机器学习人工学weekly-2018/9/23

注意下面很多链接需要科*学*上*网，无奈国情如此

1. Facebook从图片和视频extract字符，整个系统颇复杂

Rosetta: Understanding text in images and videos with machine learning

链接：https://code.fb.com/ai-research/rosetta-understanding-text-in-images-and-videos-with-machine-learning/

2. Facebook Glow，给ML硬件加速器的编译器，像XLA和NVM

Glow: A community-driven approach to AI infrastructure

链接：https://code.fb.com/ml-applications/glow-a-community-driven-approach-to-ai-infrastructure/

3. DeepMind PopArt，用adaptive rescaling target让不同的Atari游戏能被同时学习

Preserving Outputs Precisely while Adaptively Rescaling Targets

链接：https://deepmind.com/blog/preserving-outputs-precisely-while-adaptively-rescaling-targets/

4. Google新的听声辨歌功能

Google’s Next Generation Music Recognition

链接：https://ai.googleblog.com/2018/09/googles-next-generation-music.html

5. Google What-if工具

The What-If Tool: Code-Free Probing of Machine Learning Models

链接：https://ai.googleblog.com/2018/09/the-what-if-tool-code-free-probing-of.html

6. 随机梯度（非常简单的想法）

Fast, Better Training Trick — Random Gradient

链接：https://arxiv.org/pdf/1808.04293v1.pdf

7. RL transfer learning

Visual Transfer between Atari Games using Competitive Reinforcement Learning

链接：https://arxiv.org/pdf/1809.00397.pdf

8. VAE介绍

Variational Autoencoders Explained

链接：http://anotherdatum.com/vae.html

9. attention解释

Neural Attention Please, or How Attention Models Can Work for You

链接：https://insidebigdata.com/2018/09/14/neural-attention-please-attention-models-can-work/

10. The Gradient新博文-CycleGAN

Playing a game of GANstruction

链接：https://thegradient.pub/playing-a-game-of-ganstruction/