Dheeru Dua

Ph.D. Student in Computer Science

University of California, Irvine

Office:	4059 Donald Bren Hall, Irvine, California, 92617
Email:	ddua (at) uci (dot) edu

[Curriculum Vitae] [Google Scholar]

I am a PhD student in the Computer Science Department at University of California, Irvine. I am advised by Dr. Sameer Singh and Dr. Matt Gardner. My broad research interests lie in making question answering systems robust. My recent works focus on improving out-of-domain and compositional generalization in QA systems. My research is supported by HPI fellowship. In the past I have interned at Amazon (AWS) in summer 2020 and Facebook AI reasearch (FAIR) in summer 2021 and Google Research in summer 2022.

Publications

D. Dua, E. Strubell, S. Singh, P. Verga
To Adapt or to Annotate: Challenges and Interventions for Domain Adaptation in Open-Domain Question Answering [pdf]
Under review
D. Dua, S. Gupta, S. Singh, M. Gardner
Sucessive Prompting for Question Decomposition [pdf]
Empirical Methods in Natural Language Processing (EMNLP). 2022
D. Dua, S. Bhosale, V. Goswami, J. Cross, M. Lewis, A. Fan
Tricks for Training Sparse Translation Models [pdf]
North American Conference of Computational Linguistics (NAACL). 2022
D. Dua, P. Dasigi, S. Singh, M. Gardner
Learning with Instance Bundles for Reading Comprehension. [pdf]
Empirical Methods in Natural Language Processing (EMNLP). 2021
D. Dua, C. N. dos Santos, P. Ng, B. Athiwaratkun, B. Xiang, M. Gardner, S. Singh
Generative Context Pair Selection for Multi-hop Question Answering. [pdf]
Empirical Methods in Natural Language Processing (EMNLP). 2021
D. Dua, S. Singh, M. Gardner
Benefits of Intermediate Annotations in Reading Comprehension. [pdf]
Association of Computational Linguistics (ACL). 2020
A. Gottumukkala, D. Dua, S. Singh, M. Gardner
Dynamic Sampling Strategies for Multi-Task Reading Comprehension. [pdf]
Association of Computational Linguistics (ACL). 2020
D. Dua, Y. Wang, P. Dasigi, G. Stanovsky, S. Singh, M. Gardner
DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphs. [pdf]
North American Conference of Computational Linguistics (NAACL). 2019
Z. Zhao, D. Dua, S. Singh.
Generating Natural Adversarial Examples. [pdf]
International Conference on Learning Representations (ICLR). 2018
J.S. Kang, R. Logan, Z. Chu, Y. Chen, D. Dua, K. Gimpel, S. Singh, N. Balasubramanian
PoMo: Generating Entity-Specific Post-Modifiers in Context. [pdf]
North American Conference of Computational Linguistics (NAACL). 2019
D. Dua, A. Gottumukkala, A. Talmor, M. Gardner, S. Singh.
ORB: An Open Reading Benchmark for Comprehensive Multi-Dataset Evaluation of Reading Comprehension. [pdf]
Workshop on Machine Reading and Question Answering (MRQA). 2019

Performed relation classification using distantly-supervised MultiR algorithm with features extracted by doing random walks on the Freebase graph. Achieved better results over state of the art systems at sentential and aggregate predictions.
Built an event extraction system using passive- agressive conditional random fields for TAC KBP 2015.
Used reinforcement learning approaches, DQN with MCTS guided policy for abstractive document summarization
Worked on the NTCIR Question Answering task as a part of my Master's thesis and experimented with various components, some of which are listed below.

Wikipedia link graph, which was used to continually search and retrieve relevant documents to answer the questions.
Markov Logic Networks, which was used to build a sequence of events and use link-prediction for answering timeline based questions and as an event and entity graph for lookup.
A UIMA based workflow which wrapped all the components.

Designed and developed a torch-based framework for fast development and deployment of neural network models into production.
Used Bing's knowledge graph to surface information cards about various frequently queried entities
Developed an integration test framework which performed various image comparison techniques (using pHash, RMSE, Template matching algorithms) to test information cards deployed in production

PhD(CS)	MS(LTI)	B.Tech(CS)	High School

About Me

Publications

Masters Projects

Company Projects

Education