Machine Learning Weekly Review №6 — the source of latest credible papers, videos and projects on machine learning for scientists and engineers.

Recommended this week


  1. Ncnn is a high-performance neural network inference framework from Tencent. Optimized for the mobile platform and edge computing.
  2. art-DCGAN — Modified version of DCGAN with a focus on generating artworks by Robbie Barrat.
  3. Beholder — A TensorBoard plugin for visualizing arbitrary tensors in a video as your network trains.
  4. Collection of generative models, e.g. GAN, VAE in Pytorch and Tensorflow.
  5. AsciiMath is an easy-to-write markup language for mathematics. LaTeX alternative that works in a Browser.
  6. ChainerCV – a Library for Computer Vision in Deep Learning.
  7. AudioSet — alarge-scale dataset of manually annotated audio events by Google Developers.
  8. Tslearn — Python package for the time series analysis.
  9. text2vec — fast vectorization, topic modeling, distances and GloVe word embeddings in R.


  1. Filtering Variational Objectives”. Uses a particle filter to form a tighter bound then ELBO. ICML2017.
  2. PGAN — Probabilistic Generative Adversarial Networks” by Hamid. NIPS 2017 Submission.
  3. “A Neural Parametric Singing Synthesizer”. Allows conveniently modify pitch to match any target melody.
  4. Unbiased Markov chain Monte Carlo with Couplings”. With unbiased MCMC we can compute R estimators in parallel and take their average. Blog Post.
  5. Uncertainty in Deep Learning”. PhD Thesis by Yarin Gal. Research Fellow at University of Cambridge.
  6. Create Anime Characters with A.I.” Based on DRAGAN. Technical Report. Demo.
  7. Sound-Word2Vec: Learning Word Representations Grounded in Sounds”. Accepted at EMNLP 2017.
  8. What Actions are Needed for Understanding Human Actions in Video?”. ICCV2017. Code.
  9. Semantic Instance Segmentation with a Discriminative Loss Function”. CVPR2017.
  10. SurfaceNet: An End-to-end 3D Neural Network for Multiview Stereopsis”. Takes a set of images and their corresponding camera parameters as input and directly infers the 3D model. ICCV 2017 Poster.
  11. Understanding Black-box Predictions via Influence Functions”. Explains black-box model by requiring only an oracle access to gradients and Hessian-vector products. ICML2017. Code.
  12. Exploiting the Natural Exploration In Contextual Bandits”. Greedy-First — outperforms exploration-based contextual bandit algorithms Thompson sampling, UCB, or ϵ-greedy.


  1. Deep Learning for NLP Best Practices” by Sebastian Ruder.
  2. Gentle introduction to the Bregman divergences” by Mark Reid.
  3. Predictive Vision In A Nutshell” by Filip Piekniewski.
  4. Graph Convolutional Networks” by Thomas Kipf.
  5. How I Used Deep Learning To Train A Chatbot To Talk Like Me” by Adit Deshpande.


  1. Building Mobile Applications with TensorFlow” free book by Pete Warden.
  2. Introduction to Artificial Neural Networks and Deep Learning. A Practical Guide with Applications in Python ebook by Sebastian Raschka.
  3. Python Data Science Handbook” in the form of Jupyter notebooks.


  1. Convolutional Neural Networks for Visual Recognition” lectures by Stanford University.
  2. HarvardX Biomedical Data Science Open Online Training. Videos and Code for “Genomics Data Analysis Series”, “Using Python for Research” Courses.

You would also like

Facebook Group: Machine Learning Review