Robust Deep Semi-supervised Clustering with Cauchy Mixture
Loading...
Links to Files
Permanent Link
Author/Creator
Author/Creator ORCID
Date
2022-01-01
Type of Work
Department
Computer Science and Electrical Engineering
Program
Computer Science
Citation of Original Publication
Rights
Distribution Rights granted to UMBC by the author.
This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see http://aok.lib.umbc.edu/specoll/repro.php or contact Special Collections at speccoll(at)umbc.edu
This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see http://aok.lib.umbc.edu/specoll/repro.php or contact Special Collections at speccoll(at)umbc.edu
Abstract
In this paper, we present using Cauchy Mixture Model (CMM) as the activation function of deep neural networks for pseudo-labeling in semi-supervised classification. Unlike SoftMax, which is universally used as the final layer activation function, CMM allows the model to identify outliers, i.e., any data that is out of distribution from the observed data. Furthermore, using CMM for pseudo-labeling provides enhanced robustness against confounding bias by preventing the model from yielding high confidence predictions for out-of-distribution data. The proposed method is trained and tested on the CIFAR-10 dataset using only 250, 1000, and 4000 labels. Compared to the baseline of SoftMax-based only supervised and semi-supervised models, the proposed method shows substantial improvements in the same environment setting.