Search for common principles for efficient representation in both neural coding and AI | Siwei Wang

Recorded on 10/17/2025
Watch the recording without ads at https://www.nitmb.org/nitmb-seminar-s...

Speaker: Siwei Wang
Title: Search for common principles for efficient representation in both neural coding and AI
Abstract: In the wild, survival often depends on anticipating what happens next. A fly must escape a predator's strike before its visual system even finishes processing the threat. A salamander needs to track prey moving through completely different environments as it migrates from aquatic to terrestrial environments. How neural systems extract actionable insights within a split-second while neural processing is inherently laggy? Given this temporal constraint, predictive information is one such common computation. By encoding patterns predictive of future maneuvers, specialized motion sensitive neurons in the fly visual system enables them to steer animals away from visual threat within 40 ms, half the time as our eye blinks. Similarly, in the salamander retina, the encoding of predictive information sculpts the population code to establish a low-dimensional, transferable representation that seamlessly generalizes across radically different natural scenes. In addition, when this representation combines static and dynamic features synergistically to anticipate future states, its compression rivals the most state-of-the-art video compression algorithms.

The above insights from biology inspires new theories in AI. By extracting the general principles for the above biological efficient representations, the newest work from my group demonstrates that such principle also applies to feature engineering in contrastive learning writ large. The overarching goal of this walk is to establish a virtuous cycle where developing biologically plausible AI systems helps interrogate unknown encoding mechanisms in the brain, while these discovered neural principles simultaneously inspire more efficient computational architecture.

Part of the NITMB Seminar Series

The NSF-Simons National Institute for Theory and Mathematics in Biology Seminar Series aims to bring together a mix of mathematicians and biologists to foster discussion and collaboration between the two fields. The seminar series will take place on Fridays from 10am - 11am at the NITMB in the John Hancock Center in downtown Chicago. There will be both an in-person and virtual component.

The NSF-Simons National Institute for Theory and Mathematics in Biology (NITMB) aims to integrate the disciplines of mathematics and biology in order to transform the practice of biological research
and to inspire new mathematical discoveries. NITMB is a partnership between Northwestern University and the University of Chicago. It is funded by the National Science Foundation DMS-2235451 and the Simons Foundations MP-TMPS-00005320.

The mission of the NITMB is to create a nationwide collaborative research community that will generate new mathematical results and uncover the “rules of life” through theories, data-informed mathematical models, and computational and statistical tools. The NITMB leverages close collaborations between experimentalists and theorists to synergize discovery. The fundamental research done by NITMB will stimulate advances in areas as diverse as the environment, medicine, and technology development. NITMB members and visitors share space in downtown Chicago that is readily accessible to collaborators across the U.S. and the world.

NITMB uses an interlocking set of strategies and initiatives aimed at broad impacts for the mathematical and biological research communities. Targeted research bringing together mathematicians and biologists to collaborate and train the next generation of interdisciplinary scientists. Scientific long programs, workshops, and conferences enhancing collaboration between mathematics and biology. An innovative research program organized around five interrelated themes, selected because they reflect key capabilities of biological systems and interconnect with open mathematical problems.