{"id":15716349,"url":"https://github.com/onetaken/awesome_deep_learning_interpretability","last_synced_at":"2025-12-30T23:16:47.146Z","repository":{"id":38202029,"uuid":"231166651","full_name":"oneTaken/awesome_deep_learning_interpretability","owner":"oneTaken","description":"深度学习近年来关于神经网络模型解释性的相关高引用/顶会论文(附带代码)","archived":false,"fork":false,"pushed_at":"2024-04-08T09:36:12.000Z","size":160,"stargazers_count":757,"open_issues_count":2,"forks_count":125,"subscribers_count":29,"default_branch":"master","last_synced_at":"2025-09-30T10:02:15.405Z","etag":null,"topics":["awesome","awesome-list","chainer","computer-vision","cvpr","deep-learning","eccv","iccv","iclr","icml","interpretability","keras","matlab","neural-network","neurips","nlp","papers","pytorch","tensorflow","torch"],"latest_commit_sha":null,"homepage":"","language":null,"has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/oneTaken.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null}},"created_at":"2020-01-01T02:17:20.000Z","updated_at":"2025-09-28T12:57:48.000Z","dependencies_parsed_at":"2023-11-19T15:27:56.254Z","dependency_job_id":"ffe0f60e-67e1-4dbb-a4d2-0740c7d6611e","html_url":"https://github.com/oneTaken/awesome_deep_learning_interpretability","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/oneTaken/awesome_deep_learning_interpretability","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/oneTaken%2Fawesome_deep_learning_interpretability","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/oneTaken%2Fawesome_deep_learning_interpretability/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/oneTaken%2Fawesome_deep_learning_interpretability/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/oneTaken%2Fawesome_deep_learning_interpretability/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/oneTaken","download_url":"https://codeload.github.com/oneTaken/awesome_deep_learning_interpretability/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/oneTaken%2Fawesome_deep_learning_interpretability/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":279020650,"owners_count":26086898,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-10-14T02:00:06.444Z","response_time":60,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["awesome","awesome-list","chainer","computer-vision","cvpr","deep-learning","eccv","iccv","iclr","icml","interpretability","keras","matlab","neural-network","neurips","nlp","papers","pytorch","tensorflow","torch"],"created_at":"2024-10-03T21:45:13.158Z","updated_at":"2025-12-30T23:16:47.118Z","avatar_url":"https://github.com/oneTaken.png","language":null,"funding_links":[],"categories":[],"sub_categories":[],"readme":"\n\n\n\n# awesome_deep_learning_interpretability\n深度学习近年来关于模型解释性的相关论文。\n\n按引用次数排序可见[引用排序](./sort_cite.md)\n\n159篇论文pdf(有2篇需要上scihub找)上传到[腾讯微云](https://share.weiyun.com/5ddB0EQ)。\n\n不定期更新。\n\n|Year|Publication|Paper|Citation|code|\n|:---:|:---:|:---:|:---:|:---:|\n|2020|CVPR|[Explaining Knowledge Distillation by Quantifying the Knowledge](https://arxiv.org/pdf/2003.03622.pdf)|81|\n|2020|CVPR|[High-frequency Component Helps Explain the Generalization of Convolutional Neural Networks](https://openaccess.thecvf.com/content_CVPR_2020/papers/Wang_High-Frequency_Component_Helps_Explain_the_Generalization_of_Convolutional_Neural_Networks_CVPR_2020_paper.pdf)|289|\n|2020|CVPRW|[Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networks](https://openaccess.thecvf.com/content_CVPRW_2020/papers/w1/Wang_Score-CAM_Score-Weighted_Visual_Explanations_for_Convolutional_Neural_Networks_CVPRW_2020_paper.pdf)|414|[Pytorch](https://github.com/haofanwang/Score-CAM)\n|2020|ICLR|[Knowledge consistency between neural networks and beyond](https://arxiv.org/pdf/1908.01581.pdf)|28|\n|2020|ICLR|[Interpretable Complex-Valued Neural Networks for Privacy Protection](https://arxiv.org/pdf/1901.09546.pdf)|23|\n|2019|AI|[Explanation in artificial intelligence: Insights from the social sciences](https://arxiv.org/pdf/1706.07269.pdf)|3248|\n|2019|NMI|[Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead](https://arxiv.org/pdf/1811.10154.pdf)|3505|\n|2019|NeurIPS|[Can you trust your model's uncertainty? Evaluating predictive uncertainty under dataset shift](https://papers.nips.cc/paper/9547-can-you-trust-your-models-uncertainty-evaluating-predictive-uncertainty-under-dataset-shift.pdf)|1052|-|\n|2019|NeurIPS|[This looks like that: deep learning for interpretable image