RNN (Recurrent Neural Network) & Question Answering Systems 2015

See also:

100 Best Recurrent Neural Network Videos | RNN (Recurrent Neural Network) & Dialog Systems 2014 | RNN (Recurrent Neural Network) & Dialog Systems 2015

Ask your neurons: A neural-based approach to answering questions about images M Malinowski, M Rohrbach, M Fritz – Proceedings of the IEEE …, 2015 – cv-foundation.org … Recurrent Neural Networks allow Neural Networks to handle sequences of flexible length. … In [20], we present a question answering system based on a semantic parser on a more var- ied set of human question-answer pairs. … Cited by 64 Related articles All 13 versions

Towards ai-complete question answering: A set of prerequisite toy tasks J Weston, A Bordes, S Chopra, AM Rush… – arXiv preprint arXiv: …, 2015 – arxiv.org … We compared the following methods on our tasks (on the English dataset): (i) an N- gram classifier baseline, (ii) LSTMs (long short term memory Recurrent Neural Networks) (Hochreiter & Schmidhuber, 1997), (iii) Memory Networks (MemNNs) (Weston et al., 2014), (iv) some … Cited by 112 Related articles All 3 versions

Exploring models and data for image question answering M Ren, R Kiros, R Zemel – Advances in Neural Information …, 2015 – papers.nips.cc … Our VIS+LSTM and Malinkowski et al.’s recurrent neural network model [14] achieved somewhat similar performance on DAQUAR. … complete set of baselines which can provide potential insight for developing more sophisticated end-to-end image question answering systems. … Cited by 55 Related articles All 10 versions

Teaching machines to read and comprehend KM Hermann, T Kocisky, E Grefenstette… – Advances in Neural …, 2015 – papers.nips.cc … 8 Page 9. References [1] Ellen Riloff and Michael Thelen. A rule-based question answering system for reading com- prehension tests. … [8] Karol Gregor, Ivo Danihelka, Alex Graves, and Daan Wierstra. DRAW: A recurrent neural network for image generation. … Cited by 110 Related articles All 9 versions

Visual7w: Grounded question answering in images Y Zhu, O Groth, M Bernstein, L Fei-Fei – arXiv preprint arXiv:1511.03416, 2015 – arxiv.org … Traditional question answering system relies on an elabo- rate pipeline of models involving natural language parsing, knowledge base querying, and … The proposed models range from MLP classifiers [1], probabilistic infer- ence [24, 41, 48] and recurrent neural networks [1, 7, 25 … Cited by 20 Related articles All 5 versions

The Goldilocks Principle: Reading Children’s Books with Explicit Memory Representations F Hill, A Bordes, S Chopra, J Weston – arXiv preprint arXiv:1511.02301, 2015 – arxiv.org … As we show, state-of-the-art language modelling architectures, Recurrent Neural Networks (RNNs) with Long-Short Term Memory … well a language model can lend semantic coherence to applications including machine translation, dialogue and question-answering systems. … Cited by 28 Related articles All 7 versions

Applying deep learning to answer selection: A study and an open task M Feng, B Xiang, MR Glass, L Wang… – 2015 IEEE Workshop …, 2015 – ieeexplore.ieee.org … potential for practical use. Index Terms— Answer Selection, Question Answering, Convolutional Neural Network (CNN), Deep Learning, Spo- ken Question Answering System 1. INTRODUCTION Natural language understanding … Cited by 12 Related articles All 4 versions

Advances in natural language processing J Hirschberg, CD Manning – Science, 2015 – science.sciencemag.org … For translation, research has focused on a particular version of recurrent neural networks, with enhanced “long short-term memory” computational … The flip side of machine reading is to provide question-answering systems, by which humans can get answers from constructed … Cited by 21 Related articles All 9 versions

Sirius: An open end-to-end voice and vision personal assistant and its implications for future warehouse scale computers J Hauswald, MA Laurenzano, Y Zhang, C Li… – ACM SIGPLAN …, 2015 – dl.acm.org … Watson system [14]. OpenEphyra’s NLP techniques, including conditional random field (CRF), have been rec- ognized as state-of-the-art and are used at Google and in other industry question-answering systems [20]. We design … Cited by 28 Related articles All 26 versions

