Tip: Click on the sub-heading starting with ▶️ for more details.
INRIA Rennes | INRIA Nancy-Grand Est
Master’s Thesis | Rennes, France | March 2021 - August 2021
Worked as a Research Intern at LACODAM INRIA, Rennes, in collaboration with ORPAILLEUR LORIA, Nancy, till August 2021, to complete my master’s thesis under the guidance of Dr Luis Galárraga and Prof. Miguel Couceiro. During this period, I worked on the project titled “Reducing Bias in Language Models using Deep Reinforcement Learning”. This project aims to assess the fairness of a model and reduce bias. I mainly worked on question answering models to formulate the problem as a multi-armed bandit environment. We proposed a novel way to fine-tune the model with the help of rewards based on bias metrics. During this tenure, I worked with the vision to make complex language models fairer and explainable.
tags:
Reinforcement Learning Natural Language Processing Question Answering Systems
Indian Institute of Science Education & Research Bhopal
Research Assistant | Bhopal, India | August 2018 - August 2020
Through my term as a research assistant at IISER B, I formulated a genre classification problem under the guidance of Prof. Kushal Shah and Dr Rajakrishnan P. Rajkumar, which was published at StoryNLP @ ACL-2019. In this work, we have explored the linguistic and cognitive aspects of a genre classification to distinguish between fiction and non-fiction articles. While analysing the various important linguistic features, I also got a thorough training in machine learning and relevant tools. To demonstrate our work, I developed a web application - “Fictometer”, which measures the imaginative content of the text. I have developed a mobile application using Flutter, and Django called “NewsChase” in continuation of this idea. The application lists news articles from various sources and gives a measure/score of manipulative content in the selected text/article. The sole purpose of this app to educate users about the manipulative language used in news articles to deceive users. This work got featured in local news as well.
tags:
Natural Language Processing Machine Learning Genre ClassificationIndian Institute of Science Education & Research Bhopal
Master’s Thesis | Bhopal, India | June2017 - May 2018
I defended my Mathematics MS thesis in the area of Game Theory under the guidance of Dr A. K. Srustidhar Chand, in which I studied the proof of Nash Equilibrium using graph theory and simplicial complexes, keeping my mathematical insights sated. In the second phase of the project, I investigated dynamic games & strategic bargaining. I studied Rubinstein Bargaining Model with the infinite horizon setting in detail. We validated a perfect equilibrium in the bargaining model, including infinite stages.
tags:
Game Theory Strategic Bargaining Nash EquilibriumIndian Institute of Science Bangalore
Summer Internship | Bangalore, India | May 2016 - Jul 2016
At the end of the sixth semester, I worked as a summer intern at the Indian Institute of Science (IISc) Bangalore, India, under the guidance of Prof. Manjunath Krishnapur in the field of “Markov chains and Probability Theory”. I explored stochastic processes like Markov Chains and various exciting examples, including Pólya’s Urn Model.
tags:
Combinatorics Discrete Probability📜 Publications and Talks
- Qureshi, M. (2021). Mitigating Unintended Bias in Masked Language Models. Inria Rennes ; Loria.
- Qureshi, M. (2019). A Simple Approach to Classify Fictional and Non-Fictional Genres. In Proceedings of the Second Workshop on Storytelling (pp. 81–89). Association for Computational Linguistics (ACL).
🗣 Talk: “Mitigating Unindented Bias in Masked Language model” - MALOTAC Seminar, INRIA Nancy
