automatic system, in affective dialogue systems tutorial and … W (2015) Relevance units machine based dimensional and continuous speech emotion prediction …

Scale development using Twitter data: applying contemporary natural language processing methods in IS research
D Agogo, TJ Hess – Analytics and Data Science, 2018 – Springer
… machine translation, speech recognition and speech synthesis have been widely applied in consumer products such as spoken dialogue systems (SDS) (eg … 3276CrossRefGoogle Scholar. Gayo-Avello D (2013) A meta-analysis of state-of-the-art electoral prediction from Twitter …

A globally generalized emotion recognition system involving different physiological signals
M Ali, F Al Machot, A Haj Mosa, M Jdeed, E Al Machot… – Sensors, 2018 – mdpi.com
Machine learning approaches for human emotion recognition have recently demonstrated high performance. However, only/mostly for subject-dependent approaches, in a variety of applications like advanced driver assisted systems, smart homes and medical environments. Therefore …

A Bi-Encoder LSTM Model for Learning Unstructured Dialogs
D Shekhar – 2018 – digitalcommons.du.edu
… growing rapidly. A Dialog System can communicate with human in text, speech or both … in the lowest cost or loss. A cost function represent the price paid for inaccuracies of prediction in classification done by the machine learning task. The brief description of the …

NeuroSpeech: An open-source software for Parkinson’s speech analysis
JR Orozco-Arroyave, JC Vásquez-Correa… – Digital Signal …, 2018 – Elsevier
… non-Parkinson’s speech signals (if the user has access to recordings of other pathologies, he/she can re-train the system to perform the detection of other diseases), and (3) the prediction of the neurological state of the patient according to the Unified Parkinson’s Disease Rating …

Deep Reinforcement Learning For Sequence to Sequence Models
Y Keneshloo, T Shi, CK Reddy… – arXiv preprint arXiv …, 2018 – arxiv.org
Page 1. 1 Deep Reinforcement Learning For Sequence to Sequence Models Yaser Keneshloo, Tian Shi, Naren Ramakrishnan, Chandan K. Reddy, Senior Member, IEEE Abstract—In recent years, sequence-to-sequence (seq2seq …

A computational model for the emergence of turn-taking behaviors in user-agent interactions
M Jégou, P Chevaillier – Journal on Multimodal User Interfaces, 2018 – Springer
… [20] created a model dedicated to the prediction of the … This issue relates to observations that dialog systems often falsely detect user’s utterances as barge-ins, whereas the latter is only providing feedback to the system by producing a vocal backchannel or is not speaking to the …

Recurrence Quantification Models of Human Conversational Grounding Processes: Informing Natural Language Human-Computer Interaction
CD Rothwell – 2018 – rave.ohiolink.edu
… Conversational grounding is also a crucial process for human-computer interaction using language-based methods, such as spoken dialogue systems … 2013, 2004; Clark and Krych, 2004), making it a high priority for increasing capabilities of spoken dialogue systems …

ICDE 2018
S Amer-Yahia, PA Bernstein, M Stonebraker, A Ilyas… – computer.org
… State University) Efficient Learning Interpretable Shapelets for Accurate Time Series Classification 497 Zicheng Fang (Fudan University), Peng Wang (Fudan University), and Wei Wang (Fudan University) Learning Association …

Towards an automatic evaluation of the dysarthria level of patients with Parkinson’s disease
JC Vásquez-Correa, JR Orozco-Arroyave… – Journal of …, 2018 – Elsevier
Skip to main content …

Unfolding Recurrent Neural Networks
P Goyal, S Pandey, K Jain – Deep Learning for Natural Language …, 2018 – Springer
… Unsuitable for sequences, time series data, video streaming, stock data, etc … cases, we want the model to consider past historical events and make a prediction regarding future … seq2seq) models are used for everything from chatbots to speech-to-text to dialog systems to QnA to …

Sentiment Analysis for Tweets
K Korovesis – aueb.gr
Page 1. Athens 2018 MSc in Computer Science Department of Computer Science Master of Science Thesis Sentiment Analysis for Tweets Konstantinos Korovesis Page 2. 1 Acknowledgements I would like to thank my supervisor Prof …

Deep Learning using Rectified Linear Units (ReLU)
AF Agarap – arXiv preprint arXiv:1803.08375, 2018 – arxiv.org
… Values of correct prediction in the matrices seem to be balanced, as in some … approach combining recurrent neural network and support vector machines for time series classification … Semantically conditioned lstm-based natural language generation for spoken dialogue systems …

