# Training Interpretable Convolutional Neural Networks by Differentiating Class-specific Filters

arXiv: https://arxiv.org/abs/2007.08194 · HTML: https://mlsanity.com/paper/2007.08194/

## Key facts

- Cited by 26 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 16 Jul 2020.

## Built on by (most cited first, 26 of 26)

- [Deformable ProtoPNet: An Interpretable Image Classifier Using Deformable Prototypes](https://mlsanity.com/paper/2111.15000/) (2021)
- [Learning Support and Trivial Prototypes for Interpretable Image Classification](https://mlsanity.com/paper/2301.04011/) (2023)
- [A Comprehensive Survey on Self-Interpretable Neural Networks](https://mlsanity.com/paper/2501.15638/) (2025)
- [Interpretable Compositional Convolutional Neural Networks](https://mlsanity.com/paper/2107.04474/) (2021)
- [Towards Interpretable Deep Networks for Monocular Depth Estimation](https://mlsanity.com/paper/2108.05312/) (2021)
- [Explaining Deep Convolutional Neural Networks via Latent Visual-Semantic Filter Attention](https://mlsanity.com/paper/2204.04601/) (2022)
- [ME-D2N: Multi-Expert Domain Decompositional Network for Cross-Domain Few-Shot Learning](https://mlsanity.com/paper/2210.05280/) (2022)
- [Take 5: Interpretable Image Classification with a Handful of Features](https://mlsanity.com/paper/2303.13166/) (2023)
- [Bort: Towards Explainable Neural Networks with Bounded Orthogonal Constraint](https://mlsanity.com/paper/2212.09062/) (2022)
- [Unsupervised Interpretable Basis Extraction for Concept-Based Visual Explanations](https://mlsanity.com/paper/2303.10523/) (2023)
- [Learning Transferable Conceptual Prototypes for Interpretable Unsupervised Domain Adaptation](https://mlsanity.com/paper/2310.08071/) (2023)
- [Path Choice Matters for Clear Attribution in Path Methods](https://mlsanity.com/paper/2401.10442/) (2024)
- [Disentangled Latent Transformer for Interpretable Monocular Height Estimation](https://mlsanity.com/paper/2201.06357/) (2022)
- [E Pluribus Unum Interpretable Convolutional Neural Networks](https://mlsanity.com/paper/2208.05369/) (2022)
- [PICNN: A Pathway towards Interpretable Convolutional Neural Networks](https://mlsanity.com/paper/2312.12068/) (2023)
- [Towards Fully Interpretable Deep Neural Networks: Are We There Yet?](https://mlsanity.com/paper/2106.13164/) (2021)
- [Model LEGO: Creating Models Like Disassembling and Assembling Building Blocks](https://mlsanity.com/paper/2203.13453/) (2022)
- [COMIX: Compositional Explanations using Prototypes](https://mlsanity.com/paper/2501.06059/) (2025)
- [Disentangling Visual Transformers: Patch-level Interpretability for Image Classification](https://mlsanity.com/paper/2502.17196/) (2025)
- [End-to-end Feature Alignment: A Simple CNN with Intrinsic Class Attribution](https://mlsanity.com/paper/2603.25798/) (2026)
- [Jellyfish: Zero-Shot Federated Unlearning Scheme with Knowledge Disentanglement](https://mlsanity.com/paper/2604.04030/) (2026)
- [Decomposing the Deep: Finding Class Specific Filters in Deep CNNs](https://mlsanity.com/paper/2112.07719/) (2021)
- [TCNL: Transparent and Controllable Network Learning Via Embedding Human-Guided Concepts](https://mlsanity.com/paper/2210.03274/) (2022)
- [Propheter: Prophetic Teacher Guided Long-Tailed Distribution Learning](https://mlsanity.com/paper/2304.04135/) (2023)
- [Understanding the Role of Pathways in a Deep Neural Network](https://mlsanity.com/paper/2402.18132/) (2024)
- [Bridge the Gap Between Visual and Linguistic Comprehension for Generalized Zero-shot Semantic Segmentation](https://mlsanity.com/paper/2503.23806/) (2025)

## Abstract (arXiv)

Convolutional neural networks (CNNs) have been successfully used in a range of tasks. However, CNNs are often viewed as "black-box" and lack of interpretability. One main reason is due to the filter-class entanglement -- an intricate many-to-many correspondence between filters and classes. Most existing works attempt post-hoc interpretation on a pre-trained model, while neglecting to reduce the entanglement underlying the model. In contrast, we focus on alleviating filter-class entanglement during training. Inspired by cellular differentiation, we propose a novel strategy to train interpretable CNNs by encouraging class-specific filters, among which each filter responds to only one (or few) class. Concretely, we design a learnable sparse Class-Specific Gate (CSG) structure to assign each filter with one (or few) class in a flexible way. The gate allows a filter's activation to pass only when the input samples come from the specific class. Extensive experiments demonstrate the fabulous performance of our method in generating a sparse and highly class-related representation of the input, which leads to stronger interpretability. Moreover, comparing with the standard training strategy, our model displays benefits in applications like object localization and adversarial sample detection. Code link: https://github.com/hyliang96/CSGCNN.