Olga Kovaleva

Olga Kovaleva

Hi! I am a PhD student at Text Machine Lab, University of Massachusetts Lowell. My background is in Applied Mathematics and I currently focus on representation learning problems in ML. My work also includes building NLP models for question answering and analyzing electronic health records.

A. Rogers O.Kovaleva A.Rumshisky A Primer in BERTology: What We Know about How BERT Works. Accepted to TACL 2020.

@article{rogers2020primer, title={A primer in bertology: What we know about how bert works}, author={Rogers, Anna and Kovaleva, Olga and Rumshisky, Anna}, journal={arXiv preprint arXiv:2002.12327}, year={2020} }

A. Rogers O. Kovaleva M. Downey A. Rumshisky Getting Closer to AI-complete Question Answering: A Set of Prerequisite Real Tasks. Proceedings of AAAI 2020.

@inproceedings{rogers2020getting, title={Getting Closer to AI Complete Question Answering: A Set of Prerequisite Real Tasks}, author={Rogers, A and Kovaleva, O and Downey, M and Rumshisky, A}, booktitle={Proceedings of the AAAI Conference on Artificial Intelligence}, year={2020} }

R. Perlis W. Boag O. Kovaleva T. McCoy A. Rumshisky P. Szolovits Hard for Humans, Hard for Machines: Predicting Readmission after Psychiatric Hospitalization Using Narrative Notes. Translational Psychiatry.
O. Kovaleva C. Shivade S. Kashyap K. Kanjaria A. Coy D. Ballah J. Wu Y. Guo A. Karargyris D. Beymer A. Rumshisky V. Mukherjee Towards Visual Dialog for Radiology. BioNLP, ACL 2020.
O. Kovaleva A. Romanov A. Rogers A. Rumshisky Revealing the Dark Secrets of BERT. Proceedings of EMNLP 2019.

@inproceedings{kovaleva2019revealing, title={Revealing the Dark Secrets of BERT}, author={Kovaleva, Olga and Romanov, Alexey and Rogers, Anna and Rumshisky, Anna}, booktitle={Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)}, pages={4356–4365}, year={2019} }

O. Kovaleva A. Romanov A. Rumshisky Similarity-Based Reconstruction Loss for Meaning Representation. (short paper). Proceedings of EMNLP 2018.

@inproceedings{kovaleva2018similarity, title={Similarity-Based Reconstruction Loss for Meaning Representation}, author={Kovaleva, Olga and Rumshisky, Anna and Romanov, Alexey}, booktitle={Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing}, pages={4875–4880}, year={2018} }

  • QuAIL - Question Answering for Artificial Intelligence

    A new kind of question-answering dataset that combines commonsense, text-based, and unanswerable questions, balanced for different genres and reasoning types. Reasoning type annotation for 9 types of reasoning: temporal, causality, factoid, coreference, character properties, their belief states, subsequent entity states, event durations, and unanswerable. Genres: CC license fiction, Voice of America news, blogs, user stories from Quora 800 texts, 18 questions for each (~14K questions).

  • The Dark Secrets of BERT

    In this project, we study self-attention patterns and the relative impact of fine-tuning and pre-training in BERT, the popular Transformer-based model that uses pre-training to learn contextualized representations.