Building fast Bayesian computing machines out of intentionally stochastic, digital parts
This work addresses the problem of computational inefficiency in Bayesian inference for AI and neuroscience, offering a novel engineering approach with broad implications for intelligent computation.
The paper tackles the efficiency gap between brain-like Bayesian inference and modern computers by building fast Bayesian computing machines using intentionally stochastic digital parts, achieving a 1000x speed advantage over commodity microprocessors for inference over 10,000+ latent variables in real-time tasks like depth and motion perception.
The brain interprets ambiguous sensory information faster and more reliably than modern computers, using neurons that are slower and less reliable than logic gates. But Bayesian inference, which underpins many computational models of perception and cognition, appears computationally challenging even given modern transistor speeds and energy budgets. The computational principles and structures needed to narrow this gap are unknown. Here we show how to build fast Bayesian computing machines using intentionally stochastic, digital parts, narrowing this efficiency gap by multiple orders of magnitude. We find that by connecting stochastic digital components according to simple mathematical rules, one can build massively parallel, low precision circuits that solve Bayesian inference problems and are compatible with the Poisson firing statistics of cortical neurons. We evaluate circuits for depth and motion perception, perceptual learning and causal reasoning, each performing inference over 10,000+ latent variables in real time - a 1,000x speed advantage over commodity microprocessors. These results suggest a new role for randomness in the engineering and reverse-engineering of intelligent computation.