Real-time Face Detection and Recognition Based on Deep Learning
H Wang – 2018 – aut.researchgateway.ac.nz
… (mAP) on object detection, but it cannot be achieved real-time detection due to requirement of two stages in each prediction (Wu, Iandola, Jin, & Keutzer, 2017) … predictions for each default boxes, respectively, a discrete class prediction for each …

Multisource Transfer Double DQN Based on Actor Learning
J Pan, X Wang, Y Cheng, Q Yu – IEEE transactions on neural …, 2018 – ieeexplore.ieee.org
… Since there is a strong correlation among time-series data obtained by the RL agent, the continuous observation and updating of online learning will create instability for the agent. The offline data storage and experience replay …

Typing Tutor: Individualized Tutoring in Text Entry for Older Adults Based on Statistical Input Stumble Detection
T Hagiya, T Horiuchi, T Yazaki… – Journal of Information …, 2018 – jstage.jst.go.jp
… However, Kurni- awan [17] reported that older adults usually dislike text-prediction features … The possible stumble opportunity was labeled when an annotator judged that an input stumble is likely to occur based on the time series information be- fore an input, such as the unfixed …

Possible Scientific-Technical Solutions to the Problem of Giving Early Warning
S Petrenko – Big Data Technologies for Monitoring of Computer …, 2018 – Springer
… Dialog systems models and methods … data classification, identifying primary and secondary signs of cyber-attack, early cyber-attack detection, multifactor prediction of cyber … Data, stored in the repository, is a series of records characterized by a time stamp (time series); thus, the …

A network analytic approach to gaze coordination during a collaborative task
S Andrist, AR Ruis, DW Shaffer – Computers in Human Behavior, 2018 – Elsevier
… analytic technique, as it permits the visualization and quantification of recurrent patterns of states between two time series—in this … Prediction accuracy could easily be improved, however, by employing more sophisticated methods than we used here for demonstration purposes …

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
… E(o, y) = ? t ||ot ? yt||2 (2) where ot is prediction and yt is labeled groundtruth … Ix,y(1 ? H(?))dxdy ? (1 ? H(?))dxdy (7) Under this formulation, the curve evolution is shown as a time series process which helps to give better visualization of reformulating LS …

An improved formula for Jacobi rotations
CF Borges, Z Majdisova, V Skala, A Monszpart… – arXiv preprint arXiv …, 2018 – arxiv.org
… Comments: Interspeech 2018. Subjects: Computation and Language (cs.CL). arXiv:1806.08047 [pdf, other] Title: Flexible Neural Representation for Physics Prediction …

Recurrent Neural Networks
CC Aggarwal – Neural Networks and Deep Learning, 2018 – Springer
… For example, in a time-series forecasting application, we might need outputs at each time-stamp in order to predict the next value in the time-series … In Figure 7.4, we have shown the probabilistic prediction of the next word at each of time-stamps from 1 to 4. Ideally, we …

Classification of Emoji Categories from Tweet Based on Deep Neural Networks
K Matsumoto, M Yoshida, K Kita – … of the 2nd International Conference on …, 2018 – dl.acm.org
… the sentences are setimated, it would be useful for emoji recommendation, text classification based on emojis and category prediction of unknown … BiGRU have advantages that they can process word sequences in chronological order, and can train the time series bi-directionally …

Database Systems for Advanced Applications
J Pei, Y Manolopoulos, S Sadiq, J Li – 2018 – books.google.com
… 738 Xuanyu Bai, Jianguo Yao, Mingyuan Yuan, Jia Zeng, and Haibing Guan Client Churn Prediction with Call Log Analysis … 764 Wenjing Fang, Jun Zhou, Xiaolong Li, and Kenny Q. Zhu Cost-Sensitive Churn Prediction in Fund Management Services …

Can We Speculate Running Application With Server Power Consumption Trace?
Y Li, H Hu, Y Wen, J Zhang – IEEE transactions on cybernetics, 2018 – ieeexplore.ieee.org
… In this paper, we propose a novel distance measurement and build a time series classification algorithm hybridizing nearest neighbor and long short term … In particular, we com- bine the prediction probability vector of 1NN-LTW and LSTM to determine the label of the test cases …

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
… 3): 1. Absence of a human expert (navigation on Mars) 2. Human are unable to explain their expertise (speech recognition, vision and language understanding) 3. The solution to the problem changes over time (tracking, weather prediction, preference, stock, price prediction) 4 …

The Analytics Lifecycle Toolkit: A Practical Guide for an Effective Analytics Capability
GS Nelson – 2018 – books.google.com
… methods: ? Concepts k ? Business intelligence and reporting k ? Big data ? Data science ? Edge (and ambient) analytics ? Informatics ? Artificial intelligence and cognitive computing ? Methods ? Applied statistics and mathematics ? Forecasting and time series ? NLP, text …