recognition](http://papers.nips.cc/paper/9095-this-looks-like-that-deep-learning-for-interpretable-image-recognition.pdf)|665|[Pytorch](https://github.com/cfchen-duke/ProtoPNet)|\n|2019|NeurIPS|[A benchmark for interpretability methods in deep neural networks](https://papers.nips.cc/paper/9167-a-benchmark-for-interpretability-methods-in-deep-neural-networks.pdf)|413|\n|2019|NeurIPS|[Full-gradient representation for neural network visualization](http://papers.nips.cc/paper/8666-full-gradient-representation-for-neural-network-visualization.pdf)|155|\n|2019|NeurIPS|[On the (In) fidelity and Sensitivity of Explanations](https://papers.nips.cc/paper/9278-on-the-infidelity-and-sensitivity-of-explanations.pdf)|226|\n|2019|NeurIPS|[Towards Automatic Concept-based Explanations](http://papers.nips.cc/paper/9126-towards-automatic-concept-based-explanations.pdf)|342|[Tensorflow](https://github.com/amiratag/ACE)|\n|2019|NeurIPS|[CXPlain: Causal explanations for model interpretation under uncertainty](http://papers.nips.cc/paper/9211-cxplain-causal-explanations-for-model-interpretation-under-uncertainty.pdf)|133|\n|2019|CVPR|[Interpreting CNNs via Decision Trees](http://openaccess.thecvf.com/content_CVPR_2019/papers/Zhang_Interpreting_CNNs_via_Decision_Trees_CVPR_2019_paper.pdf)|293|\n|2019|CVPR|[From Recognition to Cognition: Visual Commonsense Reasoning](http://openaccess.thecvf.com/content_CVPR_2019/papers/Zellers_From_Recognition_to_Cognition_Visual_Commonsense_Reasoning_CVPR_2019_paper.pdf)|544|[Pytorch](https://github.com/rowanz/r2c)|\n|2019|CVPR|[Attention branch network: Learning of attention mechanism for visual explanation](http://openaccess.thecvf.com/content_CVPR_2019/papers/Fukui_Attention_Branch_Network_Learning_of_Attention_Mechanism_for_Visual_Explanation_CVPR_2019_paper.pdf)|371|\n|2019|CVPR|[Interpretable and fine-grained visual explanations for convolutional neural networks](http://openaccess.thecvf.com/content_CVPR_2019/papers/Wagner_Interpretable_and_Fine-Grained_Visual_Explanations_for_Convolutional_Neural_Networks_CVPR_2019_paper.pdf)|116|\n|2019|CVPR|[Learning to Explain with Complemental Examples](http://openaccess.thecvf.com/content_CVPR_2019/papers/Kanehira_Learning_to_Explain_With_Complemental_Examples_CVPR_2019_paper.pdf)|36|\n|2019|CVPR|[Revealing Scenes by Inverting Structure from Motion Reconstructions](http://openaccess.thecvf.com/content_CVPR_2019/papers/Pittaluga_Revealing_Scenes_by_Inverting_Structure_From_Motion_Reconstructions_CVPR_2019_paper.pdf)|84|[Tensorflow](https://github.com/francescopittaluga/invsfm)|\n|2019|CVPR|[Multimodal Explanations by Predicting Counterfactuality in Videos](http://openaccess.thecvf.com/content_CVPR_2019/papers/Kanehira_Multimodal_Explanations_by_Predicting_Counterfactuality_in_Videos_CVPR_2019_paper.pdf)|26|\n|2019|CVPR|[Visualizing the Resilience of Deep Convolutional Network Interpretations](http://openaccess.thecvf.com/content_CVPRW_2019/papers/Explainable%20AI/Vasu_Visualizing_the_Resilience_of_Deep_Convolutional_Network_Interpretations_CVPRW_2019_paper.pdf)|2|\n|2019|ICCV|[U-CAM: Visual Explanation using Uncertainty based Class Activation Maps](http://openaccess.thecvf.com/content_ICCV_2019/papers/Patro_U-CAM_Visual_Explanation_Using_Uncertainty_Based_Class_Activation_Maps_ICCV_2019_paper.pdf)|61|\n|2019|ICCV|[Towards Interpretable Face Recognition](https://arxiv.org/pdf/1805.00611.pdf)|66|\n|2019|ICCV|[Taking a HINT: Leveraging Explanations to Make Vision and Language Models More Grounded](http://openaccess.thecvf.com/content_ICCV_2019/papers/Selvaraju_Taking_a_HINT_Leveraging_Explanations_to_Make_Vision_and_Language_ICCV_2019_paper.pdf)|163|\n|2019|ICCV|[Understanding Deep Networks via Extremal Perturbations and Smooth Masks](http://openaccess.thecvf.com/content_ICCV_2019/papers/Fong_Understanding_Deep_Networks_via_Extremal_Perturbations_and_Smooth_Masks_ICCV_2019_paper.pdf)|276|[Pytorch](https://github.com/facebookresearch/TorchRay)|\n|2019|ICCV|[Explaining Neural Networks Semantically and Quantitatively](http://openaccess.thecvf.com/content_ICCV_2019/papers/Chen_Explaining_Neural_Networks_Semantically_and_Quantitatively_ICCV_2019_paper.pdf)|49|\n|2019|ICLR|[Hierarchical interpretations for neural network predictions](https://arxiv.org/pdf/1806.05337.pdf)|111|[Pytorch](https://github.com/csinva/hierarchical-dnn-interpretations)|\n|2019|ICLR|[How Important Is a Neuron?](https://arxiv.org/pdf/1805.12233.pdf)|101|\n|2019|ICLR|[Visual Explanation by Interpretation: Improving Visual