🤖 AI Summary
Existing prototype learning methods suffer from three key limitations: weak prototype representational capacity and unreliable out-of-distribution (OoD) detection; significant performance degradation when projecting prototypes back into image space; and exclusive focus on the most salient regions, neglecting less prominent yet discriminative ones. To address these, we propose MGProto, the first distributed prototype learning framework. It (1) employs Gaussian mixture prototypes jointly optimized for generative fidelity and classification discriminability—eliminating explicit image-space projection; (2) introduces multi-granularity saliency-aware prototype mining to capture both primary and secondary discriminative regions; and (3) incorporates a Bayesian importance prior-guided differentiable prototype pruning mechanism to enhance model compactness and interpretability. On fine-grained benchmarks—including CUB-200 and Stanford Cars—MGProto achieves state-of-the-art accuracy and OoD detection performance while delivering strong visual interpretability.
📝 Abstract
Prototypical-part methods, e.g., ProtoPNet, enhance interpretability in image recognition by linking predictions to training prototypes, thereby offering intuitive insights into their decision-making. Existing methods, which rely on a point-based learning of prototypes, typically face two critical issues: 1) the learned prototypes have limited representation power and are not suitable to detect Out-of-Distribution (OoD) inputs, reducing their decision trustworthiness; and 2) the necessary projection of the learned prototypes back into the space of training images causes a drastic degradation in the predictive performance. Furthermore, current prototype learning adopts an aggressive approach that considers only the most active object parts during training, while overlooking sub-salient object regions which still hold crucial classification information. In this paper, we present a new generative paradigm to learn prototype distributions, termed as Mixture of Gaussian-distributed Prototypes (MGProto). The distribution of prototypes from MGProto enables both interpretable image classification and trustworthy recognition of OoD inputs. The optimisation of MGProto naturally projects the learned prototype distributions back into the training image space, thereby addressing the performance degradation caused by prototype projection. Additionally, we develop a novel and effective prototype mining strategy that considers not only the most active but also sub-salient object parts. To promote model compactness, we further propose to prune MGProto by removing prototypes with low importance priors. Experiments on CUB-200-2011, Stanford Cars, Stanford Dogs, and Oxford-IIIT Pets datasets show that MGProto achieves state-of-the-art image recognition and OoD detection performances, while providing encouraging interpretability results.
