hello, i'm currently a senior undergraduate student at iiit hyderabad pursuing a bachelor's degree with honors in computer science. i am a researcher at the precog research group advised by dr. ponnurangam kumaraguru. my interests lie broadly in understanding neural networks operating on graph data. more specifically, i am interested in the following topics:
- self-supervised learning on graphs
- interpretability in graph neural networks
- trustworthy machine learning
- neural algorithmic properties of graph-based networks
- foundational models for graph learning
besides the technical stuff, i run a (slightly inactive) youtube channel, i love watching football (liverpool fc; ynwa), playing video games, watching anime, playing chess, listening to music (mostly indie rock/pop), and occasionally reading sci-fi/dystopian/philosophical fiction specifically from 1960-1970s.
feel free to mail me at akshit.sinha[at]students.iiit.ac.in if you want to have a chat about ideas!
updates
| dec 2024: | our work on improving graph explainability has been accepted at AAAI'25! |
| dec 2024: | joining prof. aleksandar bojchevski's lab to work on solutions for graph foundational models |
| dec 2024: | starting joint work with prof. chirag agarwal from university of virginia on understanding explainability in multimodal language models |
| june 2024: | our work on evaluations in graph unlearning accepted at CoLLAs 2024 workshop track |
| may 2024: | joining adobe research, bangalore as an intern to work on problems related to LLM evaluations |
| jan 2024: | started working with the infonet team from deepmind (previously google research) on on leveraging topological information from user interactions to gather insight on the quality of user-generated content |
| may 2023: | i'm joining the precog research lab as an honors student, working with prof. ponnurangam kumaraguru "pk" |
| may 2022: | i'm selected as a google summer of code contributor, mentored by ankidroid |
| oct 2021: | i'm joining international institute of information technology, hyderabad as a computer science undergraduate student |
publications
Higher Order Structures for Graph Explanations
,
AAAI'25 |
The 39th Annual AAAI Conference on Artificial Intelligence
preprint
A Cognac shot to forget bad memories: Corrective Unlearning in GNNs
,
,
ArXiv Pre-print
pdf
Sanity Checks for Evaluating Graph Unlearning
,
,
CoLLAs-W
'24 |
The 3rd Conference on Lifelong Learning Agents (Workshop Track)
pdf