Per-Tensor Fixed-Point Quantization of the Back-Propagation Algorithm
This work addresses the problem of slow training and deployment on resource-constrained systems for AI practitioners, offering a systematic approach to fixed-point quantization, though it is incremental as it builds on existing fixed-point methods.
The paper tackles the challenge of designing full fixed-point training for deep neural networks by proposing a precision assignment methodology that analytically determines near-minimal bit widths for all parameters, gradients, and accumulators, achieving complexity reduction validated on CIFAR-10, CIFAR-100, and SVHN datasets.
The high computational and parameter complexity of neural networks makes their training very slow and difficult to deploy on energy and storage-constrained computing systems. Many network complexity reduction techniques have been proposed including fixed-point implementation. However, a systematic approach for designing full fixed-point training and inference of deep neural networks remains elusive. We describe a precision assignment methodology for neural network training in which all network parameters, i.e., activations and weights in the feedforward path, gradients and weight accumulators in the feedback path, are assigned close to minimal precision. The precision assignment is derived analytically and enables tracking the convergence behavior of the full precision training, known to converge a priori. Thus, our work leads to a systematic methodology of determining suitable precision for fixed-point training. The near optimality (minimality) of the resulting precision assignment is validated empirically for four networks on the CIFAR-10, CIFAR-100, and SVHN datasets. The complexity reduction arising from our approach is compared with other fixed-point neural network designs.