Bayesian probability

Electrical engineering

Neuromorphic engineering

Resistive random-access memory

Mathematics

CMOS

Reinforcement learning

Markov chain

Gene

Markov chain Monte Carlo

Statistics

Machine learning

Memristor

Engineering

Voltage

Artificial intelligence

Bayesian inference

Chemistry

Monte Carlo method

Electronic engineering

Computer science

Probabilistic logic

Artificial neural network

Biochemistry

Robustness (evolution)

Nature Electronics

Resistive memory technologies could be used to create intelligent systems that learn locally at the edge. However, current approaches typically use learning algorithms that cannot be reconciled with the intrinsic non-idealities of resistive memory, particularly cycle-to-cycle variability. Here, we report a machine learning scheme that exploits memristor variability to implement Markov chain Monte Carlo sampling in a fabricated array of 16,384 devices configured as a Bayesian machine learning model. We apply the approach experimentally to carry out malignant tissue recognition and heart arrhythmia detection tasks, and, using a calibrated simulator, address the cartpole reinforcement learning task. Our approach demonstrates robustness to device degradation at ten million endurance cycles, and, based on circuit and system-level simulations, the total energy required to train the models is estimated to be on the order of microjoules, which is notably lower than in complementary metal–oxide–semiconductor (CMOS)-based approaches. The non-ideal characteristics of resistive memory devices can be used to develop low-power and resilient probabilistic neuromorphic computing hardware, suitable for highly constrained edge-based applications.

In situ learning using intrinsic memristor variability via Markov chain Monte Carlo sampling