New P-VAE model links information theory to metabolic cost

Researchers have developed a Poisson variational autoencoder (P-VAE) that incorporates a metabolic cost into information processing theories. This model links abstract information-theoretic quantities like coding rate to biophysical variables such as firing rate, enabling a trade-off between coding fidelity and energy expenditure. Unlike standard Gaussian VAEs, the P-VAE's Kullback-Leibler divergence term becomes proportional to prior firing rates, creating an emergent metabolic cost that penalizes high baseline activity. This approach offers a foundation for a resource-constrained theory of computation. AI

RANK_REASON This is a research paper published on arXiv detailing a new model. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv stat.ML →

• arXiv
• Gaussian VAEs
• GreLU-VAE
• Hadi Vafaii
• Kullback--Leibler divergence
• Poisson variational autoencoder

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →