Large Class Separation is not what you need for Relational Reasoning-based OOD Detection
This addresses the need for reliable OOD detection in safety-critical applications like healthcare or autonomous driving by offering a computationally efficient, fine-tuning-free method, though it is incremental as it builds on existing relational reasoning approaches.
The paper tackles the problem of out-of-distribution (OOD) detection without fine-tuning by analyzing relational reasoning pre-training, finding that large inter-class feature distance correlates with accuracy, and proposes a new loss function to control this margin, showing improved performance in experiments.
Standard recognition approaches are unable to deal with novel categories at test time. Their overconfidence on the known classes makes the predictions unreliable for safety-critical applications such as healthcare or autonomous driving. Out-Of-Distribution (OOD) detection methods provide a solution by identifying semantic novelty. Most of these methods leverage a learning stage on the known data, which means training (or fine-tuning) a model to capture the concept of normality. This process is clearly sensitive to the amount of available samples and might be computationally expensive for on-board systems. A viable alternative is that of evaluating similarities in the embedding space produced by large pre-trained models without any further learning effort. We focus exactly on such a fine-tuning-free OOD detection setting. This works presents an in-depth analysis of the recently introduced relational reasoning pre-training and investigates the properties of the learned embedding, highlighting the existence of a correlation between the inter-class feature distance and the OOD detection accuracy. As the class separation depends on the chosen pre-training objective, we propose an alternative loss function to control the inter-class margin, and we show its advantage with thorough experiments.