Optimization of novelty detection through the use of information entropy
Author
Term
4. term
Education
Publication year
2024
Submitted on
2024-06-19
Pages
10
Abstract
Modern neural networks can achieve high confidence levels of categorization of well-known classes. However, they fall short when trying to categorize classes not part of their training data. This paper proposes a method of optimizing the runtime of a previously proposed outside-the-box novelty detection method, which detects inputs of an unknown nature by monitoring the output of layers in the model on training data and comparing it to the output on any incoming data. The method achieves a reduction in runtime by reducing the amount of dimensions checked in the data through the use of Shannon entropy.
Keywords
Documents