AI summaryⓘ
The authors propose a new way to do machine learning that uses physical processes which naturally involve randomness and energy changes, aiming to save energy and speed up tasks. They focus on a type called thermodynamic computing, where systems follow Langevin dynamics controlled by adjustable energy levels. The authors show how to build and train common machine learning models within this framework using energy-based models and probabilistic graphical models. They also analyze how fast and energy-efficient these models can be and demonstrate a basic physical implementation using superconducting circuits affected by thermal noise. This work points to a future approach for energy-saving machine learning hardware based on thermodynamics.
thermodynamic computingLangevin dynamicsenergy-based modelsprobabilistic graphical modelsstochastic analog processesmachine learning hardwaresuperconducting circuitsthermal noiseenergy efficiency
Authors
Owen Lockwood, Jérémy Béjanin, Joost Bus, Christopher Chamberland, Patrick Huembeli, Frank Schäfer, Guillaume Verdon
Abstract
To address the escalating energy and latency demands of machine-learning workloads, we introduce a blueprint for an energy-efficient and fast thermodynamic computing stack that leverages stochastic analog processes in physical hardware. In this work, we focus on energy-based thermodynamic computing where the stochastic process is well described by Langevin dynamics with tunable energy potentials. The implementation of such potentials in physical hardware enables us to generate and sample from basic parameterized energy-based models. We demonstrate how to construct and train popular classes of machine learning models based on these hardware-native energy-based models, using the framework of probabilistic graphical models. We analyze the runtime and energy consumption of different models in this thermodynamic paradigm based on theoretical considerations and numerical studies. As a preliminary experimental realization of such hardware, we present our stochastic analog superconducting circuits driven by thermal noise. Together, these results outline a path toward energy-efficient thermodynamic hardware for probabilistic machine learning.