Neural Self Talk: Image Understanding via Continuous Questioning and Answering Y Yang, Y Li, C Fermuller, Y Aloimonos – arXiv preprint arXiv:1512.03460, 2015 – arxiv.org … nents include a Visual Question Generation (VQG) module and a Visual Question Answering module, in which Recurrent Neural Networks (RNN) and … to generate reasonable and relevant questions, and 2) by in- corporating it with a visual question answering system, a sys- tem … Cited by 5 Related articles All 4 versions

Evaluating prerequisite qualities for learning end-to-end dialog systems J Dodge, A Gane, X Zhang, A Bordes, S Chopra… – arXiv preprint arXiv: …, 2015 – arxiv.org … Recurrent Neural Networks (RNNs) have proven successful at several tasks involving natural language, language modeling (Mikolov et al., 2011), and have been applied recently to dialog (Sordoni et al., 2015; Vinyals & Le, 2015 … 3.4 QUESTION ANSWERING SYSTEMS … Cited by 10 Related articles All 3 versions

Learning statistical scripts with LSTM recurrent neural networks K Pichotta, RJ Mooney – … of the 30th AAAI Conference on …, 2015 – pdfs.semanticscholar.org … will have a job inter- view,” and perhaps “Smith will find a job.” Inferences of this type are required for robust question-answering systems. … We describe a novel statistical script model which uses a recurrent neural network with a Long Short- Term Memory (LSTM) architecture. … Cited by 7 Related articles All 5 versions

A hybrid neural model for type classification of entity mentions L Dong, F Wei, H Sun, M Zhou, K Xu – Proceedings of the 24th International …, 2015 – ijcai.org … The mention model uses recurrent neural networks to recursively obtain the vector representation of an entity mention from the words it … Moreover, when type information provided by our method is used in a question answering system, we observe a 14.7% relative improvement … Cited by 5 Related articles All 7 versions

A Deep Architecture for Semantic Matching with Multiple Positional Sentence Representations S Wan, Y Lan, J Guo, J Xu, L Pang, X Cheng – arXiv preprint arXiv: …, 2015 – arxiv.org … Long short term memory (LSTM) is an advanced type of Recurrent Neural Network by further using memory cells and gates to learn … Answers which is a community question answering system where some users propose questions to the system and other users will submit their … Cited by 9 Related articles All 8 versions

Natural language object retrieval R Hu, H Xu, M Rohrbach, J Feng, K Saenko… – arXiv preprint arXiv: …, 2015 – arxiv.org … the user may ask to a robot to pick up “the TV remote con- trol on the shelf” and visual question answering systems where the answers … with other types of visual-linguistic models such as bag-of-words [24], one of the advantages of us- ing a recurrent neural network as scoring … Cited by 14 Related articles All 4 versions

A C-LSTM Neural Network for Text Classification C Zhou, C Sun, Z Liu, F Lau – arXiv preprint arXiv:1511.08630, 2015 – arxiv.org … Recurrent neural networks (RNNs) are able to prop- agate historical information via a chain-like neu- ral network architecture. … Question type classification: Question classifica- tion is an important step in a question answering system that classifies a question into a specific type … Cited by 3 Related articles All 3 versions

Semeval-2015 task 3: Answer selection in community question answering PNLMWM AlessandroMoschitti, J Glass, B Randeree – SemEval-2015, 2015 – aclweb.org … Another popular technique was to use word embed- dings, eg, modeled using convolution or recurrent neural networks, or with latent semantic analysis, and also vectors trained using word2vec and GloVe (Pennington et al., 2014), as pre-trained on Google News or Wikipedia … Cited by 3 Related articles All 14 versions

