2018机器学习开源资源盘点- 专知

2018机器学习开源资源盘点

【导读】本文梳理了过去一年令人惊叹的49个开源代码和项目，共分为计算机视觉、强化学习、自然语言处理、GAN、神经网络和工具包六个部分。Github 平均Star数为3566.

计算机视觉

Detectron: FAIRs research platform for object detection research, implementing popular algorithms like Mask R-CNN and RetinaNet. [18913 Stars].

Openpose: Real-time multi-person keypoint detection library for body, face, and hands estimation [11052 stars on Github].

DensePose: A real-time approach for mapping all human pixels of 2D RGB images to a 3D surface-based model of the body [4165 stars on Github].

Maskrcnn-benchmark: Fast, modular reference implementation of Semantic Segmentation and Object Detection algorithms in PyTorch. [3888 stars on Github].

SNIPER is an efficient multi-scale object detection algorithm [1963 stars on Github].

强化学习

Psychlab: Experimental paradigms implemented using the Psychlab platform (3D platform for agent-based AI) [5595 stars on Github]

ELF: An Extensive, Lightweight, and Flexible platform for game research. We have used it to build our Go playing bot, ELF OpenGo, which achieved a 14–0 record versus four global top-30 players [2406 stars on Github].

TRFL: A library of useful building blocks for writing reinforcement learning (RL) agents in TensorFlow [2312 stars on Github].

Horizon: The first open source reinforcement learning platform for large-scale products and services [1703 stars on Github].

Chess-alpha-zero: Chess reinforcement learning by AlphaGo Zero methods. [1307 stars on Github].

Dm_control: The DeepMind Control Suite and Control Package [1231 stars on Github].

MAMEToolkit: Arcade Game Reinforcement Learning Python Library [437 stars on Github].

自然语言处理

Reaver: Reaver: Modular Deep Reinforcement Learning Framework. Focused on StarCraft II. Supports Gym, Atari, and MuJoCo. Matches reference results. [355 stars on Github].

Bert: TensorFlow code and pre-trained models for BERT [11703 stars on Github].

Pytext: A natural language modeling framework based on PyTorch [4466 stars on Github].

Bert-as-service: A NLP model developed by Google for pre-training language representations. It leverages an enormous amount of plain text data publicly available on the web and is trained in an unsupervised manner. [2055 stars on Github].

UnsupervisedMT: Phrase-Based & Neural Unsupervised Machine Translation — Facebook Research [1068 stars on Github].

DecaNLP: The Natural Language Decathlon: A Multitask Challenge for NLP — Salesforce [1648 stars on Github].

Nlp-architect: NLP Architect by Intel AI Lab: Python library for exploring the state-of-the-art deep learning topologies and techniques for NLP [1751 stars on Github].

Gluon-nlp: NLP made easy [1263 stars on Github].

GAN

DeOldify: A Deep Learning based project for colorizing and restoring old images [5060 stars on Github].

Progressive_growing_of_gans: Progressive Growing of GANs for Improved Quality, Stability, and Variation [4046 stars on Github].

MUNIT: Multimodal Unsupervised Image-to-Image Translation [1339 stars on Github].

Transparent_latent_gan: Use supervised learning to illuminate the latent space of GAN for controlled generation and edit [1337 stars on Github].

Gandissect: Pytorch-based tools for visualizing and understanding the neurons of a GAN. [1065 stars on Github].

GANimation: Anatomically-aware Facial Animation from a Single Image [869 stars on Github].

神经网络

Fastai: It simplifies training fast and accurate neural nets using modern best practices [11597 stars on Github].

DeepCreamPy: Decensoring Hentai with Deep Neural Networks [7046 stars on Github].

Augmentor v0.2: Image augmentation library in Python for machine learning. [2805 stars on Github].

Graph_nets: Build Graph Nets in Tensorflow [2723 stars on Github].

Textgenrnn: Python module to easily generate text using a pretrained character-based recurrent neural network. [1900 stars on Github].

Person-blocker: Automatically “block” people in images (like Black Mirror) using a pretrained neural network. [1806 stars on Github].

Deepvariant: DeepVariant is an analysis pipeline that uses a deep neural network to call genetic variants from next-generation DNA sequencing data. [1502 stars on Github].

Video-nonlocal-net: Non-local Neural Networks for Video Classification [1049 stars on Github].

Ann-visualizer: A python library for visualizing Artificial Neural Networks (ANN) [922 stars on Github].

工具包

Tfjs: A WebGL accelerated, browser based JavaScript library for training and deploying ML models. [10268 stars on Github].

Dopamine: A research framework for fast prototyping of reinforcement learning algorithms — Google [7142 stars on Github].

Lime: Explaining the predictions of any machine learning classifier [5173 stars on Github].

Autokeras: An open source software library for automated machine learning (AutoML) [4520 stars on Github].

Shap: Explain the output of any machine learning model using expectations and Shapley values. [3496 stars on Github].

MMdnn: A set of tools to help users inter-operate among different deep learning frameworks. E.g. model conversion and visualization. Convert models between Caffe, Keras, MXNet, Tensorflow [3021 stars on Github].

Mlflow: Open source platform for the machine learning lifecycle [3013 stars on Github].

Mace: A deep learning inference framework optimized for mobile heterogeneous computing platforms. [2979 stars on Github].

PySyft: A Python library for secure, private Deep Learning. PySyft decouples private data from model training, using Multi-Party Computation (MPC) within PyTorch [2595 stars on Github].

Adanet: Fast and flexible AutoML with learning guarantees. [2293 stars on Github].

Tencent-ml-images: Largest multi-label image database; ResNet-101 model; 80.73% top-1 acc on ImageNet [2094 stars on Github].

Donkeycar: Open source hardware and software platform to build a small scale self driving car. [1207 stars on Github].

PocketFlow: An Automatic Model Compression (AutoMC) framework for developing smaller and faster AI applications. [1677 stars on Github].

DALI: A library containing both highly optimized building blocks and an execution engine for data pre-processing in deep learning applications [1013 stars on Github].