机器学习人工学weekly-2018/4/15

注意下面很多链接需要访问外国网站，无奈国情如此

1. DeepMind的新工作，不用地图在城市里导航

Learning to navigate in cities without a map

链接：https://deepmind.com/blog/

2. 微软学习资源

2.1 微软AI school的学习资源

链接：https://aischool.microsoft.com/learning-paths

2.2 微软professional program

链接：https://academy.microsoft.com/en-us/professional-program/tracks/artificial-intelligence/

3. transformer的使用技巧

Training Tips for the Transformer Model

链接：https://arxiv.org/abs/1804.00247?utm_campaign=Revue%20newsletter&utm_medium=Newsletter&utm_source=Deep%20Learning%20Weekly

4. 2017年机器学习文章总汇

Learn to Build a Machine Learning Application from Top Articles of 2017

链接：https://medium.mybridge.co/learn-to-build-a-machine-learning-application-from-top-articles-of-2017-cdd5638453fc

5. TensorFlow Probability

链接：https://medium.com/tensorflow/introducing-tensorflow-probability-dca4c304e245

6. RL相关

6.1 RL实现tips

Lessons Learned Reproducing a Deep Reinforcement Learning Paper

链接：http://amid.fish/reproducing-deep-rl

6.2 用RL玩2048

The Mathematics of 2048: Optimal Play with Markov Decision Processes

链接：http://jdlm.info/articles/2018/03/18/markov-decision-process-2048.html

7. BAIR新文章机器下惊险动作

Towards a Virtual Stuntman

链接：http://bair.berkeley.edu/blog/2018/04/10/virtual-stuntman/

8. coreML模型文件分析

Reverse Engineering Core ML

链接：https://heartbeat.fritz.ai/reverse-engineering-core-ml-6d6f1c2bdab0

9. Uber弄出来新的differentiable plasticity来做learning to learn

Differentiable Plasticity: A New Method for Learning to Learn

链接：https://eng.uber.com/differentiable-plasticity/?utm_campaign=Revue%20newsletter&utm_medium=Newsletter&utm_source=Deep%20Learning%20Weekly

10. depthwise separable convolution清楚的分析

Depthwise separable convolutions for machine learning

链接：https://eli.thegreenplace.net/2018/depthwise-separable-convolutions-for-machine-learning/