Hello!
Hi, I'm Suman Adhya, a Ph.D. candidate at the School of Mathematical & Computational Sciences (SMCS) at IACS, Kolkata, under the supervision of Dr. Debarshi Kumar Sanyal. I work on probabilistic and neural topic modeling in Natural Language Processing (NLP), with a focus on enhancing the quality of topic representations, improving interpretability, and increasing efficiency. My research has been published in venues such as ACL 2025, NAACL 2024, EACL 2023, ECIR 2023, and IEEE Transactions on Artificial Intelligence. I will submit my doctoral thesis for review in August 2025 and am currently actively exploring research-oriented opportunities in academia or industry.

For more details, check my CV or drop a mail.

 News and Updates:

  •   July 2024: Paper accepted at IEEE-TAI.
  •   May 2024: Received Microsoft Research Travel Grant to present our paper at NAACL'24.
  •   April 2024: Received ACM India-IARCS Travel Grant to present our paper at NAACL'24.
  •   March 2024: Paper on topic modeling using graph isomorphism network has been accepted at NAACL'24.
  •   January 2024: Selected to attend Research Week with Google-2024.
  •   November 2023: Selected to attend PIC-2024.
  •   July 2023: Attended ACL'23 (virtually).
  •   May 2023: Served as Volunteer coordinator at EACL'23 (in-person).
  •   March 2023: Received Microsoft Research Travel Grant to present our paper at EACL'23.
  •   January 2023: Paper on the analysis of the dropout effect in topic modeling has been accepted at EACL'23.
  •   December 2022: Paper on the Wasserstein knowledge distillation framework for topic modeling has been accepted at ECIR'23.
  •   November 2022: Paper on negative sampling for topic modeling got accepted at ICON'22.
  •   May 2022: Our paper on analysis of the question hour session of the Indian Parliament using dynamic topic modeling got accepted at PoliticalNLP'22.
  •   October 2020: Started my Ph.D. at IACS Kolkata

 Publications:

DTECT: Dynamic Topic Explorer & Context Tracker
Suman Adhya, Debarshi Kumar Sanyal
arXiv | GitHub | Live Demo | Demo Video

S2WTM: Spherical Sliced-Wasserstein Autoencoder for Topic Modeling
Suman Adhya, Debarshi Kumar Sanyal
ACL'25 | Association for Computational Linguistics
proceedings| arXiv| presentation| code

Evaluating Negative Sampling Approaches for Neural Topic Models
Suman Adhya, Avishek Lahiri, Debarshi Kumar Sanyal, Partha Pratim Das
IEEE-TAI | IEEE Transactions on Artificial Intelligence
IEEE Vol. 5, Issue. 11| arXiv| code

GINopic: Topic Modeling with Graph Isomorphism Network
Suman Adhya, Debarshi Kumar Sanyal
NAACL'24 | North American Chapter of the Association for Computational Linguistics
proceedings| arXiv| presentation| code

Do Neural Topic Models Really Need Dropout? Analysis of the Effect of Dropout in Topic Modeling
Suman Adhya, Avishek Lahiri, Debarshi Kumar Sanyal
EACL'23 | European Chapter of the Association for Computational Linguistics
proceedings| arXiv| presentation| code

Improving Neural Topic Models with Wasserstein Knowledge Distillation
Suman Adhya, Debarshi Kumar Sanyal
ECIR'23 | European Conference on Information Retrieval
proceedings| arXiv| presentation| code

Improving Contextualized Topic Models with Negative Sampling
Suman Adhya, Avishek Lahiri, Debarshi Kumar Sanyal, Partha Pratim Das
ICON'22 | International Conference on Natural Language Processing
proceedings| arXiv| presentation| code

What Does the Indian Parliament Discuss? An Exploratory Analysis of the Question Hour in the Lok Sabha
Suman Adhya, Debarshi Kumar Sanyal
PoliticalNLP @ LREC'22 | Natural Language Processing for Political Sciences
proceedings| arXiv| presentation code


 Academic Responsibilities:



 Teaching Assistant:

  • Spring 2021: Laboratory - Numerical methods.
  • Spring 2024: Introduction to Machine Learning.

  Grab this theme: Sebastin Santy