Programmable Resistances

from Inaccurate Resistors

via Random Subset-Sum Theory

Emanuele Natale, Laurent Viennot and Arthur C. W. da Cunha

16 November 2022

The Deep Learning Revolution

Turing test (1950)


1st AI winter (1974–1980)

2nd AI winter (1974–1980)

"A hand lifts a cup"

Use of GPUs in AI (2011)

Today, most AI heavy lifting is done in the cloud due to the concentration of large data sets and dedicated compute, especially when it comes to the training of machine learning (ML) models. But when it comes to the application of those models in real-world inferencing near the point where a decision is needed, a cloud-centric AI model struggles. [...] When time is of the essence, it makes sense to distribute the intelligence from the cloud to the edge.

Inference Energy

Chakraborty, I. et al. Resistive crossbars as approximate hardware building blocks for machine learning: Opportunities and challenges. Proc. IEEE 108, 2276–2310 (2020).

AI Accelerators

Resistive Crossbar Device

Analog MVM via crossbars of programmable resistances

Problem: Making precise programmable resistances is hard

Cfr. ~10k flops for digital 100x100 MVM

The Strong Lottery Ticket Hypothesis

da Cunha, A., Natale, E. & Viennot, L. Proving the Strong Lottery Ticket Hypothesis for Convolutional Neural Networks. ICLR 2022

A network with random weights contains sub-networks that can approximate any given sufficiently-smaller neural network

"Résistance équivalente modulable

à partir de résistances imprécises"

INRIA Patent deposit FR2210217

Leverage noise itself

to increase precision



effective resistance

Random Subset Sum Theory

bits of precision  for any target value are linear w.r.t. number of resistances

Worst case among

2.5k instances

Thank you