About Mitchell Ostrow
Computational Neuroscience PhD Student at MIT. Reverse-Engineering the Brain and Deep Neural Networks
I seek to bridge the gaps between nervous systems and deep learning by developing a quantitative understanding of how + what the brain computes. I design and use methods from statistics, dynamical systems theory, and representation learning to analyze the computations performed by both biological and artificial neural networks. When used as a scientific model, deep neural networks can be used to study brain function in ways beyond contemporary experimental techniques. Through these approaches, I want to design brain-inspired artificial intelligence systems, build brain-machine interfaces, and identify the etiology of psychiatric disorders. Iâve worked in medicine (as an EMT), experimental neuroscience, computational neuroscience, and artificial intelligence (in industry).
Iâm fortunate to have been supported at MIT by the Computationally-Enabled Integrative Neuroscience Fellowship and the Praecis Presidential Fellowship.
Yale Symposia was kind enough to profile me and my undergraduate journey in neuroscience research.
Simply Neuroscience interviewed me on their podcast, The Synapse, about doing neuroscience research at Yale. Link here.
news
Jan 24, 2024 | My paper, âBeyond Geometry: Comparing the Temporal Structure of Computation in Neural Circuits with Dynamical Similarity Analysisâ was accepted to the Cognitive Computational Neuroscience conference as an oral presentation (4%) and to COSYNE (Computational Systems Neuroscience conference) as an oral, (2%)! |
---|---|
Jun 20, 2023 | My first full paper preprint is out! https://arxiv.org/abs/2306.10168, and was accepted to NeurIPS, 2023! |
Mar 31, 2023 | After four excellent rotations, I am joining Ila Fieteâs lab for my thesis work! |
Mar 28, 2023 | Our abstract âDo Deep Neural Networks Have Concepts?â was selected as a talk at the Philosophy of Deep Learning Conference, hosted at NYU |
Aug 28, 2022 | Starting PhD! đšđ»âđ |
selected publications
- Beyond Geometry: Comparing the Temporal Structure of Computation in Neural Circuits with Dynamic Similarity Analysis.In Cognitive Computational Neuroscience (CCN), selected as an oral presentation (5%) 2023
- Beyond Geometry: Comparing the Temporal Structure of Computation in Neural Circuits with Dynamic Similarity Analysis.In Computational Systems Neuroscience Conference (COSYNE), selected as an oral presentation (2%) 2023
- Beyond Geometry: Comparing the Temporal Structure of Computation in Neural Circuits with Dynamic Similarity Analysis.2023