SelectiveNet: A Deep Neural Network with an Integrated Reject Option

Yonatan Geifman; Ran El-Yaniv

SelectiveNet: A Deep Neural Network with an Integrated Reject Option

Yonatan Geifman, Ran El-Yaniv ;

Proceedings of the 36th International Conference on Machine Learning, PMLR 97:2151-2159, 2019.

Abstract

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.

Related Material

@InProceedings{pmlr-v97-geifman19a, title = {{S}elective{N}et: A Deep Neural Network with an Integrated Reject Option}, author = {Geifman, Yonatan and El-Yaniv, Ran}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {2151--2159}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, address = {Long Beach, California, USA}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/geifman19a/geifman19a.pdf}, url = {http://proceedings.mlr.press/v97/geifman19a.html}, abstract = {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.} }

%0 Conference Paper %T SelectiveNet: A Deep Neural Network with an Integrated Reject Option %A Yonatan Geifman %A Ran El-Yaniv %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-geifman19a %I PMLR %J Proceedings of Machine Learning Research %P 2151--2159 %U http://proceedings.mlr.press %V 97 %W PMLR %X 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.