A central pattern generator for controlling sequential activation in a neural architecture for sentence processing D van Dijk, F van der Velde – Neurocomputing, 2015 – Elsevier … The model illustrated in Fig. 1 is not the only neural model of sentence processing. Alternative models are models based on dynamics systems (eg, [12]), reservoir computing (eg, [21]) or recurrent neural networks (eg, [37]). But … Cited by 1 Related articles All 4 versions

Second Exam: Natural Language Parsing with Neural Networks J Cross – 2015 – pdfs.semanticscholar.org … nouns, proper nouns, etc.). 4 Recurrent Neural Networks Another powerful instance of a neural network architecture is the recurrent neural net- work. … selected aspects of the earlier inputs. The general structure of a recurrent neural network layer can be seen in Figure 6. … Related articles All 2 versions

Image-Based Question Answering with Visual-Semantic Embedding M Ren – 2015 – cs.utoronto.ca … In recent years, computational models, pariticularly large-scale deep convolutional and recurrent neural networks, have achieved em- … to achieve good performance. Building an image-based question answering system has profound implication in the field of … Related articles All 3 versions

Classifying responses on online discussion forums A Abajian – stanford.edu … A variety of models have been trained using decision trees [1], support vectors machines (SVMs) [6] [7], and recurrent neural networks [4]. The majority of these models are trained using … This leniency makes it difficult to use raw forum data to train a question answering system. … Related articles All 4 versions

Backbone Language Modeling for Constrained Natural Language Generation L Mou, R Yan, G Li, L Zhang, Z Jin – arXiv preprint arXiv:1512.06612, 2015 – arxiv.org … The recurrent neural network (RNN) is a prevailing class of language models; it is suitable for modeling time-series data (eg, a … For example, a question answering system may involve analyzing the question and querying an existing knowledge base, to the point of which, a … Related articles All 2 versions

A Novel Hierarchical Convolutional Neural Network for Question Answering over Paragraphs S Zheng, H Bao, J Zhao, J Zhang… – 2015 IEEE/WIC/ACM …, 2015 – ieeexplore.ieee.org … are Recurrent/Recursive Neural Networks (RNN) and Convo- lutional Neural Network (CNN). Recurrent Neural Networks deal successfully with time- series data and they were also applied on NLP [18], [19] by modeling a sentence as tokens processed sequentially. … Related articles

Named Entity Recognition and Question Answering Using Word Vectors and Clustering ZAR Veluri – pdfs.semanticscholar.org … The results obtained in this work suggests that the NER system used in aiding question answering system benefits from including … Named Entity Recognition using recurrent neural networks (RNN) and Long Short Term Memory (LSTM) is also a promising future direction and … Related articles All 2 versions

[BOOK] Natural Language Processing and Chinese Computing: 4th CCF Conference, NLPCC 2015, Nanchang, China, October 9-13, 2015, Proceedings J Li, H Ji, D Zhao, Y Feng – 2015 – books.google.com … 12 Chenxi Zhu, Xipeng Qiu, and Xuanjing Huang Recurrent Neural Networks with External Memory for Spoken Language Understanding … 520 Kerui Min, Chenggang Ma, Tianmei Zhao, and Haiyan Li Research on Open Domain Question Answering System….. …

Statistical Script Learning with Recurrent Neural Nets K Pichotta – 2015 – cs.utexas.edu … 15 4 Script Learning with Recurrent Neural Networks 18 … What we mean by “we would like to infer” this fact is that a competent automated Question Answering system should be able to answer “yes” to the query “Did Octavian defeat Antony?” A Statistical Script model encodes … Related articles All 3 versions

Identifying synonymy between relational phrases using word embeddings NTH Nguyen, M Miwa, Y Tsuruoka, S Tojo – Journal of biomedical …, 2015 – Elsevier Many text mining applications in the biomedical domain benefit from automatic clustering of relational phrases into synonymous groups, since it alleviates the p. Related articles All 8 versions

CS224N Final Project: QA TL Wu, DA Huang – nlp.stanford.edu … In principle this is pos- sibly achievable by a language modeler such as a recurrent neural network (Hochreiter and Schmid- huber, 1997), as … For example, an impor- tant cue for question answering system to pick the correct supporting facts is to recognize the subject in the … Related articles All 2 versions

