Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Mitesh M. Khapra

Mitesh M. Khapra

The heads hypothesis: A unifying statistical approach towards understanding multi-headed attention in BERT


Jan 22, 2021
Madhura Pande, Aakriti Budhraja, Preksha Nema, Pratyush Kumar, Mitesh M. Khapra

Figure 1 for The heads hypothesis: A unifying statistical approach towards understanding multi-headed attention in BERT
Figure 2 for The heads hypothesis: A unifying statistical approach towards understanding multi-headed attention in BERT
Figure 3 for The heads hypothesis: A unifying statistical approach towards understanding multi-headed attention in BERT
Figure 4 for The heads hypothesis: A unifying statistical approach towards understanding multi-headed attention in BERT
* accepted at AAAI 2021 (Main conference) 
  Access Paper or Ask Questions

Unsupervised Deep Video Denoising


Dec 01, 2020
Dev Yashpal Sheth, Sreyas Mohan, Joshua L. Vincent, Ramon Manzorro, Peter A. Crozier, Mitesh M. Khapra, Eero P. Simoncelli, Carlos Fernandez-Granda

Figure 1 for Unsupervised Deep Video Denoising
Figure 2 for Unsupervised Deep Video Denoising
Figure 3 for Unsupervised Deep Video Denoising
Figure 4 for Unsupervised Deep Video Denoising
* Dev and Sreyas contributed equally 
  Access Paper or Ask Questions

Towards Interpreting BERT for Reading Comprehension Based QA


Oct 18, 2020
Sahana Ramnath, Preksha Nema, Deep Sahni, Mitesh M. Khapra

Figure 1 for Towards Interpreting BERT for Reading Comprehension Based QA
Figure 2 for Towards Interpreting BERT for Reading Comprehension Based QA
Figure 3 for Towards Interpreting BERT for Reading Comprehension Based QA
Figure 4 for Towards Interpreting BERT for Reading Comprehension Based QA
* 7 pages including references and appendix. Accepted at EMNLP 2020 
  Access Paper or Ask Questions

A Survey of Evaluation Metrics Used for NLG Systems


Oct 05, 2020
Ananya B. Sai, Akash Kumar Mohankumar, Mitesh M. Khapra

Figure 1 for A Survey of Evaluation Metrics Used for NLG Systems
Figure 2 for A Survey of Evaluation Metrics Used for NLG Systems
Figure 3 for A Survey of Evaluation Metrics Used for NLG Systems
Figure 4 for A Survey of Evaluation Metrics Used for NLG Systems
* A condensed version of this paper is submitted to ACM CSUR 
  Access Paper or Ask Questions

Evaluating a Generative Adversarial Framework for Information Retrieval


Oct 01, 2020
Ameet Deshpande, Mitesh M. Khapra

Figure 1 for Evaluating a Generative Adversarial Framework for Information Retrieval
Figure 2 for Evaluating a Generative Adversarial Framework for Information Retrieval
Figure 3 for Evaluating a Generative Adversarial Framework for Information Retrieval
Figure 4 for Evaluating a Generative Adversarial Framework for Information Retrieval

  Access Paper or Ask Questions

Improving Dialog Evaluation with a Multi-reference Adversarial Dataset and Large Scale Pretraining


Sep 23, 2020
Ananya B. Sai, Akash Kumar Mohankumar, Siddhartha Arora, Mitesh M. Khapra

Figure 1 for Improving Dialog Evaluation with a Multi-reference Adversarial Dataset and Large Scale Pretraining
Figure 2 for Improving Dialog Evaluation with a Multi-reference Adversarial Dataset and Large Scale Pretraining
Figure 3 for Improving Dialog Evaluation with a Multi-reference Adversarial Dataset and Large Scale Pretraining
Figure 4 for Improving Dialog Evaluation with a Multi-reference Adversarial Dataset and Large Scale Pretraining
* Accepted for publication in TACL 
  Access Paper or Ask Questions

On the Importance of Local Information in Transformer Based Models


Aug 13, 2020
Madhura Pande, Aakriti Budhraja, Preksha Nema, Pratyush Kumar, Mitesh M. Khapra

Figure 1 for On the Importance of Local Information in Transformer Based Models
Figure 2 for On the Importance of Local Information in Transformer Based Models
Figure 3 for On the Importance of Local Information in Transformer Based Models
Figure 4 for On the Importance of Local Information in Transformer Based Models
* 10 pages, 4 figures 
  Access Paper or Ask Questions