A Case-based Reasoning Model for Depression based on Three-electrode EEG Data
H Cai, X Zhang, Y Zhang, Z Wang… – IEEE Transactions on …, 2018 – ieeexplore.ieee.org
Page 1. 1949-3045 (c) 2018 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 …

Data-Driven Input Feature Augmentation for Named Entity Recognition
??? – 2018 – s-space.snu.ac.kr
… For example, in an automatic restaurant search query system, the dialog system may have … arate model for each label is trained, and the class with the highest predicted probability is chosen as the final output prediction, or a variation known as …

Event2Mind: Commonsense Inference on Events, Intents, and Reactions
H Rashkin, M Sap, E Allaway, NA Smith… – arXiv preprint arXiv …, 2018 – arxiv.org
… For example, an ideal dialogue system should react in empathetic ways by reasoning about the human user’s mental state based on the events the … and “OReact” columns), we see that much of the im- provement in using these two models is within the prediction of PersonX’s …

RNN-Stega: Linguistic Steganography Based on Recurrent Neural Networks
Z Yang, X Guo, Z Chen, Y Huang… – IEEE Transactions on …, 2018 – ieeexplore.ieee.org
… related work, like image cap- tioning [34], [37], neural machine translation [35], dialogue systems [38], and … the characteristics of RNN that we can model each sentence as a time series signal, and … To be more specific, we define the Prediction Weight(PW) as matrix WP ? Rnl ×N …

Real World User Model: Evolution of User Modeling Triggered by Advances in Wearable and Ubiquitous Computing
F Cena, S Likavec, A Rapp – Information Systems Frontiers, 2018 – Springer
… The most commonly mea- sured aspect of personalisation quality is accuracy of rating prediction … among them, for example among physical states and context, or habits and chronic diseases, as well as to examine their evolution over time, eg through time-series analyses and …

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, detection of … Research in New York City, and his team built core machine learning libraries, released the Google Prediction API, and …

Direction of Arrival Estimation and Localization of Multi-Speech Sources
N Dey, AS Ashour – 2018 – Springer
… Some of the topics covered in this series include the presentation of real life commercial deployment of spoken dialog systems, contemporary methods of speech parameterization, developments in information … Prediction of binaural speech intelligibility against noise in rooms …

Machine learning: the new’big thing’for competitive advantage
M Attaran, P Deb – … Journal of Knowledge Engineering and Data …, 2018 – researchgate.net
… Statistical algorithms, ML, data warehousing, text analytics, and time series analytics are just … from speech recognition, natural language generation and understanding, dialog systems, and connecting … chain operations and more accurate demand and price forecasting can be …

Analysis of Delays in Continuous Annotations and Physiological Signals for Affective State Characterization using Machine Learning Algorithms
YA Pirkani, D Braun, PDDF Schwenker – researchgate.net
… of human emotion might vary from person to person which gives inconsistency in the emotion prediction … dimension. Dimensional emotion models are very effective in the continuous emotion prediction where, the emotions can change significantly over time …

Exploration and optimized siting of geothermal wells using a web-based spatial decision support system
DW Mwaura – 2018 – depositonce.tu-berlin.de
Page 1. Exploration and Optimized Siting of Geothermal Wells Using a Web-Based Spatial Decision Support System A Case Study of the Olkaria Geothermal Field vorgelegt von M.Sc. Daniel Waweru Mwaura geb. in Nairobi, Kenia von der Fakultät VI – Planen Bauen Umwelt …

Data-driven development of Virtual Sign Language Communication Agents
A Balayn, H Brock, K Nakadai – 2018 27th IEEE International …, 2018 – ieeexplore.ieee.org
… Recognition of complex and rare sentences is not accurate enough yet and a sentence language model could support the sentence prediction … 38, no. 3, pp. 2661–2667, 2011. [8] S. Lu, S. Igi, H. Matsuo, and Y. Nagashima, “Towards a dialogue system based on recognition and …

Driver Behavior and Environment Interaction Modeling for Intelligent Vehicle Advancements
Y Zheng – 2018 – libtreasures.utdallas.edu
… investigated in the design of a navigation dialogue system. The accuracy for intent detection (ie … Model (HMM) has been created. It considers time series dynamic changes in the lower level, and consecutive maneuver relationship in the upper level. This model has been …

Advanced Data Analytics Using Python
S Mukhopadhyay – Springer
… Curve Fitting …..122 Removing Trends from a Time Series …..123 … Removing Seasonality from a Time Series …..125 …

