Yash Mehta's Website

Located in Boston, USA

Hi! I am currently a researcher at DBMI, Harvard Medical School, working with Pranav Rajpurkar. I’m excited about multimodal foundation models, focused on applications in healthcare. Throughout my academic journey, I have gained valuable experience in various domains, including transformers, LLMs, neuroscience, and cognition. I am very fortunate to have gotten the opportunity to work with brilliant researchers, who have really inspired me and shaped my attitude towards research.

Previously, I was a research engineer at the amazing HHMI Janelia Research Campus, working on inferring synaptic plasticity rules using deep learning with James Fitzgerald and Jan Funke. I have worked on Neural Architecture Search in Frank Hutter’s AutoML lab in the picturesque town of Freiburg! Before that, I had an amazing time at the Gatsby Computational Neuroscience Unit at UCL, where I was working on evaluating biologically plausible perturbation-based learning algorithms to train deep networks with Tim Lillicrap at DeepMind and Peter Latham. I did my undergraduate thesis at NTU Singapore with Erik Cambria on using large language models for personality prediction. I quit my job as a software developer at Amazon to dive into the world of research ;)

I thoroughly enjoy coding and working on hard algorithmic problems.

News

Oct 23, 2023	Shifted to Boston, joined the Rajpurkar lab to work on AI+health
Jul 20, 2023	Attended ICML 2023 in Hawaii! ✨🍻
Dec 1, 2022	Attended NeurIPS 2022 in New Orleans! ✨🎷
Oct 1, 2022	Visiting student researcher in Larry Abbott’s lab at the Zuckerman Institute, Columbia University
Sep 14, 2022	Manuscript on node perturbation learning accepted to NeurIPS!
May 27, 2022	Got Married!
Apr 25, 2022	Presented our paper, NAS-BenchSuite in ICLR’22.
Jan 10, 2022	Started at HHMI Janelia Research Campus in the Funke Lab

Selected Publications

NeurIPS

Towards Biologically Plausible Convolutional Networks

Roman Pogodin, Yash Mehta, Timothy Lillicrap, and Peter Latham

In Advances in Neural Information Processing Systems (NeurIPS) 2021

Bib PDF Code

@inproceedings{pogodin2021towards,
  title = {Towards Biologically Plausible Convolutional Networks},
  author = {Pogodin, Roman and Mehta, Yash and Lillicrap, Timothy and Latham, Peter},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  volume = {34},
  year = {2021},
}

ICLR

NAS-Bench-Suite: NAS Evaluation is (Now) Surprisingly Easy

Yash Mehta, Colin White, Arber Zela, Arjun Krishnakumar, and 5 more authors

In International Conference on Learning Representations (ICLR) 2022

Bib PDF Code

@inproceedings{mehta2022bench,
  title = {NAS-Bench-Suite: NAS Evaluation is (Now) Surprisingly Easy},
  author = {Mehta, Yash and White, Colin and Zela, Arber and Krishnakumar, Arjun and Zabergja, Guri and Moradian, Shakiba and Safari, Mahmoud and Yu, Kaicheng and Hutter, Frank},
  booktitle = {International Conference on Learning Representations (ICLR)},
  year = {2022},
}

NeurIPS

Stability and Scalability of Node Perturbation Learning

Naoki Hiratani, Yash Mehta, Timothy Lillicrap, and Peter Latham

In Advances in Neural Information Processing Systems (NeurIPS) 2022

Bib PDF Code

@inproceedings{hiratani2022stability,
  title = {Stability and Scalability of Node Perturbation Learning},
  author = {Hiratani, Naoki and Mehta, Yash and Lillicrap, Timothy and Latham, Peter},
  booktitle = {Advances in Neural Information Processing Systems (NeurIPS)},
  year = {2022},
}