awesome_deep_learning_interpretability
1.0.0
Deep Learning related papers on model explanatory in recent years.
Sort by reference number
159 papers pdf (2 of them need to be found on Scihub) were uploaded to Tencent Weiyun.
Updated from time to time.
| Year | Publication | Paper | Citation | code |
|---|---|---|---|---|
| 2020 | CVPR | Explaining Knowledge Distillation by Quantifying the Knowledge | 81 | |
| 2020 | CVPR | High-frequency Component Helps Explain the Generalization of Convolutional Neural Networks | 289 | |
| 2020 | CVPRW | Score-CAM: Score-Weighted Visual Explanations for Convolutional Neural Networks | 414 | Pytorch |
| 2020 | ICLR | Knowledge consistency between neural networks and beyond | 28 | |
| 2020 | ICLR | Interpretable Complex-Valued Neural Networks for Privacy Protection | twenty three | |
| 2019 | AI | Explanation in artistic intelligence: Insights from the social sciences | 3248 | |
| 2019 | NMI | Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models instead | 3505 | |
| 2019 | NeurIPS | Can you trust your model's uncertainty? Evaluating predictive uncertainty under dataset shift | 1052 | - |
| 2019 | NeurIPS | This looks like that: deep learning for interpretable image recognition | 665 | Pytorch |
| 2019 | NeurIPS | A benchmark for interpretation methods in deep neural networks | 413 | |
| 2019 | NeurIPS | Full-gradient representation for neural network visualization | 155 | |
| 2019 | NeurIPS | On the (In) fidelity and Sensitivity of Explanations | 226 | |
| 2019 | NeurIPS | Towards Automatic Concept-based Explanations | 342 | Tensorflow |
| 2019 | NeurIPS | CXPlain: Causal explanations for model interpretation under uncertainty | 133 | |
| 2019 | CVPR | Interpreting CNNs via Decision Trees | 293 | |
| 2019 | CVPR | From Recognition to Cognition: Visual Commonsense Reasoning | 544 | Pytorch |
| 2019 | CVPR | Attention branch network: Learning of attention mechanism for visual explanation | 371 | |
| 2019 | CVPR | Interpretable and fine-grained visual explanations for convolutional neural networks | 116 | |
| 2019 | CVPR | Learning to Explain with Complemental Examples | 36 | |
| 2019 | CVPR | Revealing Scenes by Inverting Structure from Motion Reconstructions | 84 | Tensorflow |
| 2019 | CVPR | Multimodal Explanations by Predicting Counterfactuality in Videos | 26 | |
| 2019 | CVPR | Visualizing the Resilience of Deep Convolutional Network Interpretations | 2 | |
| 2019 | ICCV | U-CAM: Visual Explanation using Uncertainty based Class Activation Maps | 61 | |
| 2019 | ICCV | Towards Interpretable Face Recognition | 66 | |
| 2019 | ICCV | Taking a HINT: Leveraging Explanations to Make Vision and Language Models More Grounded | 163 | |
| 2019 | ICCV | Understanding Deep Networks via Extremal Perturbations and Smooth Masks | 276 | Pytorch |
| 2019 | ICCV | Explaining Neural Networks Semantically and Quantitatively | 49 | |
| 2019 | ICLR | Hierarchical interpretations for neural network predictions | 111 | Pytorch |
| 2019 | ICLR | How Important Is a Neuron? | 101 | |
| 2019 | ICLR | Visual Explanation by Interpretation: Improving Visual Feedback Capabilities of Deep Neural Networks | 56 | |
| 2018 | ICML | Extracting Automata from Recurrent Neural Networks Using Queries and Counterexamples | 169 | Pytorch |
| 2019 | ICML | Towards A Deep and Unified Understanding of Deep Neural Models in NLP | 80 | Pytorch |
| 2019 | ICAIS | Interpreting black box predictions using fisher kernels | 80 | |
| 2019 | ACMFAT | Explaining explanations in AI | 558 | |
| 2019 | AAAI | Interpretation of neural networks is fragile | 597 | Tensorflow |
| 2019 | AAAI | Classifier-agnostic saliency map extraction | twenty three | |
| 2019 | AAAI | Can You Explain That? Lucid Explanations Help Human-AI Collaborative Image Retrieval | 11 | |
| 2019 | AAAIW | Unsupervised Learning of Neural Networks to Explain Neural Networks | 28 | |
| 2019 | AAAIW | Network Transplanting | 4 | |
| 2019 | CSUR | A Survey of Methods for Explaining Black Box Models | 3088 | |
| 2019 | JVCIR | Interpretable convolutional neural networks via feedforward design | 134 | Keras |
| 2019 | ExplainAI | The (Un)reliability of salivency methods | 515 | |
| 2019 | ACL | Attention is not Explanation | 920 | |
| 2019 | EMNLP | Attention is not not Explanation | 667 | |
| 2019 | arxiv | Attention Interpretability Across NLP Tasks | 129 | |
| 2019 | arxiv | Interpretable CNNs | 2 | |
| 2018 | ICLR | Towards better understanding of gradient-based attribute methods for deep neural networks | 775 | |
| 2018 | ICLR | Learning how to explain neural networks: PatternNet and PatternAttribution | 342 | |
| 2018 | ICLR | On the importance of single directions for generalization | 282 | Pytorch |
| 2018 | ICLR | Detecting statistical interactions from neural network weights | 148 | Pytorch |
| 2018 | ICLR | Interpretable counting for visual question answering | 55 | Pytorch |
| 2018 | CVPR | Interpretable Convolutional Neural Networks | 677 | |
| 2018 | CVPR | Tell me where to look: Guided attention inference network | 454 | Chainer |
| 2018 | CVPR | Multimodal Explanations: Justifying Decisions and Pointing to the Evidence | 349 | Caffe |
| 2018 | CVPR | Transparency by design: Close the gap between performance and interpretation in visual reasoning | 180 | Pytorch |
| 2018 | CVPR | Net2vec: Quantifying and explaining how concepts are encoded by filters in deep neural networks | 186 | |
| 2018 | CVPR | What have we learned from deep representations for action recognition? | 52 | |
| 2018 | CVPR | Learning to Act Properly: Predicting and Explaining Affordances from Images | 57 | |
| 2018 | CVPR | Teaching Categories to Human Learners with Visual Explanations | 64 | Pytorch |
| 2018 | CVPR | What do deep networks like to see? | 36 | |
| 2018 | CVPR | Interpret Neural Networks by Identifying Critical Data Routing Paths | 73 | Tensorflow |
| 2018 | ECCV | Deep clustering for unsupervised learning of visual features | 2056 | Pytorch |
| 2018 | ECCV | Explainable neural computing via stack neural module networks | 164 | Tensorflow |
| 2018 | ECCV | Grounding visual explanations | 184 | |
| 2018 | ECCV | Textual explanations for self-driving vehicles | 196 | |
| 2018 | ECCV | Interpretable basis decomposition for visual explanation | 228 | Pytorch |
| 2018 | ECCV | Convnets and imagenet beyond accuracy: Understanding mistakes and uncovering biases | 147 | |
| 2018 | ECCV | Vqa-e: Explaining, elaborating, and enhancing your answers for visual questions | 71 | |
| 2018 | ECCV | Choose Your Neuron: Incorporating Domain Knowledge through Neuron-Importance | 41 | Pytorch |
| 2018 | ECCV | Diverse feature visualizations reveal invariances in early layers of deep neural networks | twenty three | Tensorflow |
| 2018 | ECCV | ExplainGAN: Model Explanation via Decision Boundary Crossing Transformations | 36 | |
| 2018 | ICML | Interpretability beyond feature attribute: Quantitative testing with concept activation vectors | 1130 | Tensorflow |
| 2018 | ICML | Learning to explain: An information-theoretic perspective on model interpretation | 421 | |
| 2018 | ACL | Did the Model Understand the Question? | 171 | Tensorflow |
| 2018 | FITEE | Visual interpretation for deep learning: a survey | 731 | |
| 2018 | NeurIPS | Sanity Checks for Salivation Maps | 1353 | |
| 2018 | NeurIPS | Explanations based on the missing: Towards contrary explanations with pertinent negatives | 443 | Tensorflow |
| 2018 | NeurIPS | Towards robust interpretation with self-explaining neural networks | 648 | Pytorch |
| 2018 | NeurIPS | Attacks meet interpretability: Attribute-steered detection of adversarial samples | 142 | |
| 2018 | NeurIPS | DeepPINK: reproducible feature selection in deep neural networks | 125 | Keras |
| 2018 | NeurIPS | Representer point selection for explaining deep neural networks | 182 | Tensorflow |
| 2018 | NeurIPS Workshop | Interpretable convolutional filters with sincNet | 97 | |
| 2018 | AAAI | Anchors: High-precision model-agnostic explanations | 1517 | |
| 2018 | AAAI | Improving the adversarial robustness and interpretation of deep neural networks by regularizing their input gradients | 537 | Tensorflow |