A Systematic Evaluation of Object Detection Networks for Scientific Plots


Jul 05, 2020
Pritha Ganguly, Nitesh Methani, Mitesh M. Khapra, Pratyush Kumar

Figure 1 for A Systematic Evaluation of Object Detection Networks for Scientific Plots
Figure 2 for A Systematic Evaluation of Object Detection Networks for Scientific Plots
Figure 3 for A Systematic Evaluation of Object Detection Networks for Scientific Plots
Figure 4 for A Systematic Evaluation of Object Detection Networks for Scientific Plots

  Access Paper or Ask Questions

On Incorporating Structural Information to improve Dialogue Response Generation


May 28, 2020
Nikita Moghe, Priyesh Vijayan, Balaraman Ravindran, Mitesh M. Khapra

Figure 1 for On Incorporating Structural Information to improve Dialogue Response Generation
Figure 2 for On Incorporating Structural Information to improve Dialogue Response Generation
Figure 3 for On Incorporating Structural Information to improve Dialogue Response Generation
Figure 4 for On Incorporating Structural Information to improve Dialogue Response Generation

  Access Paper or Ask Questions

AI4Bharat-IndicNLP Corpus: Monolingual Corpora and Word Embeddings for Indic Languages


Apr 30, 2020
Anoop Kunchukuttan, Divyanshu Kakwani, Satish Golla, Gokul N. C., Avik Bhattacharyya, Mitesh M. Khapra, Pratyush Kumar

Figure 1 for AI4Bharat-IndicNLP Corpus: Monolingual Corpora and Word Embeddings for Indic Languages
Figure 2 for AI4Bharat-IndicNLP Corpus: Monolingual Corpora and Word Embeddings for Indic Languages
Figure 3 for AI4Bharat-IndicNLP Corpus: Monolingual Corpora and Word Embeddings for Indic Languages
Figure 4 for AI4Bharat-IndicNLP Corpus: Monolingual Corpora and Word Embeddings for Indic Languages
* 7 pages, 8 tables, https://github.com/ai4bharat-indicnlp/indicnlp_corpus 
  Access Paper or Ask Questions

Towards Transparent and Explainable Attention Models


Apr 29, 2020
Akash Kumar Mohankumar, Preksha Nema, Sharan Narasimhan, Mitesh M. Khapra, Balaji Vasan Srinivasan, Balaraman Ravindran

Figure 1 for Towards Transparent and Explainable Attention Models
Figure 2 for Towards Transparent and Explainable Attention Models
Figure 3 for Towards Transparent and Explainable Attention Models
Figure 4 for Towards Transparent and Explainable Attention Models
* Accepted at ACL 2020 
  Access Paper or Ask Questions

Scene Graph based Image Retrieval -- A case study on the CLEVR Dataset


Nov 03, 2019
Sahana Ramnath, Amrita Saha, Soumen Chakrabarti, Mitesh M. Khapra

* 3 pages including references, Accepted at the ICCV 2019 Workshop - 'Linguistics Meets Image and Video Retrieval' (received Best Paper Award) 
  Access Paper or Ask Questions

Data Interpretation over Plots


Sep 03, 2019
Nitesh Methani, Pritha Ganguly, Mitesh M. Khapra, Pratyush Kumar

Figure 1 for Data Interpretation over Plots
Figure 2 for Data Interpretation over Plots
Figure 3 for Data Interpretation over Plots
Figure 4 for Data Interpretation over Plots

  Access Paper or Ask Questions

Let's Ask Again: Refine Network for Automatic Question Generation


Aug 31, 2019
Preksha Nema, Akash Kumar Mohankumar, Mitesh M. Khapra, Balaji Vasan Srinivasan, Balaraman Ravindran

Figure 1 for Let's Ask Again: Refine Network for Automatic Question Generation
Figure 2 for Let's Ask Again: Refine Network for Automatic Question Generation
Figure 3 for Let's Ask Again: Refine Network for Automatic Question Generation
Figure 4 for Let's Ask Again: Refine Network for Automatic Question Generation
* accepted in EMNLP 2019 in Main Conference, (10 pages) 
  Access Paper or Ask Questions

Frustratingly Poor Performance of Reading Comprehension Models on Non-adversarial Examples


Apr 04, 2019
Soham Parikh, Ananya B. Sai, Preksha Nema, Mitesh M. Khapra

Figure 1 for Frustratingly Poor Performance of Reading Comprehension Models on Non-adversarial Examples
Figure 2 for Frustratingly Poor Performance of Reading Comprehension Models on Non-adversarial Examples
Figure 3 for Frustratingly Poor Performance of Reading Comprehension Models on Non-adversarial Examples
Figure 4 for Frustratingly Poor Performance of Reading Comprehension Models on Non-adversarial Examples
* 8 pages 
  Access Paper or Ask Questions