Feedback Capabilities of Deep Neural Networks](https://arxiv.org/pdf/1712.06302.pdf)|56|\n|2018|ICML|[Extracting Automata from Recurrent Neural Networks Using Queries and Counterexamples](https://arxiv.org/pdf/1711.09576.pdf)|169|[Pytorch](https://github.com/tech-srl/lstar_extraction)|\n|2019|ICML|[Towards A Deep and Unified Understanding of Deep Neural Models in NLP](http://proceedings.mlr.press/v97/guan19a/guan19a.pdf)|80|[Pytorch](https://github.com/icml2019paper2428/Towards-A-Deep-and-Unified-Understanding-of-Deep-Neural-Models-in-NLP)|\n|2019|ICAIS|[Interpreting black box predictions using fisher kernels](https://arxiv.org/pdf/1810.10118.pdf)|80|\n|2019|ACMFAT|[Explaining explanations in AI](https://s3.amazonaws.com/academia.edu.documents/57692790/Mittelstadt__Russell_and_Wachter_-_2019_-_Explaining_Explanations_in_AI.pdf?response-content-disposition=inline%3B%20filename%3DExplaining_Explanations_in_AI.pdf\u0026X-Amz-Algorithm=AWS4-HMAC-SHA256\u0026X-Amz-Credential=ASIATUSBJ6BAJW2TMFXG%2F20200528%2Fus-east-1%2Fs3%2Faws4_request\u0026X-Amz-Date=20200528T052420Z\u0026X-Amz-Expires=3600\u0026X-Amz-SignedHeaders=host\u0026X-Amz-Security-Token=IQoJb3JpZ2luX2VjEHUaCXVzLWVhc3QtMSJIMEYCIQDCCKV%2FpUmJZHn03yzTquQ%2FNMtaXW%2FC63WPmQd%2FhImmYAIhAMelsFwqb9IfV4W2xlfL%2FHk4qeovouLdYbXKf%2B1%2FMwvyKr0DCM7%2F%2F%2F%2F%2F%2F%2F%2F%2F%2FwEQABoMMjUwMzE4ODExMjAwIgytA%2BM6OWOGN4XLrlUqkQN2f8ywZT0AEUzKdbVDyGvZN%2B1repdgXrfgT2rAJiGacTK8IRCoyECvRgcgS%2BWJWYpjS7CjoL%2BlTm1c%2BWDWdo%2FYnVM0U6shk9OQivK089W064ZR64AQCCkBDutI3vYhP%2BOJ8AtEUDE%2B7W5EWVQ4zeUDG4ryxzdomFnrHpzA5fp05qWrOmPS0vd%2FFabC%2FPKXO34bpfgyRzz3PHrIsUC2%2BPB0EAo7CPKS0Ux%2FlxmiIOYOIj5u1ZKoP8NVLgOfueQe7%2F%2F3VJUnUXSAIsAThszDTnbi0AJEjvNvUHjm8E%2F7zqBApJ6YVd39NkKl8%2BTE7MRwKuITAOIq8jsyta%2FcmIY5igpHpVCkYcG395rHfScDu3CODXIAcKRLX%2F7brNz%2FRHuGhddK3Q2XuGTjQaeLTEYTmTj2e7VDDmEOt%2BpxvXx7UaImPakzpVZ1Ks6APy1JHupKgBhM6JJkeFprlK62e4sf09wqwxk9KsJSot3TMLVwM63yGr7VmXdg61ETsg0D%2BO1DOnnMprsFhEkb%2Bt%2FpCVafebolsjCN%2Frz2BTrqAZiqy6Obte6J%2BeHJ5bzB1sy1oF%2Fi7ueF56nd1C9ObB%2FXLx930j8wqmakO%2FnoaUiYM6gHh1jZbl8cCeLr8Xu0YSGecpe1J5HECU0A5%2Fq68zoBDfyY6UGNZJ%2B87Br6crqpfaHFkP5g4zXvuN2%2F0fp6S9m2iuSRBr%2B%2Bh2Z1rXmvb3Vequ2qgqeJBS2nHOX8pLp2LhJsVMqdl218jeQDsjYnbxJKq86peVGr66Cuv7TmNiimVl0c0dPr1jgjr25N9hvMnpX83n2Xa%2Fz%2BHUmaYfwFLrD0YLkUWaS2Khcpm0%2BwvrcYsQEyOmYkVG8x5Q%3D%3D\u0026X-Amz-Signature=4fcca52f4ae92746068ea2164846aca05c2bb44e04c1330947ba70f75e676171)|558|\n|2019|AAAI|[Interpretation of neural networks is fragile](https://machine-learning-and-security.github.io/papers/mlsec17_paper_18.pdf)|597|[Tensorflow](https://github.com/amiratag/InterpretationFragility)|\n|2019|AAAI|[Classifier-agnostic saliency map extraction](https://arxiv.org/pdf/1805.08249.pdf)|23|\n|2019|AAAI|[Can You Explain That? Lucid Explanations Help Human-AI Collaborative Image Retrieval](https://arxiv.org/pdf/1904.03285.pdf)|11|\n|2019|AAAIW|[Unsupervised Learning of Neural Networks to Explain Neural Networks](https://arxiv.org/pdf/1805.07468.pdf)|28|\n|2019|AAAIW|[Network Transplanting](https://arxiv.org/pdf/1804.10272.pdf)|4|\n|2019|CSUR|[A Survey of Methods for Explaining Black Box Models](https://kdd.isti.cnr.it/sites/kdd.isti.cnr.it/files/csur2018survey.pdf)|3088|\n|2019|JVCIR|[Interpretable convolutional neural networks via feedforward design](https://arxiv.org/pdf/1810.02786)|134|[Keras](https://github.com/davidsonic/Interpretable_CNNs_via_Feedforward_Design)|\n|2019|ExplainAI|[The (Un)reliability of saliency methods](https://arxiv.org/pdf/1711.00867.pdf)|515|\n|2019|ACL|[Attention is not Explanation](https://arxiv.org/pdf/1902.10186.pdf)|920|\n|2019|EMNLP|[Attention is not not Explanation](https://arxiv.org/pdf/1908.04626.pdf)|667|\n|2019|arxiv|[Attention Interpretability Across NLP Tasks](https://arxiv.org/pdf/1909.11218.pdf)|129|\n|2019|arxiv|[Interpretable CNNs](https://arxiv.org/pdf/1901.02413.pdf)|2|\n|2018|ICLR|[Towards better understanding of gradient-based attribution methods for deep neural networks](https://arxiv.org/pdf/1711.06104.pdf)|775|\n|2018|ICLR|[Learning how to explain neural networks: PatternNet and PatternAttribution](https://arxiv.org/pdf/1705.05598.pdf)|342|\n|2018|ICLR|[On the importance of single directions for generalization](https://arxiv.org/pdf/1803.06959.pdf)|282|[Pytorch](https://github.com/1Konny/class_selectivity_index)|\n|2018|ICLR|[Detecting statistical interactions from neural network weights](https://arxiv.org/pdf/1705.04977.pdf)|148|[Pytorch](https://github.com/mtsang/neural-interaction-detection)|\n|2018|ICLR|[Interpretable