Deep neural networks have become a pervasive tool in science and engineering. However, modern deep neural networks’ growing energy requirements now increasingly limit their scaling and broader use. We propose a radical alternative for implementing deep neural network models: Physical Neural Networks. We introduce a hybrid physical-digital algorithm called Physics-Aware Training to efficiently train sequences of controllable physical systems to act as deep neural networks. This method automatically trains the functionality of any sequence of real physical systems, directly, using backpropagation, the same technique used for modern deep neural networks. To illustrate their generality, we demonstrate physical neural networks with three diverse physical systems-optical, mechanical, and electrical. Physical neural networks may facilitate unconventional machine learning hardware that is orders of magnitude faster and more energy efficient than conventional electronic processors.
Subjects: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Emerging Technologies (cs.ET); Optics (phys
— Lees op arxiv.org/abs/2104.13386
Blijf op de hoogte
Wekelijks inzichten over AI governance, cloud strategie en NIS2 compliance — direct in je inbox.
[jetpack_subscription_form show_subscribers_total="false" button_text="Inschrijven" show_only_email_and_button="true"]Bescherm AI-modellen tegen aanvallen
Agentic AI ThreatsRisico's van autonome AI-systemen
AI Governance Publieke SectorVerantwoorde AI voor overheden
Cloud SoevereiniteitSoeverein in de cloud — het kan
NIS2 Compliance ChecklistStap-voor-stap naar NIS2-compliance
Klaar om van data naar doen te gaan?
Plan een vrijblijvende kennismaking en ontdek hoe Djimit uw organisatie helpt.
Plan een kennismaking →Ontdek meer van Djimit
Abonneer je om de nieuwste berichten naar je e-mail te laten verzenden.