| 2018 | AAAI | Deep learning for case-based reasoning through prototypes: A neural network that explains its predictions | 396 | Tensorflow |
| 2018 | AAAI | Interpreting CNN Knowledge via an Explanatory Graph | 199 | Matlab |
| 2018 | AAAI | Examining CNN Representations with respect to Dataset Bias | 88 | |
| 2018 | WACV | Grad-cam++: Generalized gradient-based visual explanations for deep convolutional networks | 1459 | |
| 2018 | IJCV | Top-down neural attention by excitation backprop | 778 | |
| 2018 | TPAMI | Interpreting deep visual representations via network dissection | 252 | |
| 2018 | DSP | Methods for interpreting and understanding deep neural networks | 2046 | |
| 2018 | Access | Peeking inside the black-box: A survey on Explainable Artificial Intelligence (XAI) | 3110 | |
| 2018 | JAIR | Learning Explanatory Rules from Noisy Data | 440 | Tensorflow |
| 2018 | MIPRO | Explainable artistic intelligence: A survey | 794 | |
| 2018 | BMVC | Rise: Randomized input sampling for explanation of black-box models | 657 | |
| 2018 | arxiv | Distill-and-Compare: Auditing Black-Box Models Using Transparent Model Distillation | 194 | |
| 2018 | arxiv | Manipulating and measuring model interpretation | 496 | |
| 2018 | arxiv | How convolutional neural network see the world-A survey of convolutional neural network visualization methods | 211 | |
| 2018 | arxiv | Revisiting the importance of individual units in cnns via ablation | 93 | |
| 2018 | arxiv | Computationally Efficient Measures of Internal Neuron Importance | 10 | |
| 2017 | ICML | Understanding Black-box Predictions via Influence Functions | 2062 | Pytorch |
| 2017 | ICML | Axiomatic attribute for deep networks | 3654 | Keras |
| 2017 | ICML | Learning Important Features Through Propagating Activation Differenties | 2835 | |
| 2017 | ICLR | Visualizing deep neural network decisions: Prediction difference analysis | 674 | Caffe |
| 2017 | ICLR | Exploring LOTS in Deep Neural Networks | 34 | |
| 2017 | NeurIPS | A Unified Approach to Interpreting Model Predictions | 11511 | |
| 2017 | NeurIPS | Real time image saliency for black box classes | 483 | Pytorch |
| 2017 | NeurIPS | SVCCA: Singular Vector Canonical Correlation Analysis for Deep Learning Dynamics and Interpretability | 473 | |
| 2017 | CVPR | Mining Object Parts from CNNs via Active Question-Answering | 29 | |
| 2017 | CVPR | Network dissection: Quantifying interpretability of deep visual representations | 1254 | |
| 2017 | CVPR | Improving Interpretability of Deep Neural Networks with Semantic Information | 118 | |
| 2017 | CVPR | MDNet: A Semantically and Visually Interpretable Medical Image Diagnosis Network | 307 | Torch |
| 2017 | CVPR | Making the V in VQA matter: Elevating the role of image understanding in Visual Question Answering | 1686 | |
| 2017 | CVPR | Knowing when to look: Adaptive attention via a visual sentinel for image captioning | 1392 | Torch |
| 2017 | CVPRW | Interpretable 3d human action analysis with temporal convolutional networks | 539 | |
| 2017 | ICCV | Grad-cam: Visual explanations from deep networks via gradient-based localization | 13006 | Pytorch |
| 2017 | ICCV | Interpretable Explanations of Black Boxes by Meaningful Perturbation | 1293 | Pytorch |
| 2017 | ICCV | Interpretable Learning for Self-Driving Cars by Visualizing Causal Attention | 323 | |
| 2017 | ICCV | Understanding and comparing deep neural networks for age and gender classification | 130 | |
| 2017 | ICCV | Learning to disambiguate by asking discriminative questions | 26 | |
| 2017 | IJCAI | Right for the right reasons: Training differentiable models by constraining their explanations | 429 | |
| 2017 | IJCAI | Understanding and improving convolutional neural networks via concatenated rectified linear units | 510 | Caffe |
| 2017 | AAAI | Growing Interpretable Part Graphs on ConvNets via Multi-Shot Learning | 67 | Matlab |
| 2017 | ACL | Visualizing and Understanding Neural Machine Translation | 179 | |
