Kaushal Kumar Maurya

(he/him)| कौशल | கௌஷல் | スキル | కౌశల్ | Навык | ਹੁਨਰ |

kaushal_pf.jpg

I am a Ph.D. candidate in the Department of Computer Science and Engineering at Indian Institute of Technology Hyderabad and advised by Dr. Maunendra Sankar Desarkar. I am an active collaborator at Microsoft India, working with Manish Gupta and Anoop Kunchukuttan. I spent the last three wonderful summers interning at Microsoft (Translation and Auto-suggest teams) and Nvidia-AI.

Research Interest

My research interest is applied Machine Learning for Natural Language Processing, with a particular focus on Multilingual Natural Language Processing. Simultaneously, I am also working towards improving knowledge augmentation and trustworthy generation with large language models.

As a multilingual myself, I perceive a pressing need for the development of cross-lingual/multilingual models that can facilitate a range of end-user applications in low-resource languages. In this regard, my research trajectory is primarily geared toward the development of novel models that can enable technology for low-resource languages that have limited or no data. To achieve this overarching objective, I am actively engaged in enhancing multilingual transfer learning models for language generation with limited supervision. My approach to addressing these issues is largely rooted in the linguistic perspective. In particular, my Ph.D. research focuses on anchoring three narrative properties, namely (1) language structure, (2) context, and (3) transferability from/to diverse typological languages. Furthermore, as the use of large language models has increased in recent years, it is crucial to ensure that these models are deployed safely and ethically for the audience. I believe that the aforementioned endeavors constitute a pivotal stride in the direction of accomplishing my research objective, which is to democratize and ensure the safe deployment of NLP technologies.

Prior to commencing my Ph.D. studies, I served as a Data Scientist at NTWIST, an AI startup. Before this, I completed a Master’s degree (M.Tech) in Artificial Intelligence from the University of Hyderabad under the supervision of Prof. K. Narayana Murthy. I obtained a Bachelor’s degree (B.Tech) in Computer Science and Engineering from Uttar Pradesh Technical University. During my Ph.D., I am humbled to receive a Suzuki Foundation Fellowship (two years consecutively) to visit Shizuoka University, Japan for a short research stay.

News

Oct 2023 I talked about AI Research and My PhD research Experience with Piya on my first PodCast. Thank you, Priya, for the invitation and for being a wonderful host.
Oct 2023 Two papers have been accepted at EMNLP 2023 (Findings and The BigPicture workshop). The Finding paper is authored with the awesome collaborator, Maharaj.
Jul 2023 Our paper “Trie-NLG: Trie Context Augmentation to Improve Personalized Query Auto-Completion for Short and Unseen Prefixes” is accepted in the ECML-PKDD 2023 (Journal Track: DAMI). This work is a collaborative effort with Microsoft India, supported by the Microsoft Academic Partnership Grant.
May 2023 Our paper “DIVHSK: Diverse Headline Generation using Self-Attention based Keyword Selection” is accepted in the Findings of ACL 2023.
Sep 2022 Received a grant of Rs. 100k INR to attend conferences by IIT Hyderabad in Exceptional Research Scholar category.

Selected Publications

  1. UnsupervisedMachine TranslationExtremely LRLs
    Unsupervised Noise Injection to Enable Zero-Shot Machine Translation for Extremely Low-resource Languages
    Maharaj Brahma, Kaushal Kumar Maurya, and Maunendra Sankar Desarkar
    In Findings of EMNLP, 2023
  2. Auto-CompletionTrie QACNLG Augmentation
    Trie-NLG: Trie Context Augmentation to Improve Personalized Query Auto-Completion for Short and Unseen Prefixes
    Kaushal Kumar Maurya, Maunendra Sankar Desarkar, Manish Gupta, and 1 more author
    In ECML-PKDD (Journal Track: DAMI), 2023
  3. NLGDiverse HeadlinesSelf-attention
    DIVHSK: Diverse Headline Generation using Self-Attention based Keyword Selection
    Venkatesh E, Kaushal Kumar Maurya, Deepak Kumar, and 1 more author
    In Findings of the Association for Computational Linguistics: ACL 2023, Jul 2023
  4. Cross-LingualMeta-LearningTypology
    Meta-X_NLG: A Meta-Learning Approach Based on Language Clustering for Zero-Shot Cross-Lingual Transfer and Generation
    Kaushal Kumar Maurya, and Maunendra Desarkar
    In Findings of the Association for Computational Linguistics: ACL 2022, May 2022
  5. Cross-LingualUnsupervisedTransfer-Learning
    ZmBART: An Unsupervised Cross-lingual Transfer Framework for Language Generation
    Kaushal Kumar Maurya, Maunendra Sankar Desarkar, Yoshinobu Kano, and 1 more author
    In Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021, Aug 2021
  6. Q&A, MCQMulti-DecoderLSTM
    Learning to Distract: A Hierarchical Multi-Decoder Network for Automated Generation of Long Distractors for Multiple-Choice Questions for Reading Comprehension
    Kaushal Kumar Maurya, and Maunendra Sankar Desarkar
    In Proceedings of the 29th ACM International Conference on Information & Knowledge Management, Aug 2020