arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2002.07259cs.LGcs.NEmath.OCstat.ML
收藏

使用混合整数规划识别人工神经网络架构中的关键神经元

Identifying Critical Neurons in ANN Architectures using Mixed Integer Programming

Université de Montréal(蒙特利尔大学);Mila(米拉研究所)

Mostafa ElAraby, Guy Wolf, Margarida Carvalho

AI总结:

本研究提出基于混合整数规划的神经元重要性评分方法,可最小化维持网络精度所需的关键神经元数量,推广彩票优化并实现可扩展剪枝,在常用数据集和架构上以极小精度损失完成剪枝。

摘要:

我们提出一种混合整数规划(MIP,Mixed Integer Program)方法,用于为深度神经网络架构中的每个神经元分配重要性分数,该方法以同时剪枝这些神经元对网络主要学习任务的影响为指导。通过精心设计MIP的目标函数,我们引导求解器最小化为维持训练后神经网络的整体准确率而需要保留的关键神经元(即重要性分数高的神经元)的数量。此外,所提出的公式推广了近期被研究的彩票优化(lottery ticket optimization),可识别多个“幸运”子网络,由此得到的优化架构不仅在单个数据集上表现良好,而且在重新训练网络权重后能在多个数据集上泛化。最后,我们通过使用辅助网络解耦各层的重要性分数,实现了该方法的可扩展实现。我们在常用数据集和架构上验证了,该公式能够在准确率和泛化性损失极小的情况下完成神经网络剪枝。

英文摘要:

We introduce a mixed integer program (MIP) for assigning importance scores to each neuron in deep neural network architectures which is guided by the impact of their simultaneous pruning on the main learning task of the network. By carefully devising the objective function of the MIP, we drive the solver to minimize the number of critical neurons (i.e., with high importance score) that need to be kept for maintaining the overall accuracy of the trained neural network. Further, the proposed formulation generalizes the recently considered lottery ticket optimization by identifying multiple "lucky" sub-networks resulting in optimized architecture that not only performs well on a single dataset, but also generalizes across multiple ones upon retraining of network weights. Finally, we present a scalable implementation of our method by decoupling the importance scores across layers using auxiliary networks. We demonstrate the ability of our formulation to prune neural networks with marginal loss in accuracy and generalizability on popular datasets and architectures.

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。

↑