counting for visual question answering](https://arxiv.org/pdf/1712.08697.pdf)|55|[Pytorch](https://github.com/sanyam5/irlc-vqa-counting)|\n|2018|CVPR|[Interpretable Convolutional Neural Networks](http://openaccess.thecvf.com/content_cvpr_2018/papers/Zhang_Interpretable_Convolutional_Neural_CVPR_2018_paper.pdf)|677|\n|2018|CVPR|[Tell me where to look: Guided attention inference network](http://openaccess.thecvf.com/content_cvpr_2018/papers/Li_Tell_Me_Where_CVPR_2018_paper.pdf)|454|[Chainer](https://github.com/alokwhitewolf/Guided-Attention-Inference-Network)|\n|2018|CVPR|[Multimodal Explanations: Justifying Decisions and Pointing to the Evidence](http://openaccess.thecvf.com/content_cvpr_2018/papers/Park_Multimodal_Explanations_Justifying_CVPR_2018_paper.pdf)|349|[Caffe](https://github.com/Seth-Park/MultimodalExplanations)|\n|2018|CVPR|[Transparency by design: Closing the gap between performance and interpretability in visual reasoning](http://openaccess.thecvf.com/content_cvpr_2018/papers/Mascharka_Transparency_by_Design_CVPR_2018_paper.pdf)|180|[Pytorch](https://github.com/davidmascharka/tbd-nets)|\n|2018|CVPR|[Net2vec: Quantifying and explaining how concepts are encoded by filters in deep neural networks](http://openaccess.thecvf.com/content_cvpr_2018/papers/Fong_Net2Vec_Quantifying_and_CVPR_2018_paper.pdf)|186|\n|2018|CVPR|[What have we learned from deep representations for action recognition?](http://openaccess.thecvf.com/content_cvpr_2018/papers/Feichtenhofer_What_Have_We_CVPR_2018_paper.pdf)|52|\n|2018|CVPR|[Learning to Act Properly: Predicting and Explaining Affordances from Images](http://openaccess.thecvf.com/content_cvpr_2018/papers/Chuang_Learning_to_Act_CVPR_2018_paper.pdf)|57|\n|2018|CVPR|[Teaching Categories to Human Learners with Visual Explanations](http://openaccess.thecvf.com/content_cvpr_2018/papers/Aodha_Teaching_Categories_to_CVPR_2018_paper.pdf)|64|[Pytorch](https://github.com/macaodha/explain_teach)|\n|2018|CVPR|[What do deep networks like to see?](http://openaccess.thecvf.com/content_cvpr_2018/papers/Palacio_What_Do_Deep_CVPR_2018_paper.pdf)|36|\n|2018|CVPR|[Interpret Neural Networks by Identifying Critical Data Routing Paths](http://openaccess.thecvf.com/content_cvpr_2018/papers/Wang_Interpret_Neural_Networks_CVPR_2018_paper.pdf)|73|[Tensorflow](https://github.com/lidongyue12138/CriticalPathPruning)|\n|2018|ECCV|[Deep clustering for unsupervised learning of visual features](http://openaccess.thecvf.com/content_ECCV_2018/papers/Mathilde_Caron_Deep_Clustering_for_ECCV_2018_paper.pdf)|2056|[Pytorch](https://github.com/asanakoy/deep_clustering)|\n|2018|ECCV|[Explainable neural computation via stack neural module networks](http://openaccess.thecvf.com/content_ECCV_2018/papers/Ronghang_Hu_Explainable_Neural_Computation_ECCV_2018_paper.pdf)|164|[Tensorflow](https://github.com/ronghanghu/snmn)|\n|2018|ECCV|[Grounding visual explanations](http://openaccess.thecvf.com/content_ECCV_2018/papers/Lisa_Anne_Hendricks_Grounding_Visual_Explanations_ECCV_2018_paper.pdf)|184|\n|2018|ECCV|[Textual explanations for self-driving vehicles](http://openaccess.thecvf.com/content_ECCV_2018/papers/Jinkyu_Kim_Textual_Explanations_for_ECCV_2018_paper.pdf)|196|\n|2018|ECCV|[Interpretable basis decomposition for visual explanation](http://openaccess.thecvf.com/content_ECCV_2018/papers/Antonio_Torralba_Interpretable_Basis_Decomposition_ECCV_2018_paper.pdf)|228|[Pytorch](https://github.com/CSAILVision/IBD)|\n|2018|ECCV|[Convnets and imagenet beyond accuracy: Understanding mistakes and uncovering biases](http://openaccess.thecvf.com/content_ECCV_2018/papers/Pierre_Stock_ConvNets_and_ImageNet_ECCV_2018_paper.pdf)|147|\n|2018|ECCV|[Vqa-e: Explaining, elaborating, and enhancing your answers for visual questions](http://openaccess.thecvf.com/content_ECCV_2018/papers/Qing_Li_VQA-E_Explaining_Elaborating_ECCV_2018_paper.pdf)|71|\n|2018|ECCV|[Choose Your Neuron: Incorporating Domain Knowledge through Neuron-Importance](http://openaccess.thecvf.com/content_ECCV_2018/papers/Ramprasaath_Ramasamy_Selvaraju_Choose_Your_Neuron_ECCV_2018_paper.pdf)|41|[Pytorch](https://github.com/ramprs/neuron-importance-zsl)|\n|2018|ECCV|[Diverse feature visualizations reveal invariances in early layers of deep neural networks](http://openaccess.thecvf.com/content_ECCV_2018/papers/Santiago_Cadena_Diverse_feature_visualizations_ECCV_2018_paper.pdf)|23|[Tensorflow](https://github.com/sacadena/diverse_feature_vis)|\n|2018|ECCV|[ExplainGAN: Model Explanation via Decision Boundary Crossing Transformations](http://openaccess.thecvf.com/content_ECCV_2018/papers/Nathan_Silberman_ExplainGAN_Model_Explanation_ECCV_2018_paper.pdf)|36|\n|2018|ICML|[Interpretability beyond feature attribution: Quantitative testing with concept activation vectors](https://arxiv.org/pdf/1711.11279.pdf)|1130|[Tensorflow](https://github.com/fursovia/tcav_nlp)|\n|2018|ICML|[Learning to explain: An information-theoretic perspective on model interpretation](https://arxiv.org/pdf/1802.07814.pdf)|421|\n|2018|ACL|[Did the Model Understand the Question?](https://arxiv.org/pdf/1805.05492.pdf)|171|[Tensorflow](https://github.com/pramodkaushik/acl18_results)|\n|2018|FITEE|[Visual interpretability for deep learning: a survey](https://arxiv.org/pdf/1802.00614)|731|\n|2018|NeurIPS|[Sanity Checks for Saliency Maps](http://papers.nips.cc/paper/8160-sanity-checks-for-saliency-maps.pdf)|1353|\n|2018|NeurIPS|[Explanations based on the missing: Towards contrastive explanations with pertinent negatives](http://papers.nips.cc/paper/7340-explanations-based-on-the-missing-towards-contrastive-explanations-with-pertinent-negatives.pdf)|443|[Tensorflow](https://github.com/IBM/Contrastive-Explanation-Method)|\n|2018|NeurIPS|[Towards robust interpretability with self-explaining neural networks](http://papers.nips.cc/paper/8003-towards-robust-interpretability-with-self-explaining-neural-networks.pdf)|648|[Pytorch](https://github.com/raj-shah/senn)|\n|2018|NeurIPS|[Attacks meet interpretability: Attribute-steered detection of adversarial samples](https://papers.nips.cc/paper/7998-attacks-meet-interpretability-attribute-steered-detection-of-adversarial-samples.pdf)|142|\n|2018|NeurIPS|[DeepPINK: reproducible feature selection in deep neural networks](https://papers.nips.cc/paper/8085-deeppink-reproducible-feature-selection-in-deep-neural-networks.pdf)|125|[Keras](https://github.com/younglululu/DeepPINK)|\n|2018|NeurIPS|[Representer point selection for explaining deep neural networks](https://papers.nips.cc/paper/8141-representer-point-selection-for-explaining-deep-neural-networks.pdf)|182|[Tensorflow](https://github.com/chihkuanyeh/Representer_Point_Selection)|\n|2018|NeurIPS Workshop|[Interpretable convolutional filters with sincNet](https://arxiv.org/pdf/1811.09725)|97|\n|2018|AAAI|[Anchors: High-precision model-agnostic explanations](https://dm-gatech.github.io/CS8803-Fall2018-DML-Papers/anchors.pdf)|1517|\n|2018|AAAI|[Improving the adversarial robustness and interpretability of deep neural networks by regularizing their input gradients](https://asross.github.io/publications/RossDoshiVelez2018.pdf)|537|[Tensorflow](https://github.com/dtak/adversarial-robustness-public)|\n|2018|AAAI|[Deep learning for case-based reasoning through prototypes: A neural network that explains its predictions](https://arxiv.org/pdf/1710.04806.pdf)|396|[Tensorflow](https://github.com/OscarcarLi/PrototypeDL)|\n|2018|AAAI|[Interpreting CNN Knowledge via an Explanatory Graph](https://arxiv.org/pdf/1708.01785.pdf)|199|[Matlab](https://github.com/zqs1022/explanatoryGraph)|\n|2018|AAAI|[Examining CNN Representations with respect to Dataset Bias](http://www.stat.ucla.edu/~sczhu/papers/Conf_2018/AAAI_2018_DNN_Learning_Bias.pdf)|88|\n|2018|WACV|[Grad-cam++: Generalized gradient-based visual explanations for deep convolutional networks](https://www.researchgate.net/profile/Aditya_Chattopadhyay2/publication/320727679_Grad-CAM_Generalized_Gradient-based_Visual_Explanations_for_Deep_Convolutional_Networks/links/5a3aa2e5a6fdcc3889bd04cb/Grad-CAM-Generalized-Gradient-based-Visual-Explanations-for-Deep-Convolutional-Networks.pdf)|1459|\n|2018|IJCV|[Top-down neural attention by excitation backprop](https://arxiv.org/pdf/1608.00507)|778|\n|2018|TPAMI|[Interpreting deep visual representations via network dissection](https://arxiv.org/pdf/1711.05611)|252|\n|2018|DSP|[Methods for interpreting and understanding deep neural networks](http://iphome.hhi.de/samek/pdf/MonDSP18.pdf)|2046|\n|2018|Access|[Peeking inside the black-box: A survey on Explainable Artificial Intelligence (XAI)](https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8466590)|3110|\n|2018|JAIR|[Learning Explanatory Rules from Noisy Data](https://www.ijcai.org/Proceedings/2018/0792.pdf)|440|[Tensorflow](https://github.com/ai-systems/DILP-Core)|\n|2018|MIPRO|[Explainable artificial intelligence: A