2022-11-28-ji22a.md

title	abstract	video	layout	series	publisher	issn	id	month	tex_title	firstpage	lastpage	page	order	cycles	bibtex_author	author	date	address	container-title	volume	genre	issued	pdf	extras
Test Sample Accuracy Scales with Training Sample Density in Neural Networks	Intuitively, one would expect accuracy of a trained neural network’s prediction on test samples to correlate with how densely the samples are surrounded by seen training samples in representation space. We find that a bound on empirical training error smoothed across linear activation regions scales inversely with training sample density in representation space. Empirically, we verify this bound is a strong predictor of the inaccuracy of the network’s prediction on test samples. For unseen test sets, including those with out-of-distribution samples, ranking test samples by their local region’s error bound and discarding samples with the highest bounds raises prediction accuracy by up to 20% in absolute terms for image classification datasets, on average over thresholds.	https://youtu.be/GF_-LcdudXI	inproceedings	Proceedings of Machine Learning Research	PMLR	2640-3498	ji22a	0	Test Sample Accuracy Scales with Training Sample Density in Neural Networks	629	646	629-646	629	false	Ji, Xu and Pascanu, Razvan and Hjelm, R. Devon and Lakshminarayanan, Balaji and Vedaldi, Andrea	given family Xu Ji given family Razvan Pascanu given family R. Devon Hjelm given family Balaji Lakshminarayanan given family Andrea Vedaldi	2022-11-28		Proceedings of The 1st Conference on Lifelong Learning Agents	199	inproceedings	date-parts 2022 11 28	https://proceedings.mlr.press/v199/ji22a/ji22a.pdf