Researchers from Washington University in St. Louis present an adaptive synaptic array that can be used to improve the energy-efficiency of training machine learning (ML) systems
Abstract
“In this paper we present an adaptive synaptic array that can be used to improve the energy-efficiency of training machine learning (ML) systems. The synaptic array comprises of an ensemble of analog memory elements, each of which is a micro-scale dynamical system in its own right, storing information in its temporal state trajectory. The state trajectories are then modulated by a system level learning algorithm such that the ensemble trajectory is guided towards the optimal solution. We show that the extrinsic energy required for state trajectory modulation can be matched to the dynamics of neural network learning which leads to a significant reduction in energy-dissipated for memory updates during ML training. Thus, the proposed synapse array could have significant implications in addressing the energy-efficiency imbalance between the training and the inference phases observed in artificial intelligence (AI) systems.”
Find the open access technical paper here. Published 03-2022.
Mehta, D., Rahman, M., Aono, K. et al. An adaptive synaptic array using Fowler–Nordheim dynamic analog memory. Nat Commun 13, 1670 (2022). https://doi.org/10.1038/s41467-022-29320-6.
Visit Semiconductor Engineering’s Technical Paper library here and discover many more chip industry academic papers.
Suppliers are investing new 300mm capacity, but it’s probably not enough. And despite burgeoning 200mm demand, only Okmetic and new players in China are adding capacity.
Different interconnect standards and packaging options being readied for mass chiplet adoption.
Continued expansion in new and existing markets points to massive and sustained growth.
Experts at the Table: Designing for context, and geopolitical impacts on a global supply chain.
Funding rolls in for photonics and batteries; 88 startups raise $1.3B.
Interest in this particular ISA is expanding, but the growth of other open-source hardware is less certain.
Nanosheets are likeliest option throughout this decade, with CFETs and other exotic structures possible after that.
Hybrid bonding opens up whole new level of performance in packaging, but it’s not the only improvement.
Why this is becoming a bigger issue, and what can be done to mitigate the effects.
Suppliers are investing new 300mm capacity, but it’s probably not enough. And despite burgeoning 200mm demand, only Okmetic and new players in China are adding capacity.
From low resistance vias to buried power rails, it takes multiple strategies to usher in 2nm chips.
Manufacturing 3D structures will require atomic-level control of what’s removed and what stays on a wafer.
Disaggregation and the wind-down of Moore’s Law have changed everything.
Leave a Reply