survey](https://www.researchgate.net/profile/Mario_Brcic/publication/325398586_Explainable_Artificial_Intelligence_A_Survey/links/5b0bec90a6fdcc8c2534d673/Explainable-Artificial-Intelligence-A-Survey.pdf)|794|\n|2018|BMVC|[Rise: Randomized input sampling for explanation of black-box models](https://arxiv.org/pdf/1806.07421.pdf)|657|\n|2018|arxiv|[Distill-and-Compare: Auditing Black-Box Models Using Transparent Model Distillation](https://arxiv.org/pdf/1710.06169.pdf)|194|\n|2018|arxiv|[Manipulating and measuring model interpretability](https://arxiv.org/pdf/1802.07810.pdf)|496|\n|2018|arxiv|[How convolutional neural network see the world-A survey of convolutional neural network visualization methods](https://arxiv.org/pdf/1804.11191.pdf)|211|\n|2018|arxiv|[Revisiting the importance of individual units in cnns via ablation](https://arxiv.org/pdf/1806.02891.pdf)|93|\n|2018|arxiv|[Computationally Efficient Measures of Internal Neuron Importance](https://arxiv.org/pdf/1807.09946.pdf)|10|\n|2017|ICML|[Understanding Black-box Predictions via Influence Functions](https://dm-gatech.github.io/CS8803-Fall2018-DML-Papers/influence-functions.pdf)|2062|[Pytorch](https://github.com/nimarb/pytorch_influence_functions)|\n|2017|ICML|[Axiomatic attribution for deep networks](https://mit6874.github.io/assets/misc/sundararajan.pdf)|3654|[Keras](https://github.com/hiranumn/IntegratedGradients)|\n|2017|ICML|[Learning Important Features Through Propagating Activation Differences](https://mit6874.github.io/assets/misc/shrikumar.pdf)|2835|\n|2017|ICLR|[Visualizing deep neural network decisions: Prediction difference analysis](https://arxiv.org/pdf/1702.04595.pdf)|674|[Caffe](https://github.com/lmzintgraf/DeepVis-PredDiff)|\n|2017|ICLR|[Exploring LOTS in Deep Neural Networks](https://openreview.net/pdf?id=SkCILwqex)|34|\n|2017|NeurIPS|[A Unified Approach to Interpreting Model Predictions](http://papers.nips.cc/paper/7062-a-unified-approach-to-interpreting-model-predictions.pdf)|11511|\n|2017|NeurIPS|[Real time image saliency for black box classifiers](https://papers.nips.cc/paper/7272-real-time-image-saliency-for-black-box-classifiers.pdf)|483|[Pytorch](https://github.com/karanchahal/SaliencyMapper)|\n|2017|NeurIPS|[SVCCA: Singular Vector Canonical Correlation Analysis for Deep Learning Dynamics and Interpretability](http://papers.nips.cc/paper/7188-svcca-singular-vector-canonical-correlation-analysis-for-deep-learning-dynamics-and-interpretability.pdf)|473|\n|2017|CVPR|[Mining Object Parts from CNNs via Active Question-Answering](http://openaccess.thecvf.com/content_cvpr_2017/papers/Zhang_Mining_Object_Parts_CVPR_2017_paper.pdf)|29|\n|2017|CVPR|[Network dissection: Quantifying interpretability of deep visual representations](http://openaccess.thecvf.com/content_cvpr_2017/papers/Bau_Network_Dissection_Quantifying_CVPR_2017_paper.pdf)|1254|\n|2017|CVPR|[Improving Interpretability of Deep Neural Networks with Semantic Information](http://openaccess.thecvf.com/content_cvpr_2017/papers/Dong_Improving_Interpretability_of_CVPR_2017_paper.pdf)|118|\n|2017|CVPR|[MDNet: A Semantically and Visually Interpretable Medical Image Diagnosis Network](http://openaccess.thecvf.com/content_cvpr_2017/papers/Zhang_MDNet_A_Semantically_CVPR_2017_paper.pdf)|307|[Torch](https://github.com/zizhaozhang/mdnet-cvpr2017)|\n|2017|CVPR|[Making the V in VQA matter: Elevating the role of image understanding in Visual Question Answering](http://openaccess.thecvf.com/content_cvpr_2017/papers/Goyal_Making_the_v_CVPR_2017_paper.pdf)|1686|\n|2017|CVPR|[Knowing when to look: Adaptive attention via a visual sentinel for image captioning](http://openaccess.thecvf.com/content_cvpr_2017/papers/Lu_Knowing_When_to_CVPR_2017_paper.pdf)|1392|[Torch](https://github.com/jiasenlu/AdaptiveAttention)|\n|2017|CVPRW|[Interpretable 3d human action analysis with temporal convolutional networks](http://openaccess.thecvf.com/content_cvpr_2017_workshops/w20/papers/Kim_Interpretable_3D_Human_CVPR_2017_paper.pdf)|539|\n|2017|ICCV|[Grad-cam: Visual explanations from deep networks via gradient-based localization](http://openaccess.thecvf.com/content_ICCV_2017/papers/Selvaraju_Grad-CAM_Visual_Explanations_ICCV_2017_paper.pdf)|13006|[Pytorch](https://github.com/leftthomas/GradCAM)|\n|2017|ICCV|[Interpretable Explanations of Black Boxes by Meaningful Perturbation](http://openaccess.thecvf.com/content_ICCV_2017/papers/Fong_Interpretable_Explanations_of_ICCV_2017_paper.pdf)|1293|[Pytorch](https://github.com/jacobgil/pytorch-explain-black-box)|\n|2017|ICCV|[Interpretable