TY  - JOUR
T1  - Novelty Detection in Underwater Acoustic Environments Using Out-of-Distribution Detector for Neural Networks
AU  - Kim, Nayeon 
AU  - Chun, Chanjun 
AU  - Kim, Hong Kook 
JO  - The Journal of Korean Institute of Communications and Information Sciences
PY  - 2025
DA  - 2025/1/1
DO  - 10.7840/kics.2025.50.12.1839
KW  - Novelty detection
KW  - Out-of-Distribution
KW  - Detector for Neural Networks
KW  - 
KW  - Temperature scaling
KW  - Input
KW  - perturbation
AB  - In this paper, we propose an ODIN-based novelty detection framework to effectively identify unknownacoustic signals in underwater environments. Specifically, temperature scaling and input perturbation are applied to the softmax output of a pre-trained classifier to induce differences between known and unknown samples, and the calibrated maximum softmax probability is used as a novelty score to perform novelty detection.