顶级机器学习论文的项目实践 | 年度 Top50
前言
过去一年，机器学习领域涌现出多篇重量级论文，其中一些技术已经有了表现上佳的项目实践。这里整理了50个年度最佳项目，涵盖图像处理、风格转换、图像分类、面部识别、视频防抖、目标检测、自动驾驶、智能推荐、游戏、下棋、医疗、语音生成、音乐、自然语言处理、预测等15个应用领域。AI 技术进展，一文尽览。
50个项目在15个领域的分布情况
No.1
High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs
利用条件 GANs 的高分辨率图像合成及语义操作
项目地址：https://github.com/facebookresearch/fastText
No.2
Using Deep Learning to Create Professional-Level Photographs
利用深度学习创作专业级照片
项目地址：https://research.googleblog.com/2017/07/using-deep-learning-to-create.html
No.3
High Dynamic Range (HDR) Imaging using OpenCV (Python)
利用 OpenCV (Python)高动态范围成像(HDR)
项目地址：https://www.learnopencv.com/high-dynamic-range-hdr-imaging-using-opencv-cpp-python
No.4
Visual Attribute Transfer through Deep Image Analogy
通过深度图像类比进行视觉属性转换（像素级别）
论文地址：https://arxiv.org/abs/1705.01088
No.5
Deep Photo Style Transfer: A deep-learning approach to photographic style transfer that handles a large variety of image content while faithfully transferring the reference style
深度摄影风格转换: 一种处理大量图像内容，同时忠实传递参考风格的深度学习方法
论文地址：https://arxiv.org/abs/1703.07511
No.6
Deep Image Prior:Super-resolution, inpainting, denoising without learning on a dataset and pretrained networks
Deep Image Prior：使用随机初始化神经网络实现图片去噪、超分辨率和修补
项目地址：https://dmitryulyanov.github.io/deep_image_prior
No.7
Feature Visualization: How neural networks build up their understanding of images
特征可视化: 神经网络如何建立对图像的理解
论文地址：https://distill.pub/2017/feature-visualization
No.8
An absolute beginner's guide to Image Classification with Neural Networks
初学者指南：神经网络图像分类
项目地址：https://github.com/humphd/have-fun-with-machine-learning
No.9
Background removal with deep learning
用深度学习清除图像背景
项目地址：https://medium.com/towards-data-science/background-removal-with-deep-learning-c4f2104b3157
No.10
Large Pose 3D Face Reconstruction from a Single Image via Direct Volumetric CNN Regression
基于直接体积回归的单幅图像大姿态三维人脸重建
项目地址：http://aaronsplace.co.uk/papers/jackson2017recon
No.11
Eye blink detection with OpenCV, Python, and dlib
用 OpenCV，Python 和 dlib 捕捉眨眼动作
项目地址：http://www.pyimagesearch.com/2017/04/24/eye-blink-detection-opencv-python-dlib
No.12
DEAL WITH IT in Python with Face Detection
DEAL WITH IT：基于 Python 的人脸识别，实现自动戴墨镜效果
项目地址：https://www.makeartwithpython.com/blog/deal-with-it-generator-face-recognition
No.13
Fused Video Stabilization on the Pixel 2 and Pixel 2 XL
Google 解读 Pixel 2 和 Pixel 2 XL 两款手机的视频拍摄防抖技术
文章地址：https://research.googleblog.com/2017/11/fused-video-stabilization-on-pixel-2.html
No.14
How HBO’s Silicon Valley built “Not Hotdog” with mobile TensorFlow, Keras &
Not Hotdog：热狗识别器
项目地址：https://medium.com/@timanglade/how-hbos-silicon-valley-built-not-hotdog-with-mobile-tensorflow-keras-react-native-ef03260747f3
No.15
Object detection: an overview in the age of Deep Learning
综述：深度学习时代的目标检测技术
文章地址：https://tryolabs.com/blog/2017/08/30/object-detection-an-overview-in-the-age-of-deep-learning
No.16
How to train your own Object Detector with TensorFlow’s Object Detector API
如何使用 TensorFlow 的对象检测器 API 来训练你的对象检测器
文章地址：https://medium.com/towards-data-science/how-to-train-your-own-object-detector-with-tensorflows-object-detector-api-bec72ecfe1d9
No.17
Real-time object detection with deep learning and OpenCV
利用深度学习和 OpenCV 实时监测目标
项目地址：http://www.pyimagesearch.com/2017/09/18/real-time-object-detection-with-deep-learning-and-opencv
No.18
Self-driving Grand Theft Auto V with Python : Intro [Part I] - Sentdex
用 Python 实现“侠盗飞车5”自动驾驶
视频地址：https://www.youtube.com/watch?v=ks4MPfMq8aQ
No.19
Recognizing Traffic Lights With Deep Learning: How I learned deep learning in 10 weeks and won $5,000
用深度学习识别交通信号灯
文章地址：https://medium.freecodecamp.com/recognizing-traffic-lights-with-deep-learning-23dae23287cc
No.20
Spotify’s Discover Weekly: How machine learning finds your new music
机器学习如何找到你喜欢的新歌
项目地址：https://hackernoon.com/spotifys-discover-weekly-how-machine-learning-finds-your-new-music-19a41ab76efe
No.21
Artwork Personalization at Netflix
Netflix 上影视作品的个性化推荐
文章地址：https://medium.com/netflix-techblog/artwork-personalization-c589f074ad76
No.22
MariFlow - Self-Driving Mario Kart w/Recurrent Neural Network
MariFlow：在游戏中利用 RNN 自动驾驶马里奥卡丁车
视频地址：https://www.youtube.com/watch?v=Ipi40cb_RsI
No.23
OpenAI Baselines: DQN. Reproduce reinforcement learning algorithms with performance on par with published results
OpenAI Baselines: DQN：强化学习算法的高质量复现
项目地址：https://blog.openai.com/openai-baselines-dqn
No.24
Reinforcement Learning on Dota 2 [Part II]
在 Dota 2 游戏中，强化学习算法如何击败人类玩家
项目地址：https://blog.openai.com/more-on-dota-2
No.25
Creating an AI DOOM bot
在毁灭战士（DOOM）游戏中创建一个 AI 机器人
项目地址：https://www.codelitt.com/blog/doom-ai
No.26
Phase-Functioned Neural Networks for Character Control
用于角色控制的相位函数神经网络
项目地址：http://theorangeduck.com/page/phase-functioned-neural-networks-character-control
No.27
The Game Imitation: Deep Supervised Convolutional Networks for Quick Video Game AI
用有监督的深度卷积神经网络实现游戏仿真
论文地址：https://arxiv.org/abs/1702.05663
No.28
Introducing: Unity Machine Learning Agents
Unity 3D 游戏中的机器学习智能体
文章地址：https://blogs.unity3d.com/cn/2017/09/19/introducing-unity-machine-learning-agents
No.29
Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm
利用强化学习算法自我博弈，掌握国际象棋和日本将棋
论文地址：https://arxiv.org/abs/1712.01815
No.30
AlphaGo Zero: Learning from scratch | DeepMind
AlphaGo Zero：从零开始学习，掌握围棋
论文地址：https://deepmind.com/blog/alphago-zero-learning-scratch
No.31
How Does DeepMind' AlphaGo Zero Work?
