SphOR: A Representation Learning Perspective on Open-set Recognition for Identifying Unknown Classes in Deep Learning Models
This addresses the need for efficient identification of unknown classes in deep learning models, which is incremental as it builds on existing representation learning approaches.
The paper tackles the problem of open-set recognition in deep learning by proposing SphOR, a computationally efficient method using spherical embeddings and von Mises-Fisher distributions, which achieves state-of-the-art results with improvements up to 6% on benchmarks.
The widespread use of deep learning classifiers necessitates Open-set recognition (OSR), which enables the identification of input data not only from classes known during training but also from unknown classes that might be present in test data. Many existing OSR methods are computationally expensive due to the reliance on complex generative models or suffer from high training costs. We investigate OSR from a representation-learning perspective, specifically through spherical embeddings. We introduce SphOR, a computationally efficient representation learning method that models the feature space as a mixture of von Mises-Fisher distributions. This approach enables the use of semantically ambiguous samples during training, to improve the detection of samples from unknown classes. We further explore the relationship between OSR performance and key representation learning properties which influence how well features are structured in high-dimensional space. Extensive experiments on multiple OSR benchmarks demonstrate the effectiveness of our method, producing state-of-the-art results, with improvements up-to 6% that validate its performance. Code at https://github.com/nadarasarbahavan/SpHOR