Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
139 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
46 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

PICNN: A Pathway towards Interpretable Convolutional Neural Networks (2312.12068v1)

Published 19 Dec 2023 in cs.CV, cs.AI, and cs.LG

Abstract: Convolutional Neural Networks (CNNs) have exhibited great performance in discriminative feature learning for complex visual tasks. Besides discrimination power, interpretability is another important yet under-explored property for CNNs. One difficulty in the CNN interpretability is that filters and image classes are entangled. In this paper, we introduce a novel pathway to alleviate the entanglement between filters and image classes. The proposed pathway groups the filters in a late conv-layer of CNN into class-specific clusters. Clusters and classes are in a one-to-one relationship. Specifically, we use the Bernoulli sampling to generate the filter-cluster assignment matrix from a learnable filter-class correspondence matrix. To enable end-to-end optimization, we develop a novel reparameterization trick for handling the non-differentiable Bernoulli sampling. We evaluate the effectiveness of our method on ten widely used network architectures (including nine CNNs and a ViT) and five benchmark datasets. Experimental results have demonstrated that our method PICNN (the combination of standard CNNs with our proposed pathway) exhibits greater interpretability than standard CNNs while achieving higher or comparable discrimination power.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (48)
  1. Network dissection: Quantifying interpretability of deep visual representations. In CVPR.
  2. Detect what you can: Detecting and representing objects using holistic models and body parts. In CVPR.
  3. An analysis of single-layer networks in unsupervised feature learning. In AISTATS.
  4. Underspecification presents challenges for credibility in modern machine learning. JMLR.
  5. Imagenet: A large-scale hierarchical image database. In CVPR.
  6. An image is worth 16x16 words: Transformers for image recognition at scale.
  7. Inverting visual representations with convolutional networks. In CVPR.
  8. Net2vec: Quantifying and explaining how concepts are encoded by filters in deep neural networks. In CVPR.
  9. Generative adversarial nets. In NeurIPS.
  10. Seg-XRes-CAM: Explaining Spatially Local Regions in Image Segmentation. In CVPR.
  11. Deep residual learning for image recognition. In CVPR.
  12. Generating visual explanations. In ECCV.
  13. Densely connected convolutional networks. In CVPR.
  14. Categorical Reparameterization with Gumbel-Softmax. In ICLR.
  15. Learning discriminative features via label consistent neural network. In WACV.
  16. Auto-Encoding Variational Bayes. In ICLR.
  17. Krizhevsky, A. 2014. One weird trick for parallelizing convolutional neural networks. CoRR.
  18. Learning multiple layers of features from tiny images.
  19. Unlocking the potential of ordinary classifier: Class-specific adversarial erasing framework for weakly supervised semantic segmentation. In ICCV.
  20. Relevance-cam: Your model already knows where to look. In CVPR.
  21. Collaging class-specific gans for semantic image synthesis. In ICCV.
  22. Training interpretable convolutional neural networks by differentiating class-specific filters. In ECCV.
  23. Understanding deep image representations by inverting them. In CVPR.
  24. Action recognition with spatial-temporal discriminative filter banks. In ICCV.
  25. Pytorch: An imperative style, high-performance deep learning library. In NeurIPS.
  26. Rudin, C. 2019. Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead. Nat. Mac. Intell.
  27. Mobilenetv2: Inverted residuals and linear bottlenecks. In CVPR.
  28. RL-CAM: Visual Explanations for Convolutional Networks Using Reinforcement Learning. In CVPR.
  29. Grad-cam: Visual explanations from deep networks via gradient-based localization. In ICCV.
  30. Interpretable compositional convolutional neural networks. In IJCAI.
  31. Normalized cuts and image segmentation. TPAMI.
  32. Deep inside convolutional networks: Visualising image classification models and saliency maps. In ICLR.
  33. Very deep convolutional networks for large-scale image recognition. In ICLR.
  34. Striving for simplicity: The all convolutional net. In ICLR.
  35. Efficientnet: Rethinking model scaling for convolutional neural networks. In ICML.
  36. Local class-specific and global image-level generative adversarial networks for semantic-guided scene generation. In CVPR.
  37. Interpretable image recognition by constructing transparent embedding space. In ICCV.
  38. Learning a discriminative filter bank within a cnn for fine-grained recognition. In CVPR.
  39. Explaining deep convolutional neural networks via latent visual-semantic filter attention. In CVPR.
  40. Understanding neural networks through deep visualization.
  41. Explainability of deep vision-based autonomous driving systems: Review and challenges. IJCV.
  42. Visualizing and understanding convolutional networks. In ECCV.
  43. Wide residual networks for mitosis detection. In ISBI. IEEE.
  44. Interpreting CNN knowledge via an explanatory graph. In AAAI.
  45. Interpretable convolutional neural networks. In CVPR.
  46. Interpreting cnns via decision trees. In CVPR.
  47. Object Detectors Emerge in Deep Scene CNNs. In ICLR.
  48. Learning deep features for discriminative localization. In CVPR.

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now