Neural Task Planning with And-Or Graph Representations
T Chen, R Chen, L Nie, X Luo, X Liu… – IEEE Transactions on …, 2018 – ieeexplore.ieee.org
… Index Terms—Scene understanding, Task planning, Action prediction, Recurrent neural network … II. RELATED WORK We review the related works following three main research streams: task planning, action recognition and prediction, and recurrent sequence prediction …

Fact or Fiction
A Lu, CJ Lovering, N Dinh, C Tri, T Nguyen, H Bui – 2018 – digitalcommons.wpi.edu
… RNNs have had success with text generation, classification, and time series forecasting (?Gers?, 1999) … Any client application that converses with users, such as a dialog system or a chatbot, can pass user input to a LUIS app and receive results that provide natural language …

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 …

The Natural Auditor: How To Tell If Someone Used Your Words To Train Their Model
C Song, V Shmatikov – arXiv preprint arXiv:1811.00513, 2018 – arxiv.org
… computes the probability Pr(yj|y1,…,yj?1; x) as f(y1,…,yj?1; x). Similar to the next-word prediction task, the … of P. Successful membership inference attacks against aggregate statistics have been demonstrated in the context of genomic studies [3], [16], location time-series [28], and …

A dynamic deep learning approach for intonation modeling
F Tombini – 2018 – publikationen.sulb.uni-saarland.de
… Page 14. 2 is bound to become more critical in the coming years, especially as the demand for TTS applications such as dialog systems, virtual agents, and personal assistants increases … of time-series prediction, the contextual information is limited by the size of the input layer …

A Review of Artificial Intelligence in the Internet of Things
CG García, ER Núñez-Valdez, V García-Díaz… – researchgate.net
… Some examples of the use of these algorithms are in the field of automated designs of equipment and machinery, in game theory, in the learning of Fuzzy Logic rules, natural language processing and prediction, among others …

Nav view search
H Li, HA Patil, N Evans, K Evanini… – 2018 – isca-speech.org
… language processing. q Speech-to-text, text-to-speech and speech-to-speech processing. q Machine translation and dialogue systems. q Application of spoken language technologies for under-resourced languages. Important …

Nonlinear Semi-Supervised Metric Learning Via Multiple Kernels and Local Topology
X Li, Y Bai, Y Peng, S Du, S Ying – International journal of neural …, 2018 – World Scientific
Page 1. 2 October 4, 2017 19:5 1750040 International Journal of Neural Systems, Vol. 27, No. 0 (2017) 1750040 (22 pages) c© World Scientific Publishing Company DOI: 10.1142/S012906571750040X Nonlinear Semi-Supervised …

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
… Moreover, MOE-DNN can also be compact because the prediction is relatively easy when given prior awareness of the auxiliary information … We used MFCC features to improve the prediction of clean LPS features in our previous work [25] …

System-Level Design of GPU-Based Embedded Systems
A Maghazeh – 2018 – books.google.com
Page 1. Linköping Studies in Science and Technology. Dissertations. No. 1964 System-Level Design of GPU-Based Embedded Systems Arian Maghazeh Page 2. Linköping Studies in Science and Technology. Dissertations. No …

Natural Language Generation with Neural Variational Models
H Pallikara Bahuleyan – 2018 – uwspace.uwaterloo.ca
… 19 2.2.9 Dialog Systems … The chapter concludes with the related work in the tasks of question generation and dialog systems, which are the two experiments conducted in this study, to evaluate the proposed model. 2.1 Machine Learning …

Natural Language Generation with Neural Variational Models
H Bahuleyan – arXiv preprint arXiv:1808.09012, 2018 – arxiv.org
… 17 2.2.7 Variational Inference . . . . . 18 2.2.8 Question Generation . . . . . 19 2.2.9 Dialog Systems . . . . . 19 3 Sequence to Sequence Models 21 3.1 Word Embeddings …

Improving and Scaling Mobile Learning via Emotion and Cognitive-state Aware Interfaces
P Pham – 2018 – d-scholarship.pitt.edu
… 52 Table 5. Quiz error prediction performance. The standard deviation showed in parenthesis ….. 53 … 28 Figure 7. From the PPG time series (above), multiple sliding signal windows were extracted (below) …

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 …

Deep Learning Approaches to Feature Extraction, Modelling and Compensation for Short Duration Language Identification
S Fernando – 2018 – researchgate.net
… The BLSTM framework is also evaluated using different front-end features and a frequency domain linear prediction based front-end which is robust to channel mismatch is proposed … FDLP Frequency Domain Linear Prediction FFT Fast Fourier Transform …

