Leakage-Robust Classifier via Mask-Enhanced Training (Student Abstract)

Authors

  • Damian Stachura Jagiellonian University
  • Christopher Galias Jagiellonian University
  • Konrad Żołna Jagiellonian University

DOI:

https://doi.org/10.1609/aaai.v34i10.7234

Abstract

We synthetically add data leakage to well-known image datasets, which results in predictions of convolutional neural networks trained naively on these spoiled datasets becoming wildly inaccurate. We propose a method, dubbed Mask-Enhanced Training, that automatically identifies the possible leakage and makes the classifier robust. The method enables the model to focus on all features needed to solve the task, making its predictions on the original validation set accurate, even if the whole training dataset is spoiled with the leakage.

Downloads

Published

2020-04-03

How to Cite

Stachura, D., Galias, C., & Żołna, K. (2020). Leakage-Robust Classifier via Mask-Enhanced Training (Student Abstract). Proceedings of the AAAI Conference on Artificial Intelligence, 34(10), 13923-13924. https://doi.org/10.1609/aaai.v34i10.7234

Issue

Section

Student Abstract Track