Information-Set Deep Learning Architectures for Improving Classification of Noisy Patterns

Abstract

Deep learning models have been widely used in many supervised learning applications. These models, however, suffer from the problem of overfitting due to various types of uncertainty with deteriorating performance when facing data biases, class imbalance, or noise propagation. To address the critical problem of the absence of robust deep learning models, we propose Information-Set Deep learning (ISDL) architectures with four variants by integrating information set theory and deep learning principle. We describe the ISDL architectures, learning algorithms, and analytic workflows. We evaluate and compare the performance of the ISDL models and standard architectures using a noise-corrupted benchmark dataset. The experimental results show that the ISDL models can efficiently handle noise-dominated uncertainty and outperform peer architectures.