Eiji Kawasaki

A biologically grounded framework in which noisy neural and synaptic dynamics perform inference and learning via stochastic sampling from an internal energy function, capturing uncertainty over latent states and model parameters through neural and synaptic variability, respectively is discussed.

This report demonstrates using a detailed transaction-level model (TLM) of a probabilistic analogue in-memory computing (AIMC) processor that the same energy-based model can execute well over 1000x faster than data-center-grade hardware by eliminating the HBM interface and performing computation directly within on-chip memory.

MSE training can yield chemically accurate potentials that still fail at sampling, but a simple contrastive correction can fix this critical issue.

AI can now estimate thermodynamic properties of complex materials with chemical disorder, rivaling traditional Monte Carlo methods in accuracy but with significantly reduced computational cost.