Jayanth Koushik

Email GitHub Google Scholar Twitter

I am a researcher interested in computer vision. I am particularly interested in object categorization from limited data. I have a PhD in neural computation and machine learning from Carnegie Mellon University, where I was advised by Mike Tarr and Aarti Singh.

Publications and Preprints

Deep learning powered real-time identification of insects using citizen science data [pnas]
S. Chiranjeevi, M. Sadaati, Z. K. Deng, J. Koushik, T. Z. Jubery, D. Mueller, M. E. O’Neal, N. Merchant, A. Singh, A. K. Singh, S. Sarkar, A. Singh, B. Ganapathysubramanian.
In Proceedings of the National Academy of Sciences (PNAS Nexus 2024).

AlphaNet: Improving Long-Tail Classification By Combining Classifiers [arxiv] [web] [code]
N. Chang*, J. Koushik*, A. Singh, M. Hebert, Y.-X. Wang, M. J. Tarr.
arXiv:2008.07073 (2023).

Deep Black-Box Optimization with Influence Functions [web]
J. Koushik, Michael J. Tarr, Aarti Singh.
Preprint (2020).

Eve: A Gradient Based Optimization Method with Locally and Globally Adaptive Learning Rates [arxiv] [web] [code]
H. Hayashi*, J. Koushik*, G. Neubig.
arXiv:1611.01505 (2018).

A Brain Phenotype for Stressor-Evoked Blood Pressure Reactivity [jaha]
P. J. Gianaros, L. K. Sheu, F. Uyar, J. Koushik, J. R. Jennings, T. D. Wager, A. Singh, T. D. Verstynen.
In Journal of the American Heart Association (2017).

Hypothesis Transfer Learning via Transformation Functions [arxiv]
S. S. Du, J. Koushik, A. Singh, B. Póczos.
In 31st Conference on Neural Information Processing Systems (NIPS 2017).

Deep Multimodal Fusion for Persuasiveness Prediction [acm dl]
B. Nojavanasghari*, D. Gopinath*, J. Koushik*, T. Baltrušaitis, L.-P. Morency.
In 18th International Conference on Multimodal Interaction (ICMI 2016).

*Equal contribution.

Projects

  • shiny-pyseed: opinionated bootstrapper for Python projects [code] [docs]
  • corgy: feature rich Python data classes [code] [docs]
  • paje: website creation and deployment system [code] [demo]
  • torch-gel: PyTorch implementation of group elastic-net [code]
  • neural-style: Theano/Keras implementation of style transfer algorithms [code] [demo]

Presentations

Influence Functions for Black-Box Optimization
J. Koushik, A. Singh, M. J. Tarr.
Talk at 2nd Annual Conference on Machine Learning and Engineering (MS&T 2019).

Influence Functions for Adaptive Stimulus Selection
J. Koushik, A. Marcus, A. Singh, M. J. Tarr.
Poster at 18th Annual Meeting of the Vision Sciences Society (VSS 2018).