Artificial Intelligence for All: An Abiding Destination
V Pathak, P Tiwari – 2018 – books.google.com
Page 1. ARTIFICIAL INTELLIGENCE FOR ALL AN ABIDING DESTINATION º Nº. -º- 2 º nº 2 wº º: DR. PANKAT TIWARI VIKAS PATHAIK Page 2. Artificial Intelligence For All Artificial Intelligence For All i Page 3. Artificial Intelligence …

Investigating multi-modal features for continuous affect recognition using visual sensing
H Wei – 2018 – doras.dcu.ie
… 137 A.3.2 Affect Prediction . . . . . 144 … 87 4.7 Plot of valence delay for dynamic features . . . . . 87 4.8 Plot of post processed prediction on arousal dimension using LBP- TOP features …

Autonomous agents modelling other agents: A comprehensive survey and open problems
SV Albrecht, P Stone – Artificial Intelligence, 2018 – Elsevier

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
… 53 Figure 23. Framework of the proposed neural network with VGG Net and ROI Nets 53 Figure 24. Feature extraction via CNN and prediction using LSTM 55 Figure 25. Comparison of emotion recognition accuracies on all modalities using DBN and LSTM 58 Figure 26 …

Using Polarity Classification Model to Assess Customer Attitudes: the Case of Russian E-Commerce Companies on Twitter
T Alexander – 2018 – dspace.spbu.ru
… ones may be limited to nine: classification and class probability estimation, regression analysis, similarity matching, clustering, co-occurrence grouping, profiling, link prediction, data reduction, and causal modeling (Provost, 2013) …

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 …

On the Interplay between Global Optimization and Machine Learning
F Bagattini, F Schoen, M Sciandrone, L Chisci – flore.unifi.it
… In particular, the only output of a GO algorithm is the prediction of the global optimum … This allows to get rid of the unknown label variables by replacing them with the expression of their prediction. The semi-supervised objective function (2.1) becomes …

Scalable and Efficient Probabilistic Topic Model Inference for Textual Data
M Magnusson – 2018 – books.google.com
… In addition, in this thesis a supervised topic model for high-dimensional text classification is also proposed, with emphasis on interpretable document prediction using the horseshoe shrinkage prior in supervised topic models …

Detecting New, Informative Propositions in Social Media
N Dewdney – 2018 – etheses.whiterose.ac.uk
… 43 3.6 Statisticalapproachesinknowledgediscovery . . . . . 45 3.7 Application of probabilistic models . . . . . 47 3.8 Machine Learning in Classification and Prediction . . . . . 49 3.9 AdvancesinMachineLearning …

Matrix Factorization Methods For Training Embeddings In Selected Machine Learning Problems
A Fonarev, I Oseledets – skoltech.ru
… rithms are widely used in applications related to text meaning understanding, eg, in automated dialog systems … complex structures and entailments in natural language, which also is widely used in automated dialog systems …

A New Approach to the Analysis of Cooperation Under the Shadow of the Future: Theory and Experimental Evidence
M Kartal, W Müller – 2018 – papers.ssrn.com
… Our results also show that standard theory may have poor predictive power even if it makes a unique prediction … uniquely predicts defection in all of our treatments, observed behavior can be far from this prediction even after players gain a large amount of experience …

From Synchronous to Asynchronous Event-driven Fusion Approaches in Multi-modal Affect Recognition
F Lingenfelser – 2018 – opus.bibliothek.uni-augsburg.de
… from facial and vocal modalities (smiles and laughs). . . . 134 8.6 Frequency of correctly classified frames according to laugh / smile confidence. Similar prediction behaviour allows to directly combine confidence values during the fusion process. . . . . 135 …

Artificial Intelligence And Machine Learning And Marketing Management
J Seligman – 2018 – books.google.com
… A difficult task in business is the prediction of future sales which in turn defines future revenues and profits … From managing leads as a conduit for improving forecasting, a digital assistant takes maintenance of data management …

PacGAN: The power of two samples in generative adversarial networks
AK Khetan – 2018 – ideals.illinois.edu
… crisp, and original examples of images [3, 6] and text [7]. This is useful in image and video processing (eg frame prediction [8], image super-resolution [9], and image-to-image transla- tion [10]), as well as dialogue systems or chatbots—applications where one may need realistic …

Contextual Recurrent Level Set Networks and Recurrent Residual Networks for Semantic Labeling
NTH Le – 2018 – search.proquest.com
… network cover more relevant information. This is very important for tasks which need. a larger receptive eld when making the prediction. Transposed convolution [66] can be. seen as the backward pass of a corresponding traditional convolution. It is also known as. deconvolution …