Fault-Tolerant Neural Networks from Biological Error Correction Codes
This addresses the challenge of reliable computation in noisy analog systems, such as the mammalian cortex and artificial intelligence applications, by providing a mechanism for fault tolerance, though it is incremental in applying known biological codes to neural networks.
The paper tackled the problem of achieving fault-tolerant computation in neural networks using unreliable neurons, by developing a universal fault-tolerant neural network based on biological error correction codes, which achieves reliable computation when neuron faultiness is below a sharp threshold, with noisy biological neurons found to be below this threshold.
It has been an open question in deep learning if fault-tolerant computation is possible: can arbitrarily reliable computation be achieved using only unreliable neurons? In the grid cells of the mammalian cortex, analog error correction codes have been observed to protect states against neural spiking noise, but their role in information processing is unclear. Here, we use these biological error correction codes to develop a universal fault-tolerant neural network that achieves reliable computation if the faultiness of each neuron lies below a sharp threshold; remarkably, we find that noisy biological neurons fall below this threshold. The discovery of a phase transition from faulty to fault-tolerant neural computation suggests a mechanism for reliable computation in the cortex and opens a path towards understanding noisy analog systems relevant to artificial intelligence and neuromorphic computing.