< Back to previous page
Project
Smart chips: embedding machine learning at circuit level.
The project targets to enable every circuit designer to use machine learning towards a 100-fold improvement of the energy efficiency of integrated circuits developed for sensing, computing and communication. This will be achieved by creating an automated way for the circuit designer to embed advanced machine learning techniques deeply into the hardware, while shielding him from the complexity of the machine learning algorithms. The introduced on-die intelligence will be capable of autonomously learning the chips limitations and its optimal operating modes to let the chip dynamically self-adapt towards an optimal energy-performance operating point.
Date:1 Oct 2013 → 30 Sep 2015
Keywords:Reconfiguration, Circuit-level, Chip, Energy-efficiency, Sensor interface, Complexity, Machine learning
Disciplines:Nanotechnology, Design theories and methods, Artificial intelligence, Cognitive science and intelligent systems