The paper investigates how temperature affects analog deep neural network (DNN) inference, focusing on both stochastic and systematic non‑idealities in analog hardware. Experiments show that temperature‑induced performance loss is mainly driven by systematic errors rather than random noise. The study evaluates various mitigation techniques, finding that noise‑aware training and temperature‑aware calibration—especially hardware‑in‑the‑loop training—best preserve inference accuracy across different thermal conditions.
By Niklas Summ, Xiao Wang, Hendrik Borras, Bernhard Klein, Holger Fr\"oning
The article explores how biological learning and decision-making, often modeled as Bayesian processes, can be replicated in computing systems by leveraging noisy neural and synaptic dynamics for stochastic sampling. It proposes a biologically grounded framework where internal energy functions capture uncertainty over latent states and model parameters, enabling predictive coding networks to perform Markov chain Monte Carlo sampling. By drawing parallels between intrinsic biological noise and electrical noise in emerging probabilistic analogue memory technologies, the authors argue that analogue in‑memory computing hardware offers a massively scalable and energy‑efficient solution for probabilistic inference.
By Thomas Dalgaty, Eiji Kawasaki, Miguel de Prado, Devendra Vyas, Tommaso Salvatori
arXiv:2608. 01615v1 Announce Type: cross Abstract: We present a set of tools for mapping general stochastic programs to thermodynamic hardware designed for energy-efficient stochastic sampling.
By Mirko Amico, Andra\v{z} Jelin\v{c}i\v{c}, Colin Oscar Nancarrow, Leo Tyrpak, David Roberts, Seth Morton, Dalton Sakthivadivel, Ashwin Gopal, Guillaume Verdon
arXiv:2606. 27294v1 Announce Type: cross Abstract: Analog hardware platforms such as coupled oscillators and Analog Ising Machines naturally solve differential equations at a fraction of the energy cost of digital computation, making them attractive for low-power generative modeling, yet a fundamental mismatch exists: modern generative models assume flexible, software-defined dynamics, whereas analog hardware imposes fixed, physics-determined differential equations with limited approximation capacity.
By Yu-Neng Wang, Sara Achour
Thermodynamic computing devices based on the Ising model show great promise for low-power AI inference and edge computing, but scalable methods for training large models for such hardware remain limited. Prior theory shows that the time-averaged behavior of high-temperature Gibbs-sampled Ising systems can implement feed-forward neural inference.
arXiv:2607. 00170v1 Announce Type: cross Abstract: Thermodynamic computing devices based on the Ising model show great promise for low-power AI inference and edge computing, but scalable methods for training large models for such hardware remain limited.
By Andrew G. Moore
The paper introduces Langevin simulated bifurcation (LSB), a fast, parallel Boltzmann sampler that matches the accuracy of sequential MCMC methods. It also proposes conditional expectation matching (CEM), an efficient technique for estimating the effective temperature of samples from energy‑based models with conditional independence. Building on these, the authors develop sampler adaptive learning (SAL), which adjusts the model temperature to align with the distribution produced by LSB, enabling efficient training of semi‑restricted Boltzmann machines (SRBMs) and outperforming conventional methods on synthetic spin‑glass datasets.
By Kentaro Kubo, Hayato Goto
arXiv:2606. 04582v1 Announce Type: cross Abstract: Real-time monitoring of the temperature distribution within components and sub-structures is a challenging topic in many systems due to restrictions on feasible sensor locations.
By Monika Stipsitz, H\`elios Sanchis-Alepuz, Jacob Reynvaan, Silvester Sabathiel
arXiv:2609.15439v1 Announce Type: cross
Abstract: Generative thermodynamic computers turn thermal noise into structured data through Langevin dynamics. We train these systems with a local update at e...
By Huilin Wang, Weibing Deng
The paper investigates the thermodynamic cost of inference and learning in physical neural networks. It shows that quasi‑static inference requires no work, while finite‑speed inference incurs work bounded by the Wasserstein‑2 distance between thermal states, roughly $k_B T$ per dimension of the widest layer. Learning, however, has an irreducible cost of a few $k_B T$ per parameter, independent of speed, indicating that memory dominates the thermodynamic price.
By Alexei V. Tkachenko
arXiv:2607. 05187v1 Announce Type: new Abstract: As CMOS technology scales into the deep nanometer regime, digital circuit reliability is increasingly threatened by the combined stochastic effects of Bias Temperature Instability (BTI) and Process Variation (PV).
By Arash Esshaghi, Siavash Es'haghi, Gholamreza Shahabadi, Alireza Moradi
arXiv:2608. 12624v1 Announce Type: new Abstract: Structure-preserving machine learning embeds physical structure directly into model architectures, yet uncertainty quantification (UQ) for such hard-constrained models remains limited because standard UQ methods may violate the encoded admissibility conditions, require architectural modifications, or impose substantial computational costs.
By Zequn He, Celia Reina