Learning, Information Theory, and Nonequilibrium Statistical Mechanics
Learning, Information Theory, and Nonequilibrium Statistical Mechanics
Berkeley workgroup on
Learning, Information Theory, & Nonequilibrium
Thermodynamics
Coordinates
EMAIL lineq (at) lists.berkeley.edu
LOCATION 560 Evans Halls, UC Berkeley
TIme 3:30 PM every other Friday (kinda)
WEB https://calmail.berkeley.edu/manage/list/listinfo/lineq@lists.berkeley.edu
Nix Barnett (UCD): Structured Transformations of Structured Processes: The ε-Transducer
12 February 2016
Computational mechanics quantifies structure in a stochastic process via its causal states, leading to the process’s minimal, optimal predictor---the ε-machine. We extend computational mechanics to communication channels coupling two processes, obtaining an analogous optimal model---the ε-transducer---of the stochastic mapping between them. Here, we lay the foundation of a structural analysis of communication channels, treating joint processes and processes with input. The result is a principled structural analysis of mechanisms that support information flow between processes.
Reference: http://arxiv.org/abs/1412.2690