[BOOK] Artificial Intelligence: A Systems Approach MT Jones – 2015 – books.google.com … Backpropagation 274 Training Variants 274 Weight Adjustment Variants 274 Probabilistic Neural Networks 275 PNN Algorithm 276 PNN Implementation 277 Other Neural Network Architectures 281 Time Series Processing Architecture 281 Recurrent Neural Network 283 Tips … Cited by 201 Related articles All 9 versions

Temporal Information Extraction Extracting Events and Temporal Expressions A Literature Survey N Gupta – 2015 – cfilt.iitb.ac.in … 1 ) as shown in Figure-3.1. The function g is implemented either by feed-forward neural network or recurrent neural network or another parametrized function, with parameters ?. So overall parameter set is ? = (C, ?). Training … Related articles

Cross-Lingual Cross-Media Content Linking: Annotations and Joint Representations (Dagstuhl Seminar 15201) AG Hauptmann, J Hodson, J Li, N Sebe… – Dagstuhl …, 2015 – drops.dagstuhl.de … when used to jointly model images or videos along with captions using deep learning approaches like Recurrent neural networks and multimodal log … developed and are the basis of applications ranging from the support of search to the realization of question answering systems. … Related articles All 3 versions

LN-Annote: An Alternative Approach to Information Extraction from Emails using Locally-Customized Named-Entity Recognition YH Jung, K Stratos, LP Carloni – … of the 24th International Conference on …, 2015 – dl.acm.org … remote access to the cloud). LN-Annote enables third-party service providers to build a question-answering system on top of the local personal information without having to own the user data. In addition, LN-Annote mitigates … Cited by 2 Related articles All 10 versions

Lexical Resource for Medical Events: A Polarity Based Approach A Mondal, I Chaturvedi, D Das, R Bajpai… – … Conference on Data …, 2015 – ieeexplore.ieee.org … Module-3 No. of Event B No. of Event B 217 Y N 136 Y N A Y 199 4 A Y 127 2 N 6 8 N 3 4 C. Sentic WSD using Deep RNN Lastly, we evaluate the proposed lexical resource for training a Deep Recurrent Neural Networks (DRNN) previously described in [24]. … Cited by 1 Related articles All 2 versions

Learning semantic hierarchies: A continuous vector space approach R Fu, J Guo, B Qin, W Che, H Wang, T Liu – IEEE Transactions on Audio, …, 2015 – dl.acm.org … guistic regularities. [8] first demonstrated that many syntactic/ semantic relations of words can be recovered by means of vector arithmetic in the embedding space learned by recurrent neural network language models. For example … Cited by 2 Related articles All 8 versions

Fast and Large-scale Unsupervised Relation Extraction S Takase, N Okazaki, K Inui – 2015 – bcmi.sjtu.edu.cn Page 1. Fast and Large-scale Unsupervised Relation Extraction Sho Takase † Naoaki Okazaki †‡ Kentaro Inui † Graduate School of Information Sciences, Tohoku University † Japan Science and Technology Agency (JST) ‡ {takase, okazaki, inui}@ecei.tohoku.ac.jp Abstract … Related articles All 8 versions

Framing QA as Building and Ranking Intersentence Answer Justifications P Jansen, R Sharp, M Surdeanu, P Clark – pdfs.semanticscholar.org Page 1. Framing QA as Building and Ranking Intersentence Answer Justifications Peter Jansen? University of Arizona Rebecca Sharp?? University of Arizona Mihai Surdeanu† University of Arizona Peter Clark‡ Allen Institute for Artificial Intelligence … Related articles All 2 versions

Learning for Spoken Dialog Systems with Discriminative Graphical Models Y Ma – 2015 – etd.ohiolink.edu … system that maintains the entire dialog history from the very beginning of the interaction. Instead, they are more like question answering systems that will ignore all previous user input except the most recent user’s utterance during the course of a dialog. … Related articles All 3 versions