ElimiNet: A Model for Eliminating Options for Reading Comprehension with Multiple Choice Questions


Apr 04, 2019
Soham Parikh, Ananya B. Sai, Preksha Nema, Mitesh M. Khapra

Figure 1 for ElimiNet: A Model for Eliminating Options for Reading Comprehension with Multiple Choice Questions
Figure 2 for ElimiNet: A Model for Eliminating Options for Reading Comprehension with Multiple Choice Questions
Figure 3 for ElimiNet: A Model for Eliminating Options for Reading Comprehension with Multiple Choice Questions
Figure 4 for ElimiNet: A Model for Eliminating Options for Reading Comprehension with Multiple Choice Questions
* Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (2018) Main track. Pages 4272-4278 
 * IJCAI-18 
  Access Paper or Ask Questions

Efficient Video Classification Using Fewer Frames


Feb 27, 2019
Shweta Bhardwaj, Mukundhan Srinivasan, Mitesh M. Khapra

Figure 1 for Efficient Video Classification Using Fewer Frames
Figure 2 for Efficient Video Classification Using Fewer Frames
Figure 3 for Efficient Video Classification Using Fewer Frames
Figure 4 for Efficient Video Classification Using Fewer Frames
* To Appear in Proceedings of IEEE International Conference on Computer Vision and Pattern Recognition (CVPR'2019) 
  Access Paper or Ask Questions

Re-evaluating ADEM: A Deeper Look at Scoring Dialogue Responses


Feb 23, 2019
Ananya B. Sai, Mithun Das Gupta, Mitesh M. Khapra, Mukundhan Srinivasan

Figure 1 for Re-evaluating ADEM: A Deeper Look at Scoring Dialogue Responses
Figure 2 for Re-evaluating ADEM: A Deeper Look at Scoring Dialogue Responses
Figure 3 for Re-evaluating ADEM: A Deeper Look at Scoring Dialogue Responses
Figure 4 for Re-evaluating ADEM: A Deeper Look at Scoring Dialogue Responses
* Accepted as a long paper in the proceedings of AAAI-2019 
  Access Paper or Ask Questions

Studying the Plasticity in Deep Convolutional Neural Networks using Random Pruning


Dec 26, 2018
Deepak Mittal, Shweta Bhardwaj, Mitesh M. Khapra, Balaraman Ravindran

Figure 1 for Studying the Plasticity in Deep Convolutional Neural Networks using Random Pruning
Figure 2 for Studying the Plasticity in Deep Convolutional Neural Networks using Random Pruning
Figure 3 for Studying the Plasticity in Deep Convolutional Neural Networks using Random Pruning
Figure 4 for Studying the Plasticity in Deep Convolutional Neural Networks using Random Pruning
* To appear in the Journal of Machine Vision and Applications, Springer. This work is an extended version of our previous work arXiv:1801.10447, "Recovering from Random Pruning: On the Plasticity of Deep Convolutional Neural Networks", accepted at WACV 2018 
  Access Paper or Ask Questions

On Controllable Sparse Alternatives to Softmax


Oct 30, 2018
Anirban Laha, Saneem A. Chemmengath, Priyanka Agrawal, Mitesh M. Khapra, Karthik Sankaranarayanan, Harish G. Ramaswamy

Figure 1 for On Controllable Sparse Alternatives to Softmax
Figure 2 for On Controllable Sparse Alternatives to Softmax
Figure 3 for On Controllable Sparse Alternatives to Softmax
Figure 4 for On Controllable Sparse Alternatives to Softmax
* To appear in NIPS 2018, Total 16 pages including appendix 
  Access Paper or Ask Questions

DuoRC: Towards Complex Language Understanding with Paraphrased Reading Comprehension


Oct 10, 2018
Amrita Saha, Rahul Aralikatte, Mitesh M. Khapra, Karthik Sankaranarayanan

Figure 1 for DuoRC: Towards Complex Language Understanding with Paraphrased Reading Comprehension
Figure 2 for DuoRC: Towards Complex Language Understanding with Paraphrased Reading Comprehension
Figure 3 for DuoRC: Towards Complex Language Understanding with Paraphrased Reading Comprehension
Figure 4 for DuoRC: Towards Complex Language Understanding with Paraphrased Reading Comprehension
* Accepted in ACL 2018 
  Access Paper or Ask Questions

