Machine Learning Weekly Review №5 — the source of latest credible papers, videos and projects on machine learning for scientists and engineers.

Recommended this week


  1. Stanford CoreNLP — production grade natural language software. API available in Python, Scala, JavaScript and more.
  2. Vectorflow is a minimalist neural network library optimized for sparse data and single machine environments. Blog Post.
  3. Merlin — Open Source Neural Network Speech Synthesis System.
  4. Academicpages — ready-to-fork GitHub Pages template for Academic Personal Websites. By R. Stuart Geiger.
  5. TensorFire is a framework for running neural networks in the browser, accelerated by WebGL.
  6. Deeplearn.js — javasript hardware-accelerated machine learning library from Google Brain(Daniel Smilkov).


  1. Interpretable Active Learning”. Presented at 2017 ICML Workshop. Expands on LIME framework.
  2. ClothCap: Seamless 4D Clothing Capture and Retargeting”. Segments and tracks garment pieces and estimates body shape. Demo.
  3. Learning to Reinforcement Learn”. Earns Prize for Applied Cognition at CogSci 2017.
  4. Photographic Image Synthesis with Cascaded Refinement Networks”. Synthesizes photographic images at 2-megapixel resolution. TensorFlow code.
  5. Causal Transfer Learning”. NIPS 2017. Generic method identifies sets of features that lead to transferable predictions.
  6. Proceedings of International Machine Learning Research (ICML 2017).
  7. SmoothGrad: removing noise by adding noise” from Google Brain (Daniel Smilkov). PyTorch code.


  1. A Beginner’s Guide to Optimizing Pandas Code for Speed” by Sofia Heisler.
  2. A 2017 Guide to Semantic Segmentation with Deep Learning”.
  3. Guides and best practices for effective use of Tensorflow” by Vahid Kazemi.
  4. A Step-by-Step Guide to Synthesizing Adversarial Examples. By Anish Athalye.


  1. CVPR 2017 Video Recordings.
  2. Concise introductory course on probabilistic graphical models.

You would also like

Facebook Group: Machine Learning Review