机器学习人工学weekly-2018/8/19

注意下面很多链接需要访问外国网站，无奈国情如此

1. 大神的profile，最近的optimizer search, AutoAugment，device placement, MnasNet, Swish，ENAS全部参与。n久前有幸1:1过一次聊seq2seq

An Unassuming Genius: the Man behind Google’s AutoML

链接：

https://medium.com/@aifrontiers/an-unassuming-genius-the-man-behind-google-brains-automl-4ddc801f3e9b

2. Saleforce开源给结构化数据做ML的项目，支持自动选模型，自动调参，炼丹必备

Open Sourcing TransmogrifAI: Automated Machine Learning for Structured Data

链接：

https://engineering.salesforce.com/open-sourcing-transmogrifai-4e5d0e098da2

3. Netflix给Jupyter Notebook做了很多定制的项目

Beyond Interactive: Notebook Innovation at Netflix

链接：

https://medium.com/@NetflixTechBlog/notebook-innovation-591ee3221233

4. Lyft的自驾车stack，搞了那么一堆东西最后就是为了输出steering/speed control :)

5. facebook ML视频

Introducing the Facebook Field Guide to Machine Learning video series

链接：

https://research.fb.com/the-facebook-field-guide-to-machine-learning-video-series

6. Pinterest的graph convolution neural network，做图片推荐用

PinSage: A New Graph Convolutional Neural Network for Web-Scale Recommender Systems

链接：

https://medium.com/@Pinterest_Engineering/pinsage-a-new-graph-convolutional-neural-network-for-web-scale-recommender-systems-88795a107f48

7. CNN调试技巧

Troubleshooting Convolutional Neural Networks

链接：

https://gist.github.com/zeyademam/0f60821a0d36ea44eef496633b4430fc

8. 用dask并行化特征工程

Parallelizing Feature Engineering with Dask

链接：

https://towardsdatascience.com/parallelizing-feature-engineering-with-dask-3db88aec33b7

9. 炼丹调参工具

Understanding Hyperparameters Optimization in Deep Learning Models: Concepts and Tools

链接：

https://towardsdatascience.com/understanding-hyperparameters-optimization-in-deep-learning-models-concepts-and-tools-357002a3338a

10. hyperas调keras参数

A guide to an efficient way to build neural network architectures- Part I: Hyper-parameter selection and tuning for Dense Networks using Hyperas on Fashion-MNIST

链接：

https://towardsdatascience.com/a-guide-to-an-efficient-way-to-build-neural-network-architectures-part-i-hyper-parameter-8129009f131b

11. 各种convolution的解释

An Introduction to different Types of Convolutions in Deep Learning

链接：

https://towardsdatascience.com/types-of-convolutions-in-deep-learning-717013397f4d

12. Google改进backprop，不直接使用derivative，用进化算法去搜索update，也算是AutoML

Backprop Evolution

链接：

https://arxiv.org/pdf/1808.02822.pdf

13. fastText做NLP教程

Text Classification is Your New Secret Weapon

链接：

https://medium.com/@ageitgey/text-classification-is-your-new-secret-weapon-7ca4fad15788

14. 游戏AI简介

The Total Beginner's Guide to Game AI

链接:

https://www.gamedev.net/articles/programming/artificial-intelligence/the-total-beginners-guide-to-game-ai-r4942

15. RL各种算法简介

Introduction to Various Reinforcement Learning Algorithms.

链接：

https://towardsdatascience.com/introduction-to-various-reinforcement-learning-algorithms-i-q-learning-sarsa-dqn-ddpg-72a5e0cb6287

https://towardsdatascience.com/introduction-to-various-reinforcement-learning-algorithms-part-ii-trpo-ppo-87f2c5919bb9