New physics-based self-learning machines could replace current artificial neural networks and save energy

https://techxplore.com/news/2023-09-physics-based-self-learning-machines-current-artificial.html

"train AI more efficiently using. physical processes instead of digital artificial neural networks... neuromorphic computing... machine parameters optimized by process itself... not requiring external feedback not only saves energy but also computing time... able to run forwards or backwards... non-linear process... processes information in the form of superimposed light waves, whereby suitable components regulate the type and strength of the interaction"

Related:

New photonic neural networks promise ultrafast computing for complex tasks

New technique based on 18th-century mathematics shows simpler AI models don't need deep learning

New computing hardware needs a theoretical basis, says study

Beyond deep learning: Advancing affective computing with diverse AI methodologies

Contrastive learning simple neural networks find communities in complex networks with theoretical optimality, questioning need fort more complex models
https://techxplore.com/news/2025-01-ai-simpler-assumptions.html

Comments

Post a Comment