Learning for Self-Driving Cars by Visualizing Causal Attention](http://openaccess.thecvf.com/content_ICCV_2017/papers/Kim_Interpretable_Learning_for_ICCV_2017_paper.pdf)|323|\n|2017|ICCV|[Understanding and comparing deep neural networks for age and gender classification](http://openaccess.thecvf.com/content_ICCV_2017_workshops/papers/w23/Lapuschkin_Understanding_and_Comparing_ICCV_2017_paper.pdf)|130|\n|2017|ICCV|[Learning to disambiguate by asking discriminative questions](http://openaccess.thecvf.com/content_ICCV_2017/papers/Li_Learning_to_Disambiguate_ICCV_2017_paper.pdf)|26|\n|2017|IJCAI|[Right for the right reasons: Training differentiable models by constraining their explanations](https://arxiv.org/pdf/1703.03717.pdf)|429|\n|2017|IJCAI|[Understanding and improving convolutional neural networks via concatenated rectified linear units](http://www.jmlr.org/proceedings/papers/v48/shang16.pdf)|510|[Caffe](https://github.com/chakkritte/CReLU)|\n|2017|AAAI|[Growing Interpretable Part Graphs on ConvNets via Multi-Shot Learning](https://arxiv.org/pdf/1611.04246.pdf)|67|[Matlab](https://github.com/zqs1022/partGraphForCNN)|\n|2017|ACL|[Visualizing and Understanding Neural Machine Translation](https://www.aclweb.org/anthology/P17-1106.pdf)|179|\n|2017|EMNLP|[A causal framework for explaining the predictions of black-box sequence-to-sequence models](https://arxiv.org/pdf/1707.01943.pdf)|192|\n|2017|CVPR Workshop|[Looking under the hood: Deep neural network visualization to interpret whole-slide image analysis outcomes for colorectal polyps](http://openaccess.thecvf.com/content_cvpr_2017_workshops/w8/papers/Korbar_Looking_Under_the_CVPR_2017_paper.pdf)|47|\n|2017|survey|[Interpretability of deep learning models: a survey of results](https://discovery.ucl.ac.uk/id/eprint/10059575/1/Chakraborty_Interpretability%20of%20deep%20learning%20models.pdf)|345|\n|2017|arxiv|[SmoothGrad: removing noise by adding noise](https://arxiv.org/pdf/1706.03825.pdf)|1479|\n|2017|arxiv|[Interpretable \u0026 explorable approximations of black box models](https://arxiv.org/pdf/1707.01154.pdf)|259|\n|2017|arxiv|[Distilling a neural network into a soft decision tree](https://arxiv.org/pdf/1711.09784.pdf)|520|[Pytorch](https://github.com/kimhc6028/soft-decision-tree)|\n|2017|arxiv|[Towards interpretable deep neural networks by leveraging adversarial examples](https://arxiv.org/pdf/1708.05493.pdf)|111|\n|2017|arxiv|[Explainable artificial intelligence: Understanding, visualizing and interpreting deep learning models](https://arxiv.org/pdf/1708.08296.pdf)|1279|\n|2017|arxiv|[Contextual Explanation Networks](https://arxiv.org/pdf/1705.10301.pdf)|77|[Pytorch](https://github.com/alshedivat/cen)|\n|2017|arxiv|[Challenges for transparency](https://arxiv.org/pdf/1708.01870.pdf)|142|\n|2017|ACMSOPP|[Deepxplore: Automated whitebox testing of deep learning systems](https://machine-learning-and-security.github.io/papers/mlsec17_paper_1.pdf)|1144|\n|2017|CEURW|[What does explainable AI really mean? A new conceptualization of perspectives](https://arxiv.org/pdf/1710.00794.pdf)|518|\n|2017|TVCG|[ActiVis: Visual Exploration of Industry-Scale Deep Neural Network Models](https://arxiv.org/pdf/1704.01942.pdf)|346|\n|2016|NeurIPS|[Synthesizing the preferred inputs for neurons in neural networks via deep generator networks](http://papers.nips.cc/paper/6519-synthesizing-the-preferred-inputs-for-neurons-in-neural-networks-via-deep-generator-networks.pdf)|659|[Caffe](https://github.com/Evolving-AI-Lab/synthesizing)|\n|2016|NeurIPS|[Understanding the effective receptive field in deep convolutional neural networks](https://papers.nips.cc/paper/6203-understanding-the-effective-receptive-field-in-deep-convolutional-neural-networks.pdf)|1356|\n|2016|CVPR|[Inverting Visual Representations with Convolutional Networks](https://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Dosovitskiy_Inverting_Visual_Representations_CVPR_2016_paper.pdf)|626|\n|2016|CVPR|[Visualizing and Understanding Deep Texture Representations](http://openaccess.thecvf.com/content_cvpr_2016/papers/Lin_Visualizing_and_Understanding_CVPR_2016_paper.pdf)|147|\n|2016|CVPR|[Analyzing Classifiers: Fisher Vectors and Deep Neural Networks](https://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Bach_Analyzing_Classifiers_Fisher_CVPR_2016_paper.pdf)|191|\n|2016|ECCV|[Generating Visual Explanations](https://arxiv.org/pdf/1603.08507)|613|[Caffe](https://github.com/LisaAnne/ECCV2016)|\n|2016|ECCV|[Design