Reverse-Engineering Neural Networks to Characterize Their Cost Functions

DOI10.1162/neco_a_01315zbMath1453.92014OpenAlexW3087248248WikidataQ99569099 ScholiaQ99569099MaRDI QIDQ3386420

Takuya Isomura, Karl J. Friston

Publication date: 4 January 2021

Published in: Neural Computation (Search for Journal in Brave)

Full work available at URL: https://doi.org/10.1162/neco_a_01315

zbMATH Keywords

Markov decision processesneural networkscost functionsBayesian inference and learningneural activity and plasticity

Mathematics Subject Classification ID

Neural networks for/in biological studies, artificial life and related topics (92B20) Markov and semi-Markov decision processes (90C40)

Related Items (1)

On the Achievability of Blind Source Separation for High-Dimensional Nonlinear Source Mixtures

Uses Software

• Adam

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Action understanding and active inference
• A complete class theorem for statistical problems with finite sample spaces
• A unifying information-theoretic framework for independent component analysis
• Codes on graphs: normal realizations
• Markov blankets, information geometry and stochastic thermodynamics
• A Novel Predictive-Coding-Inspired Variational RNN Model for Online Prediction and Recognition
• Linking Neuromodulated Spike-Timing Dependent Plasticity with the Free-Energy Principle
• Active Inference: A Process Theory
• An Approximation of the Error Backpropagation Algorithm in a Predictive Coding Network with Local Hebbian Synaptic Plasticity
• Active Inference, Curiosity and Insight
• An Essentially Complete Class of Admissible Decision Functions
• On Information and Sufficiency

This page was built for publication: Reverse-Engineering Neural Networks to Characterize Their Cost Functions