PredictiveNet: An energy-efficient convolutional neural network via zero prediction
PredictiveNet: An energy-efficient convolutional neural network via zero prediction
复制标题
DOI：
10.1109/iscas.2017.8050797
复制
发表时间：
2017-05
期刊：
影响因子：
--
通讯作者：
Yingyan Lin;Charbel Sakr;Yongjune Kim;Naresh R Shanbhag
中科院分区：
文献类型：
--
作者：
Yingyan Lin;Charbel Sakr;Yongjune Kim;Naresh R Shanbhag
卷积神经网络（CNN）因其在许多识别任务中破纪录的性能而引起了人们极大的兴趣。然而，CNN 的计算复杂性阻碍了它们在功率受限的嵌入式平台上的部署。在本文中，我们提出了预测 CNN (PredictiveNet)，它可以预测非线性层的稀疏输出，从而绕过大部分计算。 PredictiveNet 在运行时跳过 CNN 中的大部分卷积，无需修改 CNN 结构或需要额外的分支网络。提供了仿真支持的分析，以证明所提出的技术在保留非线性层输出均方误差 (MSE) 的能力方面的合理性。当应用于 CNN 进行手写数字识别时，模拟结果表明，与最先进的 CNN 相比，PredictiveNet 可以将计算成本降低 2.9 倍，同时会导致边际精度下降。
Convolutional neural networks (CNNs) have gained considerable interest due to their record-breaking performance in many recognition tasks. However, the computational complexity of CNNs precludes their deployments on power-constrained embedded platforms. In this paper, we propose predictive CNN (PredictiveNet), which predicts the sparse outputs of the non-linear layers thereby bypassing a majority of computations. PredictiveNet skips a large fraction of convolutions in CNNs at runtime without modifying the CNN structure or requiring additional branch networks. Analysis supported by simulations is provided to justify the proposed technique in terms of its capability to preserve the mean square error (MSE) of the nonlinear layer outputs. When applied to a CNN for handwritten digit recognition, simulation results show that PredictiveNet can reduce the computational cost by a factor of 2.9χ compared to a state-of-the-art CNN, while incurring marginal accuracy degradation.