Analog electronic deep networks for fast and efficient inference

We propose an efficient approach for real-time inference using deep neural networks implemented through low-power analog electronic circuits. Although analog implementations can be extremely compact, they have been largely supplanted by digital designs, partly because of device mismatch effects due to fabrication imperfections. We propose a framework that exploits the power of deep learning to compensate for this mismatch by incorporating the measured device variations as constraints in the training process. This eliminates the need for mismatch minimization strategies and allows circuit complexity and power-consumption to be reduced to a minimum. Our results, based on large-scale simulations as well as a prototype VLSI chip implementation indicate a processing efficiency comparable to current state-of-art digital implementations. This method is suitable for future technology based on nanodevices with large variability, such as memristive arrays.