Supervised and semi-supervised algorithms have been designed under a closed-world setting, with the assumption that unlabeled data consists of classes previously seen in labeled training data. However, real world is inherently open set where this assumption is often violated, and thus novel data may be encountered in test data. In this paper, we look at the problem where the model is required to discover novel classes never encountered in the labeled set. We propose a dependency measure based on Squared Mutual Information (SMI) where we simultaneously learn to classify and cluster the data. Our experiments show that our approach is able to achieve competitive performance on CIFAR and Imagenet datasets.
Mukherjee, T & Deligiannis, N 2022, NOVEL CLASS DISCOVERY: A DEPENDENCY APPROACH. in IEEE International Conference on Acoustics, Speech and Signal Processing. International Conference on Acoustics, Speech and Signal Processing, IEEE Signal Processing Society, pp. 2525-2528, 2022 International Conference on Acoustics, Speech, and Signal Processing, Singapore, 22/05/22. https://doi.org/10.1109/ICASSP43922.2022.9747827, https://doi.org/10.1109/ICASSP43922.2022.9747827
Mukherjee, T., & Deligiannis, N. (2022). NOVEL CLASS DISCOVERY: A DEPENDENCY APPROACH. In IEEE International Conference on Acoustics, Speech and Signal Processing (pp. 2525-2528). (International Conference on Acoustics, Speech and Signal Processing). IEEE Signal Processing Society. https://doi.org/10.1109/ICASSP43922.2022.9747827, https://doi.org/10.1109/ICASSP43922.2022.9747827
@inproceedings{27b0f0631d854cac88f4aa4a83653968,
title = "NOVEL CLASS DISCOVERY: A DEPENDENCY APPROACH",
abstract = "Supervised and semi-supervised algorithms have been designed under a closed-world setting, with the assumption that unlabeled data consists of classes previously seen in labeled training data. However, real world is inherently open set where this assumption is often violated, and thus novel data may be encountered in test data. In this paper, we look at the problem where the model is required to discover novel classes never encountered in the labeled set. We propose a dependency measure based on Squared Mutual Information (SMI) where we simultaneously learn to classify and cluster the data. Our experiments show that our approach is able to achieve competitive performance on CIFAR and Imagenet datasets.",
author = "Tanmoy Mukherjee and Nikos Deligiannis",
year = "2022",
doi = "10.1109/ICASSP43922.2022.9747827",
language = "English",
isbn = "9781665405416",
series = "International Conference on Acoustics, Speech and Signal Processing",
publisher = "IEEE Signal Processing Society",
pages = "2525--2528",
booktitle = "IEEE International Conference on Acoustics, Speech and Signal Processing",
note = "2022 International Conference on Acoustics, Speech, and Signal Processing ; Conference date: 22-05-2022 Through 27-05-2022",
}