Ari Benjamin

In 2015, I quit a PhD in molecular simulation after witnessing neural networks translate between human languages. I had to understand deep learning. I am now a computational neuroscientist specializing in neural networks and their relationship to the brain, asking how and why neural networks learn what they do.

About half my work is theoretical. I have shown that that networks naturally develop ‘efficient’ codes for the world, in an info-theoretic sense, when they learn with gradient descent (Nature Communications, 2022); I’ve studied how to measure and regularize networks in the space of functions they compute (ICLR, 2019); and showed in stoplighted work at NeurIPS that any single network can be understood as a Bayesian ensemble, which explains to understand patterns in catastrophic forgetting. NeurIPS, 2024.

The other half of my work is data. I develop machine learning algorithms to analyze neural data, especially data which reveals cellular diversity. Recently I built TissueFormer, a transformer that extends single-cell foundation models to predict tissue-level phenotypes from populations of single cells (BMC Bioinformatics, 2026), and have worked on uncertainty-aware objectives for post-training language models.

I think the best way to understand the brain is through the lens of machine learning. I am especially drawn to what today’s neural networks don’t yet capture: neuromodulators like serotonin, the diversity of cell types in the brain, and how brains keep learning for a lifetime without overwriting what they know.

I am currently a postdoctoral fellow in the laboratory of Tony Zador at Cold Spring Harbor Laboratory. I completed my PhD with Konrad Kording at the University of Pennsylvania.