arXiv 2312.000922025-04-11cs.CV

基于生成建模的高斯分布原型混合模型用于可解释且可信的图像识别

Mixture of Gaussian-distributed Prototypes with Generative Modelling for Interpretable and Trustworthy Image Recognition

Australian Institute for Machine Learning (AIML), University of Adelaide（阿德莱德大学澳大利亚机器学习研究所）；Stanford University（斯坦福大学）；Cornell University（康奈尔大学）；Cornell Tech（康奈尔理工学院）；University of Oxford（牛津大学）；St. Vincent’s Institute of Medical Research（圣文森特医学研究所）；Melbourne Integrative Genomics, The University of Melbourne（墨尔本大学墨尔本整合基因组学中心）；St Vincent’s Hospital Melbourne（墨尔本圣文森特医院）；Centre for Vision, Speech and Signal Processing (CVSSP), University of Surrey（萨里大学视觉、语音与信号处理中心）；The University of Adelaide（阿德莱德大学）

Chong Wang, Yuanhong Chen, Fengbei Liu, Yuyuan Liu, Davis James McCarthy, Helen Frazer, Gustavo Carneiro

备注：: IEEE TPAMI

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

针对现有原型点基学习方法表示能力有限、投影性能下降等问题，提出高斯分布原型混合模型MGProto，通过生成范式学习原型分布并优化投影，结合次显著部件挖掘与原型剪枝，在多数据集上实现SOTA识别及OoD检测性能与可解释性。

摘要：

原型部件方法（例如ProtoPNet）通过将预测与训练原型关联，增强图像识别的可解释性，从而为其决策过程提供直观见解。现有依赖原型点基学习的方法通常面临两个关键问题：1）所学原型的表示能力有限，不适合检测分布外（OoD）输入，降低决策可信度；2）将所学原型投影回训练图像空间的必要步骤导致预测性能急剧下降。此外，当前原型学习采用激进方法，训练时仅考虑最活跃的物体部件，却忽略仍包含关键分类信息的次显著物体区域。本文提出一种新的生成范式来学习原型分布，称为高斯分布原型混合模型（MGProto）。MGProto的原型分布同时支持可解释图像分类和可信的OoD输入识别。MGProto的优化过程自然将所学原型分布投影回训练图像空间，从而解决原型投影导致的性能下降问题。此外，我们开发了一种新颖有效的原型挖掘策略，不仅考虑最活跃部件，还兼顾次显著部件。为提升模型紧凑性，我们进一步提出通过移除低重要性先验的原型来剪枝MGProto。在CUB-200-2011、Stanford Cars、Stanford Dogs和Oxford-IIIT Pets数据集上的实验表明，MGProto实现了最先进的图像识别和OoD检测性能，同时提供了令人鼓舞的可解释性结果。

英文摘要：

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.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。