of kernels in convolutional neural networks for image classification](https://arxiv.org/pdf/1511.09231.pdf)|24|\n|2016|ICML|[Understanding and improving convolutional neural networks via concatenated rectified linear units](http://www.jmlr.org/proceedings/papers/v48/shang16.pdf)|510|\n|2016|ICML|[Visualizing and comparing AlexNet and VGG using deconvolutional layers](https://icmlviz.github.io/icmlviz2016/assets/papers/4.pdf)|126|\n|2016|EMNLP|[Rationalizing Neural Predictions](https://arxiv.org/pdf/1606.04155)|738|[Pytorch](https://github.com/zhaopku/Rationale-Torch)|\n|2016|IJCV|[Visualizing deep convolutional neural networks using natural pre-images](https://arxiv.org/pdf/1512.02017)|508|[Matlab](https://github.com/aravindhm/nnpreimage)|\n|2016|IJCV|[Visualizing Object Detection Features](https://arxiv.org/pdf/1502.05461.pdf)|38|[Caffe](https://github.com/cvondrick/ihog)|\n|2016|KDD|[Why should i trust you?: Explaining the predictions of any classifier](https://chu-data-lab.github.io/CS8803Fall2018/CS8803-Fall2018-DML-Papers/lime.pdf)|11742|\n|2016|TVCG|[Visualizing the hidden activity of artificial neural networks](https://www.researchgate.net/profile/Samuel_Fadel/publication/306049229_Visualizing_the_Hidden_Activity_of_Artificial_Neural_Networks/links/5b13ffa7aca2723d9980083c/Visualizing-the-Hidden-Activity-of-Artificial-Neural-Networks.pdf)|309|\n|2016|TVCG|[Towards better analysis of deep convolutional neural networks](https://arxiv.org/pdf/1604.07043.pdf)|474|\n|2016|NAACL|[Visualizing and understanding neural models in nlp](https://arxiv.org/pdf/1506.01066)|650|[Torch](https://github.com/jiweil/Visualizing-and-Understanding-Neural-Models-in-NLP)|\n|2016|arxiv|[Understanding neural networks through representation erasure](https://arxiv.org/pdf/1612.08220.pdf))|492|\n|2016|arxiv|[Grad-CAM: Why did you say that?](https://arxiv.org/pdf/1611.07450.pdf)|398|\n|2016|arxiv|[Investigating the influence of noise and distractors on the interpretation of neural networks](https://arxiv.org/pdf/1611.07270.pdf)|108|\n|2016|arxiv|[Attentive Explanations: Justifying Decisions and Pointing to the Evidence](https://arxiv.org/pdf/1612.04757)|88|\n|2016|arxiv|[The Mythos of Model Interpretability](http://www.zacklipton.com/media/papers/mythos_model_interpretability_lipton2016.pdf)|3786|\n|2016|arxiv|[Multifaceted feature visualization: Uncovering the different types of features learned by each neuron in deep neural networks](https://arxiv.org/pdf/1602.03616)|317|\n|2015|ICLR|[Striving for Simplicity: The All Convolutional Net](https://arxiv.org/pdf/1412.6806.pdf)|4645|[Pytorch](https://github.com/StefOe/all-conv-pytorch)|\n|2015|CVPR|[Understanding deep image representations by inverting them](https://www.cv-foundation.org/openaccess/content_cvpr_2015/papers/Mahendran_Understanding_Deep_Image_2015_CVPR_paper.pdf)|1942|[Matlab](https://github.com/aravindhm/deep-goggle)|\n|2015|ICCV|[Understanding deep features with computer-generated imagery](http://openaccess.thecvf.com/content_iccv_2015/papers/Aubry_Understanding_Deep_Features_ICCV_2015_paper.pdf)|156|[Caffe](https://github.com/mathieuaubry/features_analysis)|\n|2015|ICML Workshop|[Understanding Neural Networks Through Deep Visualization](https://arxiv.org/pdf/1506.06579.pdf)|2038|[Tensorflow](https://github.com/jiye-ML/Visualizing-and-Understanding-Convolutional-Networks)|\n|2015|AAS|[Interpretable classifiers using rules and Bayesian analysis: Building a better stroke prediction model](https://projecteuclid.org/download/pdfview_1/euclid.aoas/1446488742)|749|\n|2014|ECCV|[Visualizing and Understanding Convolutional Networks](https://arxiv.org/pdf/1311.2901.pdf)|18604|[Pytorch](https://github.com/huybery/VisualizingCNN)|\n|2014|ICLR|[Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps](https://arxiv.org/pdf/1312.6034.pdf)|6142|[Pytorch](https://github.com/huanghao-code/VisCNN_ICLR_2014_Saliency)|\n|2013|ICCV|[Hoggles: Visualizing object detection features](https://www.cv-foundation.org/openaccess/content_iccv_2013/papers/Vondrick_HOGgles_Visualizing_Object_2013_ICCV_paper.pdf)|352|\n \n+ [ ] 论文talk\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fonetaken%2Fawesome_deep_learning_interpretability","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fonetaken%2Fawesome_deep_learning_interpretability","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fonetaken%2Fawesome_deep_learning_interpretability/lists"}