Justin Kang
Welcome to my website. You can find my Publications, Projects and CV above. Below, you can find some information about me.
- I am a Graduate Student at the University of California, Berkeley, affiliated with the BAIR Lab and supervised by Prof. Kannan Ramchandran.
- Previously, I completed my Master’s of Applied Science at The University of Toronto supervised by Prof. Wei Yu.
- My research interests include machine learning, privacy in learning, and data valuation. Recently, I’ve been working on applying some ideas from sparse signal processing to explainable AI for LLMs. I have a strong background in signal processing and information theory, and like to view machine learning problems through that lens, which often leads to unique and exciting solutions.
- In 2019 I received my B.A.Sc. from the University of British Columbia in Engineering Physics.
Recent News
- My paper Learning to Understand: Identifying Interactions via the Mobius Transform has been accepted to NeurIPS!
- A recent article about Federated Learning that I was interviewed for.
- I’m happy to anounce that my paper has won the 2024 IEEE Communication Society and Information Theory Society Joint Paper Award. This is one of two paper awards
- I’ll be joining Google for summer 2024 as a Student Reseacher, working on a project with the Google Cloud Platforms Systems Research Group (SRG) hosted by Kun Lin and Prof. David Culler.
Selected Papers
SPEX: Scaling Feature Interaction Explanations for LLMs. Kang, J.S(e)., Butler, L.(e), Agarwal, A.(e),Erginbas Y.E., Pedarsani, R., Ramchandran, K., Yu, B., Preprint. (paper)
Learning to Understand: Identifying Interactions via the Mobius Transform. Kang, J.S., Erginbas, Y.E., Butler, L., Pedarsani, R., Ramchandran, K. (2024), NeurIPS, 2024. (paper, video)
Learning a 1-Layer Conditional Generative Model in Total Variation. Ajil Jalal, Justin Singh Kang, Ananya Uppal, Kannan Ramchandran, Eric Price. NeurIPS 2023. (paper, video, poster, code)
The Fair Value of Data Under Heterogeneous Privacy Constraints in Federated Learning. Justin Singh Kang, Ramtin Pedarsani and Kannan Ramchandran, NeurIPS FL@FM 2023, TMLR, 2024. (paper, video)
Efficiently Computing Sparse Fourier Transforms of q-ary Functions. Justin Singh Kang(e), Y. E. Erginbas(e), A. Aghazadeh and K. Ramchandran. IEEE ISIT 2023 (paper, video, code)