Complex Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with a Knowledge Graph


Oct 04, 2018
Amrita Saha, Vardaan Pahuja, Mitesh M. Khapra, Karthik Sankaranarayanan, Sarath Chandar

Figure 1 for Complex Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with a Knowledge Graph
Figure 2 for Complex Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with a Knowledge Graph
Figure 3 for Complex Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with a Knowledge Graph
Figure 4 for Complex Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with a Knowledge Graph
* Accepted in AAAI'18 
  Access Paper or Ask Questions

Towards Exploiting Background Knowledge for Building Conversation Systems


Sep 21, 2018
Nikita Moghe, Siddhartha Arora, Suman Banerjee, Mitesh M. Khapra

Figure 1 for Towards Exploiting Background Knowledge for Building Conversation Systems
Figure 2 for Towards Exploiting Background Knowledge for Building Conversation Systems
Figure 3 for Towards Exploiting Background Knowledge for Building Conversation Systems
Figure 4 for Towards Exploiting Background Knowledge for Building Conversation Systems
* Camera Ready EMNLP 2018 
  Access Paper or Ask Questions

HOPF: Higher Order Propagation Framework for Deep Collective Classification


Sep 21, 2018
Priyesh Vijayan, Yash Chandak, Mitesh M. Khapra, Srinivasan Parthasarathy, Balaraman Ravindran

Figure 1 for HOPF: Higher Order Propagation Framework for Deep Collective Classification
Figure 2 for HOPF: Higher Order Propagation Framework for Deep Collective Classification
Figure 3 for HOPF: Higher Order Propagation Framework for Deep Collective Classification
Figure 4 for HOPF: Higher Order Propagation Framework for Deep Collective Classification

  Access Paper or Ask Questions

Fusion Graph Convolutional Networks


Sep 21, 2018
Priyesh Vijayan, Yash Chandak, Mitesh M. Khapra, Srinivasan Parthasarathy, Balaraman Ravindran

Figure 1 for Fusion Graph Convolutional Networks
Figure 2 for Fusion Graph Convolutional Networks

  Access Paper or Ask Questions

Towards a Better Metric for Evaluating Question Generation Systems


Aug 31, 2018
Preksha Nema, Mitesh M. Khapra

Figure 1 for Towards a Better Metric for Evaluating Question Generation Systems
Figure 2 for Towards a Better Metric for Evaluating Question Generation Systems
Figure 3 for Towards a Better Metric for Evaluating Question Generation Systems
Figure 4 for Towards a Better Metric for Evaluating Question Generation Systems
* EMNLP, 2018 
 * 10 pages 
  Access Paper or Ask Questions

A Dataset for Building Code-Mixed Goal Oriented Conversation Systems


Jun 15, 2018
Suman Banerjee, Nikita Moghe, Siddhartha Arora, Mitesh M. Khapra

Figure 1 for A Dataset for Building Code-Mixed Goal Oriented Conversation Systems
Figure 2 for A Dataset for Building Code-Mixed Goal Oriented Conversation Systems
Figure 3 for A Dataset for Building Code-Mixed Goal Oriented Conversation Systems
Figure 4 for A Dataset for Building Code-Mixed Goal Oriented Conversation Systems
* 15 pages, 2 figures, 10 tables, Accepted in COLING - 2018 
  Access Paper or Ask Questions

A Question-Answering framework for plots using Deep learning


Jun 12, 2018
Revanth Reddy, Rahul Ramesh, Ameet Deshpande, Mitesh M. Khapra

Figure 1 for A Question-Answering framework for plots using Deep learning
Figure 2 for A Question-Answering framework for plots using Deep learning
Figure 3 for A Question-Answering framework for plots using Deep learning
Figure 4 for A Question-Answering framework for plots using Deep learning

  Access Paper or Ask Questions

I Have Seen Enough: A Teacher Student Network for Video Classification Using Fewer Frames


May 12, 2018
Shweta Bhardwaj, Mitesh M. Khapra

Figure 1 for I Have Seen Enough: A Teacher Student Network for Video Classification Using Fewer Frames
Figure 2 for I Have Seen Enough: A Teacher Student Network for Video Classification Using Fewer Frames
Figure 3 for I Have Seen Enough: A Teacher Student Network for Video Classification Using Fewer Frames
* CVPR Workshop on Brave New Ideas for Video Understanding (BIVU) 
  Access Paper or Ask Questions