Visual-analytics-and-Interpretability-in-Deep-Learning
本项目主要是通过可视分析的手段,对深度学习的可解释性做出讨论与探讨。并且记录小组成员的学习过程与工作
Part 1 分享
资料
- Deep Learning Lab:作者Paulo Rauber关于深度学习的教程资料
- Deep Learning:作者Hugo Larochelle(http://www.dmi.usherb.ca/~larocheh/)
- 深度学习的屌丝逆袭超简史:作者Hugo Larochelle
- Feature Visualization:How neural networks build up their understanding of images
- Ideas on Interpreting machine learning
- 花书:deeplearning解读
- awesome-interpretable-machine-learning
- 机器学习日报
- 知乎:为什么都说神经网络是个黑箱?
- Explainable AI: Visualizing and Interpreting Machine Learning
- 有哪些讨论深度学习、机器学习的论坛推荐?
- CV牛人牛事简介之一
- handong1587.github.io/_posts/deep_learning/2015-10-09-visulizing-interpreting-cnn.md
- Exploring Neural Networks with Activation Atlases
- Awesome-XAI
- Awesome Deep Vision
- interpretable_machine_learning_with_python
- awesome-machine-learning-interpretability
- 电子书:Interpretable Machine Learning
- xai_resources
- DALEX
课程
- Andrew Ng:CS230: Deep Learning
- cs229 MachineLearning
- Deep Learning for Computer Vision
- Getting Started in Computer VisionResearch
重要进展
- 破解AI大脑黑盒,谷歌迈出新的一步
- DeepMind:Understanding deep learning through neuron deletion
- tensorflow 2018 summit
文献
方法论
度量变量相关性
博文
Blog
- Google AI blog
- 可解释性博文
- Deep Dive into Math Behind Deep Networks
- Understanding the ‘black box’ of artificial intelligence
中文博文
英文博文
- Essentials of Deep Learning: Visualizing Convolutional Neural Networks in Python
- How to use Tensorboard with PyTorch
- Visual LSTM
- Visualizing Layer Representations in Neural Networks
- GPU-accelerated Neural Networks in JavaScript
- Deep Learning’s Uncertainty Principle
- Visualize World Trends using Seaborn in Python
- Modern Theory of Deep Learning: Why Does It Work so Well
- Why Deep Learning Works
- A Comprehensive Design Guide for Image Classification CNNs
- 深度学习三巨头等口述神经网络复兴史
- Understanding Batch Normalization with Examples in Numpy and Tensorflow with Interactive Code
- Human Interpretable Machine Learning (Part 1) — The Need and Importance of Model Interpretation
- Almost any Image Classification Problem using PyTorch
- 人工智能领域必须知道的深度学习博客!!
基础介绍博文
Part 2 写成研究论文
- How to write a Research Paper
- Ten simple rules for structuring papers
- Writing a Journal Article
- 魏秀参:A novice's guide to papers writing with LaTeX (in Chinese)
Part 3 相关研究组织
- PKU Visualization Blog
- VGA 浙大可视化研究组
- Georgia Tech : Machine Learning Visualization & Interpretation
- VALSE:视觉与青年学者研讨会
Part 4 github相关开源项目
- GAN lab
- tensorflow.js
- ShapeShop
- darksight
- CNN filters category selectivity
- JS与deeplearning关系
- ANNvisualizer
- CNN fixation
- Fabrik:Collaboratively build, visualize, and design neural nets in browser
- advis:adversarial visualization built with tensorflow.js and React
- CNN_VISUALIZING(ECCV14:Visualizing and Understanding Convolutional Networks)
- Seq2Seq-Vis 2018
- AdversarialDNN-Playground
- MXboard
- Yellowbrick
- interactive-classification
- gpu.js
4.1 DeepLearning与前端相关开源项目
- Screenshot-to-code-in-Keras自动生成html code
- sketch-code由草图生成网页
4.2 对抗样本
4.3 tensorflow.js相关资料
- Introduction to the TensorFlow.js core API
- Visualizing activations of MobileNet with TensorFlow.js
- Realtime tSNE Visualizations with TensorFlow.js
Part 5 相关顶会论文集
- IJCAI:International Joint Conferences on Artificial Intelligence Organization
- 机器学习&深度学习顶会(2017、2018)
- 2019 accept list
Part 6 理论层次
信息论与深度学习的结合
- (MEDIUM) Mukul Malik:Information Theory of Neural Networks
- 知乎:机器学习入门:重要的概念---信息熵(Shannon’s Entropy Model)
- Must know Information Theory concepts in Deep Learning (AI)
- 案例研究:InfoGAN
- Gan指南
Part 7 Visual Analytics Community Classical Project
- classilist
- [LstmVis]
- [RnnVis]
- [Revacnn]
Part 8 Model Evaluation system
Part 9 科研路漫漫
- Lessons from My First Two Years of AI Research
- 文献综述德思维导图
- 如何写好博士论文
- 死磕论文前,不如先找齐一套好用的工具
- 科研论文各章节该用什么时态?这些点90%的人都没注意到!
- 没有导师的指导,研究生如何阅读文献、提出创见、写论文?
Part 9.1 如何做好AI文献阅读
Part 10 Tutorial
Part 11 便利技巧
- Work remotely with PyCharm, TensorFlow and SSH
- Comprehensive Beginner’s Guide to Jupyter Notebooks for Data Science & Machine
- How to unit test machine learning code
Part 12 中文博客
thesis defense
Paper
机器学习
- 机器学习中回归模型的5种损失函数
- 深度学习优化导论:梯度下降
- 矩阵求导
- 机器学习中的范数规则化之(一)L0、L1与L2范数
- 求导:All the Backpropagation derivatives
- 剑桥大学:156页PPT全景展示AI过去的12个月
- State of AI
- Machine-Learning-Tutorials
深度学习
贝叶斯网络与图
- 代表作者:Thomas Kipf
- 贝叶斯网络之父:Judea Pearl
- 【信息汇总】国际“顶尖”计算机视觉、机器学习会议大搜罗--附排名&接收率
- CAM 方法
- Essentials of Deep Learning: Visualizing Convolutional Neural Networks in Python
科技报告文章
视频
会议报告
工具
评估
可视化机器学习算法
网络模型压缩
Researcher
相关议题
python 工具包
线性代数理解
计算机视觉的入门
Visualization for machine learning
- 2018 Nips Tutorial
- Martin Wattenberg Homepage
- 作品集:HINT.FM
- Image Kernels
- 作品集
- code
- 作品集2
- Effective Visualization of Multi-Dimensional Data — A Hands-on Approach
可解释性教程
Docker 相关资料
Jupyter 相关资料
- Jupyter Notebook Extensions
- awesome-jupyter
- jupyter_contrib_nbextensions
- Jupyter Superpower — Interactive Visualization Combo with Python
cs领域各顶会历届best paper(不足之处 没有vis会议)
python数据可视化
- The Next Level of Data Visualization in Python
- Author:data-visualization
- Top 50 matplotlib Visualizations – The Master Plots (with full python code)
- blog:My experiences at the VAST challenge
优秀博士论文
论文搜集网址
图像分类数据集
计算机课程资源
数据可视化教程材料
青年基金申请
经典案例
- What Causes Heart Disease? Explaining the Model
- A Complete Exploratory Data Analysis and Visualization for Text Data
- Black-box vs. white-box models
Github Trending
神经网络训练技巧
可解释性相关工程
- The official cli and python API client for W&B
- microsoft/interpret
- 可交互式:Initializing neural networks