arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1901.09192cs.LGstat.ML
收藏

SelectiveNet:一个具有集成可编程选项的深度神经网络

SelectiveNet: A Deep Neural Network with an Integrated Reject Option

Technion - Israel Institute of Technology(以色列理工学院)

Yonatan Geifman, Ran El-Yaniv

摘要:

我们考虑了深度神经网络中的选择性预测(也称为拒绝选项)问题,并引入了SelectiveNet,这是一种具有集成拒绝选项的深度神经架构。现有的拒绝机制主要基于预先训练的网络的预测置信度的阈值。相比之下,SelectiveNet经过训练,可以同时优化分类(或回归)和拒绝,端到端。结果是在覆盖域上优化的深度神经网络。在我们的实验中,我们在几个著名的分类和回归数据集上显示了持续改进的风险覆盖权衡,从而达到了深度选择性分类的最新结果。

英文摘要:

We consider the problem of selective prediction (also known as reject option) in deep neural networks, and introduce SelectiveNet, a deep neural architecture with an integrated reject option. Existing rejection mechanisms are based mostly on a threshold over the prediction confidence of a pre-trained network. In contrast, SelectiveNet is trained to optimize both classification (or regression) and rejection simultaneously, end-to-end. The result is a deep neural network that is optimized over the covered domain. In our experiments, we show a consistently improved risk-coverage trade-off over several well-known classification and regression datasets, thus reaching new state-of-the-art results for deep selective classification.

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。

↑