Towards a Theory of Quantum Gravity from Neural Networks

Abstract

Neural network is a dynamical system described by two different types of degrees of freedom: fast-changing non-trainable variables (e.g., state of neurons) and slow-changing trainable variables (e.g., weights and biases). We show that the non-equilibrium dynamics of trainable variables can be described by the Madelung equations, if the number of neurons is fixed, and by the Schrodinger equation, if the learning system is capable of adjusting its own parameters such as the number of neurons, step size and mini-batch size. We argue that the Lorentz symmetries and curved space-time can emerge from the interplay between stochastic entropy production and entropy destruction due to learning. We show that the non-equilibrium dynamics of non-trainable variables can be described by the geodesic equation (in the emergent space-time) for localized states of neurons, and by the Einstein equations (with cosmological constant) for the entire network. We conclude that the quantum description of trainable variables and the gravitational description of non-trainable variables are dual in the sense that they provide alternative macroscopic descriptions of the same learning system, defined microscopically as a neural network.

Quantization of events in the event-universe and the emergence of quantum mechanics

“…We draw attention to certain parallels between our approach, which leads to the emergence of quantum theory from DHT, and the neural network model of the universe presented in articles [35][36][37][38][39] . In that sense a machine learning, ML, (or neural network) is naturally identified with a probability density function p(x; θ ) depending on a set of parameters.…”

Quasi-Equilibrium States and Phase Transitions in Biological Evolution

“…This finding further strengthens the idea that biological evolution can be effectively modeled through learning dynamics, opening up possibilities for investigating various biological phenomena using the framework provided by the theory of learning. Indeed, numerous non-trivial emergent physical phenomena, including quantum mechanics [ 8 , 24 ] and gravity [ 8 , 25 ], as well as critical phenomena such as phase transitions [ 23 ] or scale invariance [ 26 ], have already been derived from learning dynamics. This paper, along with Refs.…”