What is a Tabby? Interpretable Model Decisions by Learning Attribute-Based Classification Criteria
| Autor: | Ruiping Wang, Xilin Chen, Haomiao Liu, Shiguang Shan |
|---|---|
| Rok vydání: | 2021 |
| Předmět: |
Contextual image classification
Artificial neural network Hierarchy (mathematics) business.industry Computer science Applied Mathematics 02 engineering and technology Machine learning computer.software_genre Convolutional neural network Visualization Set (abstract data type) Computational Theory and Mathematics Discriminative model Artificial Intelligence Scalability 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Computer Vision and Pattern Recognition Artificial intelligence business computer Software |
| Zdroj: | IEEE Transactions on Pattern Analysis and Machine Intelligence. 43:1791-1807 |
| ISSN: | 1939-3539 0162-8828 |
| Popis: | State-of-the-art classification models are usually considered as black boxes since their decision processes are implicit to humans. On the contrary, human experts classify objects according to a set of explicit hierarchical criteria. For example, “ tabby is a domestic cat with stripes, dots, or lines”, where tabby is defined by combining its superordinate category ( domestic cat ) and some certain attributes (e.g., has stripes). Inspired by this mechanism, we propose an interpretable Hierarchical Criteria Network (HCN) by additionally learning such criteria. To achieve this goal, images and semantic entities (e.g., taxonomies and attributes) are embedded into a common space, where each category can be represented by the linear combination of its superordinate category and a set of learned discriminative attributes. Specifically, a two-stream convolutional neural network (CNN) is elaborately devised, which embeds images and taxonomies with the two streams respectively. The model is trained by minimizing the prediction error of hierarchy labels on both streams. Extensive experiments on two widely studied datasets (CIFAR-100 and ILSVRC) demonstrate that HCN can learn meaningful attributes as well as reasonable and interpretable classification criteria. Therefore, the proposed method enables further human feedback for model correction as an additional benefit. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|