Critical Behavior in Physics and Probabilistic Formal Languages

2016/06/21 by Henry W. Lin, Max Tegmark · 1 voice · 11 citations
Physics and Astronomy · Computer Science · Social Sciences · #Statistical Mechanics and Entropy #Computability, Logic, AI Algorithms #Language and cultural evolution

paper · pdf · doi:10.3390/e19070299

Abstract

We show that the mutual information between two symbols, as a function of the number of symbols between the two, decays exponentially in any probabilistic regular grammar, but can decay like a power law for a context-free grammar. This result about formal languages is closely related to a well-known result in classical statistical mechanics that there are no phase transitions in dimensions fewer than two. It is also related to the emergence of power law correlations in turbulence and cosmological inflation through recursive generative processes. We elucidate these physics connections and comment on potential applications of our results to machine learning tasks like training artificial recurrent neural networks. Along the way, we introduce a useful quantity, which we dub the rational mutual information, and discuss generalizations of our claims involving more complicated Bayesian networks.

Citations

• On Information and Sufficiency

Cited by

• Language Modeling with Reduced Densities
• A scaling law of contextual persistence in human language
• Evaluating Computational Language Models with Scaling Properties of Natural Language
• Hyperbolic Deep Neural Networks: A Survey
• Gradual Forgetting: Logarithmic Compression for Extending Transformer Context Windows
• Correlation Dimension of Auto-Regressive Large Language Models
• Hyperbolic Attention Networks
• Slower is Better: Revisiting the Forgetting Mechanism in LSTM for Slower Information Decay
• Fundamentals of Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) network
• Quantum algorithms for equational reasoning
• Human-like fleeting memory improves language learning but impairs reading time prediction in transformer language models

Discussions

• Critical Behavior from Deep Dynamics: A Hidden Dimension in Natural Language [hn, 175 points, 44 comments]

Related

• First numerical observation of the Berezinskii-Kosterlitz-Thouless transition in language models
• Statistical Patterns in the Equations of Physics and the Emergence of a Meta-Law of Nature
• The physics of randomness and regularities for languages (lifetimes, family trees, and the second languages); in terms of random matrices
• Mutual Information Scaling and Expressive Power of Sequence Models
• Zipf's law emerges asymptotically during phase transitions in communicative systems