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 2106.09647cs.LGstat.ML
收藏

通过示例难度的镜头进行深度学习

Deep Learning Through the Lens of Example Difficulty

Google Research(谷歌研究院)

Robert J. N. Baldock, Hartmut Maennel, Behnam Neyshabur

摘要:

现有的理解深度学习的工作通常采用将所有数据相关信息压缩为几个数字的方法。在这项工作中,我们采用了一种基于个别例子的作用的观点。我们引入了一个衡量对给定输入进行预测的计算难度的指标:(有效)预测深度。我们的广泛调查揭示了给定输入的预测深度与模型的不确定性,置信度,准确性和该数据点的学习速度之间令人惊讶而简单的关系。我们进一步将困难的例子分类为三个可解释的组,展示这些组在深度模型中如何以不同的方式处理,并展示这种理解如何使我们提高预测准确性。从我们的研究中获得的见解导致了文献中一些单独报告的现象的连贯性:早期层概括,而后期层记忆;早期层收敛速度更快,网络首先学习简单的数据和简单的函数。

英文摘要:

Existing work on understanding deep learning often employs measures that compress all data-dependent information into a few numbers. In this work, we adopt a perspective based on the role of individual examples. We introduce a measure of the computational difficulty of making a prediction for a given input: the (effective) prediction depth. Our extensive investigation reveals surprising yet simple relationships between the prediction depth of a given input and the model's uncertainty, confidence, accuracy and speed of learning for that data point. We further categorize difficult examples into three interpretable groups, demonstrate how these groups are processed differently inside deep models and showcase how this understanding allows us to improve prediction accuracy. Insights from our study lead to a coherent view of a number of separately reported phenomena in the literature: early layers generalize while later layers memorize; early layers converge faster and networks learn easy data and simple functions first.

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。

↑