arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 2101.09671cs.CVcs.AI
收藏

用于深度神经网络加速的剪枝和Quantization:综述

Pruning and Quantization for Deep Neural Network Acceleration: A Survey

University of Science and Technology Beijing(北京科技大学);Hua Xia General Processor Technologies(华夏通用处理器技术公司);General Processor Technologies(通用处理器技术公司)

Tailin Liang, John Glossner, Lei Wang, Shaobo Shi, Xiaotong Zhang

摘要:

深度神经网络已经应用于在计算机视觉领域中表现出非凡能力的许多应用中。然而,复杂的网络架构对高效的实时部署提出了挑战,并需要大量的计算资源和能源成本。这些挑战可以通过网络压缩等优化来克服。网络压缩通常可以在精度损失很小的情况下实现。在某些情况下,准确性甚至可以提高。本文综述了两种网络压缩方法:剪枝和量化。修剪可以被分类为静态的(如果它是离线执行的)或动态的(如果它是在运行时执行的)。我们比较修剪技术,并描述用于删除冗余计算的标准。我们讨论了在元素方面,通道方面,形状方面,过滤器方面,层方面,甚至网络方面的修剪的权衡。量化通过降低数据类型的精度来减少计算。权重、偏置和激活通常可以被量化为8位整数,尽管还讨论了包括二进制神经网络的较低位宽实现。修剪和量化可以独立使用或组合使用。我们比较了当前的技术,分析了它们的优缺点,在一些框架上展示了压缩网络的准确性结果,并为压缩网络提供了实用的指导。

英文摘要:

Deep neural networks have been applied in many applications exhibiting extraordinary abilities in the field of computer vision. However, complex network architectures challenge efficient real-time deployment and require significant computation resources and energy costs. These challenges can be overcome through optimizations such as network compression. Network compression can often be realized with little loss of accuracy. In some cases accuracy may even improve. This paper provides a survey on two types of network compression: pruning and quantization. Pruning can be categorized as static if it is performed offline or dynamic if it is performed at run-time. We compare pruning techniques and describe criteria used to remove redundant computations. We discuss trade-offs in element-wise, channel-wise, shape-wise, filter-wise, layer-wise and even network-wise pruning. Quantization reduces computations by reducing the precision of the datatype. Weights, biases, and activations may be quantized typically to 8-bit integers although lower bit width implementations are also discussed including binary neural networks. Both pruning and quantization can be used independently or combined. We compare current techniques, analyze their strengths and weaknesses, present compressed network accuracy results on a number of frameworks, and provide practical guidance for compressing networks.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