First complete memristor neural network for medical AI
Cette publication existe aussi en Français
Researchers in France have developed a neural network based on memristor technology that has been used for real world medical analysis.
The team from CEA-Leti, CEA-List and two CNRS laboratories presented the first complete memristor-based Bayesian neural network implementation for a real-world AI task, classifying types of arrhythmia recordings with precise aleatoric and epistemic uncertainty.
Considering medical-diagnosis and other safety-critical, sensory-processing applications that require accurate decisions based on a small amount of noisy input data, the study highlights that while Bayesian neural networks excel at such tasks because they provide predictive uncertainty assessment, the probabilistic nature requires increased use of energy and computation. The increase is caused by the fact that implementing the networks in hardware requires a random number generator to store the probability distributions, the synaptic weights.
“Our paper presents, for the first time, a complete hardware implementation of a Bayesian neural network utilizing the intrinsic variability of memristors to store these probability distributions,” said Elisa Vianello, CEA-Leti chief scientist and co-author of the paper, “Bringing Uncertainty Quantification to the Extreme-Edge with Memristor-Based Bayesian Neural Networks”. “We exploited the intrinsic variability of memristors to store these probability distributions, instead of using random number generators.”
- Memristors drive digital reservoir neural network
- Memory circuit can learn for neuromorphic AI
- Scientists create brain synapses array using memristors
- 48 core neuromorphic AI chip uses resistive memory
A second major challenge was that performing inference in the team’s approach requires massive parallel multiply-and-accumulate (MAC) operations.
“These operations are power-hungry when carried out on CMOS-based ASICs and field-programmable gate arrays, due to the shuttling of data between processor and memory,” Vianello said. “In our solution, we use crossbars of memristors that naturally implement the multiplication between the input voltage and the probabilistic synaptic weight through Ohm’s law, and the accumulation through Kirchhoff’s current law, to significantly lower power consumption.”
The team also had to reconcile the nature of memristors, whose statistical effects adhere to the laws of device physics, with Bayesian neural networks, in which these effects can take arbitrary shapes.
“This work overcomes that challenge with a new training algorithm – variational inference augmented by a ‘technological loss’ – that accommodates device non-idealities during the learning phase,” said Damien Querlioz, a researcher at the University of Paris-Saclay, the French National Centre for Scientific Research (CNRS) and the Centre of Nanosciences and Nanotechnologies,. “Our approach enables the Bayesian neural network to be compatible with the imperfections of our memristors.”
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
