Recurrence of Optimum for Training Weight and Activation Quantized Networks

• URL: http://arxiv.org/abs/2012.05529v1
• Date: Thu, 10 Dec 2020 09:14:43 GMT
• Title: Recurrence of Optimum for Training Weight and Activation Quantized Networks
• Authors: Ziang Long, Penghang Yin, Jack Xin
• Abstract summary: Training deep learning models with low-precision weights and activations involves a demanding optimization task. We show how to overcome the nature of network quantization. We also show numerical evidence of the recurrence phenomenon of weight evolution in training quantized deep networks.
• Score: 4.103701929881022
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Deep neural networks (DNNs) are quantized for efficient inference on resource-constrained platforms. However, training deep learning models with low-precision weights and activations involves a demanding optimization task, which calls for minimizing a stage-wise loss function subject to a discrete set-constraint. While numerous training methods have been proposed, existing studies for full quantization of DNNs are mostly empirical. From a theoretical point of view, we study practical techniques for overcoming the combinatorial nature of network quantization. Specifically, we investigate a simple yet powerful projected gradient-like algorithm for quantizing two-linear-layer networks, which proceeds by repeatedly moving one step at float weights in the negation of a heuristic \emph{fake} gradient of the loss function (so-called coarse gradient) evaluated at quantized weights. For the first time, we prove that under mild conditions, the sequence of quantized weights recurrently visits the global optimum of the discrete minimization problem for training fully quantized network. We also show numerical evidence of the recurrence phenomenon of weight evolution in training quantized deep networks.

Related papers

• SPFQ: A Stochastic Algorithm and Its Error Analysis for Neural Network Quantization [5.982922468400901]
  We show that it is possible to achieve error bounds equivalent to that obtained in the order of the weights of a neural layer. We prove that it is possible to achieve full-network bounds under an infinite alphabet and minimal assumptions on the input data.
  arXiv  Detail & Related papers  (2023-09-20T00:35:16Z)
• BiTAT: Neural Network Binarization with Task-dependent Aggregated Transformation [116.26521375592759]
  Quantization aims to transform high-precision weights and activations of a given neural network into low-precision weights/activations for reduced memory usage and computation. Extreme quantization (1-bit weight/1-bit activations) of compactly-designed backbone architectures results in severe performance degeneration. This paper proposes a novel Quantization-Aware Training (QAT) method that can effectively alleviate performance degeneration.
  arXiv  Detail & Related papers  (2022-07-04T13:25:49Z)
• Cluster-Promoting Quantization with Bit-Drop for Minimizing Network Quantization Loss [61.26793005355441]
  Cluster-Promoting Quantization (CPQ) finds the optimal quantization grids for neural networks. DropBits is a new bit-drop technique that revises the standard dropout regularization to randomly drop bits instead of neurons. We experimentally validate our method on various benchmark datasets and network architectures.
  arXiv  Detail & Related papers  (2021-09-05T15:15:07Z)
• A White Paper on Neural Network Quantization [20.542729144379223]
  We introduce state-of-the-art algorithms for mitigating the impact of quantization noise on the network's performance. We consider two main classes of algorithms: Post-Training Quantization (PTQ) and Quantization-Aware-Training (QAT)
  arXiv  Detail & Related papers  (2021-06-15T17:12:42Z)
• In-Hindsight Quantization Range Estimation for Quantized Training [5.65658124285176]
  We propose a simple alternative to dynamic quantization, in-hindsight range estimation, that uses the quantization ranges estimated on previous iterations to quantize the present. Our approach enables fast static quantization of gradients and activations while requiring only minimal hardware support from the neural network accelerator. It is intended as a drop-in replacement for estimating quantization ranges and can be used in conjunction with other advances in quantized training.
  arXiv  Detail & Related papers  (2021-05-10T10:25:28Z)
• Direct Quantization for Training Highly Accurate Low Bit-width Deep Neural Networks [73.29587731448345]
  This paper proposes two novel techniques to train deep convolutional neural networks with low bit-width weights and activations. First, to obtain low bit-width weights, most existing methods obtain the quantized weights by performing quantization on the full-precision network weights. Second, to obtain low bit-width activations, existing works consider all channels equally.
  arXiv  Detail & Related papers  (2020-12-26T15:21:18Z)
• Where Should We Begin? A Low-Level Exploration of Weight Initialization Impact on Quantized Behaviour of Deep Neural Networks [93.4221402881609]
  We present an in-depth, fine-grained ablation study of the effect of different weights initialization on the final distributions of weights and activations of different CNN architectures. To our best knowledge, we are the first to perform such a low-level, in-depth quantitative analysis of weights initialization and its effect on quantized behaviour.
  arXiv  Detail & Related papers  (2020-11-30T06:54:28Z)
• Optimal Gradient Quantization Condition for Communication-Efficient Distributed Training [99.42912552638168]
  Communication of gradients is costly for training deep neural networks with multiple devices in computer vision applications. In this work, we deduce the optimal condition of both the binary and multi-level gradient quantization for textbfANY gradient distribution. Based on the optimal condition, we develop two novel quantization schemes: biased BinGrad and unbiased ORQ for binary and multi-level gradient quantization respectively.
  arXiv  Detail & Related papers  (2020-02-25T18:28:39Z)
• Gradient $\ell_1$ Regularization for Quantization Robustness [70.39776106458858]
  We derive a simple regularization scheme that improves robustness against post-training quantization. By training quantization-ready networks, our approach enables storing a single set of weights that can be quantized on-demand to different bit-widths.
  arXiv  Detail & Related papers  (2020-02-18T12:31:34Z)
• Post-Training Piecewise Linear Quantization for Deep Neural Networks [13.717228230596167]
  Quantization plays an important role in the energy-efficient deployment of deep neural networks on resource-limited devices. We propose a piecewise linear quantization scheme to enable accurate approximation for tensor values that have bell-shaped distributions with long tails. Compared to state-of-the-art post-training quantization methods, our proposed method achieves superior performance on image classification, semantic segmentation, and object detection with minor overhead.
  arXiv  Detail & Related papers  (2020-01-31T23:47:00Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.