Comparing Information-Theoretic Measures of Complexity in Boltzmann Machines

被引:11
作者
Kanwal, Maxinder S. [1 ]
Grochow, Joshua A. [2 ,3 ,4 ]
Ay, Nihat [4 ,5 ,6 ]
机构
[1] Univ Calif Berkeley, Dept Elect Engn & Comp Sci, Berkeley, CA 94720 USA
[2] Univ Colorado, Dept Comp Sci, Boulder, CO 80309 USA
[3] Univ Colorado, Dept Math, Boulder, CO 80309 USA
[4] Santa Fe Inst, Santa Fe, NM 87501 USA
[5] Max Planck Inst Math Sci, D-04103 Leipzig, Germany
[6] Univ Leipzig, Fac Math & Comp Sci, D-04009 Leipzig, Germany
基金
美国国家科学基金会;
关键词
complexity; information integration; information geometry; Boltzmann machine; Hopfield network; Hebbian learning; INTEGRATION;
D O I
10.3390/e19070310
中图分类号
O4 [物理学];
学科分类号
0702 ;
摘要
In the past three decades, many theoretical measures of complexity have been proposed to help understand complex systems. In this work, for the first time, we place these measures on a level playing field, to explore the qualitative similarities and differences between them, and their shortcomings. Specifically, using the Boltzmann machine architecture (a fully connected recurrent neural network) with uniformly distributed weights as our model of study, we numerically measure how complexity changes as a function of network dynamics and network parameters. We apply an extension of one such information-theoretic measure of complexity to understand incremental Hebbian learning in Hopfield networks, a fully recurrent architecture model of autoassociative memory. In the course of Hebbian learning, the total information flow reflects a natural upward trend in complexity as the network attempts to learn more and more patterns.
引用
收藏
页数:16
相关论文
共 23 条
[1]  
[Anonymous], 2016, INFORM GEOMETRY ITS
[2]  
[Anonymous], [No title captured]
[3]   Information Geometry on Complexity and Stochastic Interaction [J].
Ay, Nihat .
ENTROPY, 2015, 17 (04) :2432-2458
[4]   Practical Measures of Integrated Information for Time-Series Data [J].
Barrett, Adam B. ;
Seth, Anil K. .
PLOS COMPUTATIONAL BIOLOGY, 2011, 7 (01)
[5]   Predictability, complexity, and learning [J].
Bialek, W ;
Nemenman, I ;
Tishby, N .
NEURAL COMPUTATION, 2001, 13 (11) :2409-2463
[6]  
Crutchfield J. P., COMPLEX SYSTEMS THEO
[7]   Regularities unseen, randomness observed: Levels of entropy convergence [J].
Crutchfield, JP ;
Feldman, DP .
CHAOS, 2003, 13 (01) :25-54
[8]  
Csiszar I., 2004, Foundations and Trends in Communications and Information Theory, V1, P1, DOI 10.1561/0100000004
[9]  
Edlund J.A., 2011, PLOS COMPUT BIOL, V7, P1
[10]  
Gell-Mann Murray., 1994, The Quark and the Jaguar. Adventures in the Simple and the Complex