| 2017 | EMNLP | A causeal framework for explaining the predictions of black-box sequence-to-sequence models | 192 | |
| 2017 | CVPR Workshop | Looking under the hood: Deep neural network visualization to interpret whole-slide image analysis outcomes for colorful polys | 47 | |
| 2017 | Survey | Interpretability of deep learning models: a survey of results | 345 | |
| 2017 | arxiv | SmoothGrad: removing noise by adding noise | 1479 | |
| 2017 | arxiv | Interpretable & explore approximations of black box models | 259 | |
| 2017 | arxiv | Distilling a neural network into a soft decision tree | 520 | Pytorch |
| 2017 | arxiv | Towards interpretable deep neural networks by leveraging adversarial examples | 111 | |
| 2017 | arxiv | Explainable artistic intelligence: Understanding, visualizing and interpreting deep learning models | 1279 | |
| 2017 | arxiv | Contextual Explanation Networks | 77 | Pytorch |
| 2017 | arxiv | Challenges for transparency | 142 | |
| 2017 | ACMSOPP | Deepxplore: Automated whitebox testing of deep learning systems | 1144 | |
| 2017 | CEURW | What does explainable AI really mean? A new conceptualization of perspectives | 518 | |
| 2017 | TVCG | ActiVis: Visual Exploration of Industry-Scale Deep Neural Network Models | 346 | |
| 2016 | NeurIPS | Synthesizing the preferred inputs for neurons in neural networks via deep generator networks | 659 | Caffe |
| 2016 | NeurIPS | Understanding the effective receptive field in deep convolutional neural networks | 1356 | |
| 2016 | CVPR | Inverting Visual Representations with Convolutional Networks | 626 | |
| 2016 | CVPR | Visualizing and Understanding Deep Texture Representations | 147 | |
| 2016 | CVPR | Analyzing Classifiers: Fisher Vectors and Deep Neural Networks | 191 | |
| 2016 | ECCV | Generating Visual Explanations | 613 | Caffe |
| 2016 | ECCV | Design of kernels in convolutional neural networks for image classification | twenty four | |
| 2016 | ICML | Understanding and improving convolutional neural networks via concatenated rectified linear units | 510 | |
| 2016 | ICML | Visualizing and comparing AlexNet and VGG using deconvolutional layers | 126 | |
| 2016 | EMNLP | Rationalizing Neural Predictions | 738 | Pytorch |
| 2016 | IJCV | Visualizing deep convolutional neural networks using natural pre-images | 508 | Matlab |
| 2016 | IJCV | Visualizing Object Detection Features | 38 | Caffe |
| 2016 | KDD | Why should i trust you?: Explaining the predictions of any classifier | 11742 | |
| 2016 | TVCG | Visualizing the hidden activity of artistic neural networks | 309 | |
| 2016 | TVCG | Towards better analysis of deep convolutional neural networks | 474 | |
| 2016 | NAACL | Visualizing and understanding neural models in nlp | 650 | Torch |
| 2016 | arxiv | Understanding neural networks through representation erasure) | 492 | |
| 2016 | arxiv | Grad-CAM: Why did you say that? | 398 | |
| 2016 | arxiv | Investing the influence of noise and distractors on the interpretation of neural networks | 108 | |
| 2016 | arxiv | Attentive Explanations: Justifying Decisions and Pointing to the Evidence | 88 | |
| 2016 | arxiv | The Mythos of Model Interpretability | 3786 | |
| 2016 | arxiv | Multifaceted feature visualization: Uncovering the different types of features learned by each neuron in deep neural networks | 317 | |
| 2015 | ICLR | Striving for Simplicity: The All Convolutional Net | 4645 | Pytorch |
| 2015 | CVPR | Understanding deep image representations by inverting them | 1942 | Matlab |
| 2015 | ICCV | Understanding deep features with computer-generated imagery | 156 | Caffe |
| 2015 | ICML Workshop | Understanding Neural Networks Through Deep Visualization | 2038 | Tensorflow |
| 2015 | AAS | Interpretable classes using rules and Bayesian analysis: Building a better stroke prediction model | 749 | |
| 2014 | ECCV | Visualizing and Understanding Convolutional Networks | 18604 | Pytorch |
| 2014 | ICLR | Deep Inside Convolutional Networks: Visualising Image Classification Models and Salivation Maps | 6142 | Pytorch |
| 2013 | ICCV | Hoggles: Visualizing object detection features | 352 |