讲解：DeepMind 的 AlphaGo Zero 是如何工作的？
视频地址：https://www.youtube.com/watch?v=vC66XFoN4DE
No.32
A step-by-step guide to building a simple chess AI
实操指南：如何建立一个简单的国际象棋 AI
文章地址：https://medium.freecodecamp.com/simple-chess-ai-step-by-step-1d55a9266977
No.33
CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning
AI 放射科医生：用深度学习技术分析胸透 x 光，检测是否患肺炎
项目地址：https://stanfordmlgroup.github.io/projects/chexnet
No.34
Can you improve lung cancer detection? 2nd place solution for the Data Science Bowl 2017
如何改良对肺癌的检测？看2017年 kaggle 大赛第二名的解决方案
项目地址：http://juliandewit.github.io/kaggle-ndsb2017
No.35
Improving Palliative Care with Deep Learning - Andrew Ng
用深度学习技术改善临终关怀
项目地址：https://stanfordmlgroup.github.io/projects/improving-palliative-care
No.36
Heart Disease Diagnosis with Deep Learning
用深度学习技术诊断心脏病
地址：https://blog.insightdatascience.com/heart-disease-diagnosis-with-deep-learning-c2d92c27e730
No.37
Tacotron: A Fully End-to-End Text-To-Speech Synthesis Model - Data Scientists at Google
Tacotron：谷歌的全端到端语音合成技术
论文地址：https://arxiv.org/abs/1703.10135
No.38
Sequence Modeling with CTC
用 CTC 算法进行序列建模
论文地址：https://distill.pub/2017/ctc
No.39
Deep Voice: Real-time Neural Text-to-Speech - Baidu Silicon Valley AI Lab
Deep Voice：百度的实时语音生成技术
论文地址：https://arxiv.org/abs/1702.07825
No.40
Deep Learning for Siri’s Voice: On-device Deep Mixture Density Networks for Hybrid Unit Selection Synthesis - Apple
Deep Learning for Siri’s Voic：Siri 新声音背后的深度学习技术
文章地址：https://machinelearning.apple.com/2017/08/06/siri-voices.html
No.41
Computer evolves to generate baroque music!
计算机自动生成巴洛克音乐
视频地址：https://www.youtube.com/watch?v=SacogDL_4JU
No.42
Make your own music with WaveNets: Making a Neural Synthesizer Instrument
用 WaveNets 制作个性化音乐
地址：https://magenta.tensorflow.org/nsynth-instrument
No.43
Learning to communicate: Agents developing their own language - OpenAI Research
学会沟通：让 Agents 发明自己的语言
文章地址：https://openai.com/blog/learning-to-communicate
No.44
Big Picture Machine Learning: Classifying Text with Neural Networks and TensorFlow
图解机器学习：神经网络和 TensorFlow 文本分类
文章地址：https://medium.freecodecamp.org/big-picture-machine-learning-classifying-text-with-neural-networks-and-tensorflow-d94036ac2274
No.45
A novel approach to neural machine translation - Facebook AI Research
一种神经机器翻译新方法
项目地址：https://code.facebook.com/posts/1978007565818999/a-novel-approach-to-neural-machine-translation
No.46
How to make a racist AI without really trying
如何让 AI 辨识种族歧视言论
文章地址：https://blog.conceptnet.io/2017/07/13/how-to-make-a-racist-ai-without-really-trying
No.47
Using Machine Learning to Predict Value of Homes On Airbnb
利用机器学习预测 Airbnb 住宅的价值
文章地址：https://medium.com/airbnb-engineering/using-machine-learning-to-predict-value-of-homes-on-airbnb-9272d3d4739d
No.48
Engineering Uncertainty Estimation in Neural Networks for Time Series Prediction at Uber
Uber神经网络时序预测不确定性估计
项目地址：https://eng.uber.com/neural-networks-uncertainty-estimation
No.49
Using Machine Learning to make parking easier
使用机器学习让停车更容易
项目地址：https://research.googleblog.com/2017/02/using-machine-learning-to-predict.html
No.50
How to Predict Stock Prices Easily - Intro to Deep Learning #7
深度学习入门教程：如何轻松预测股票价格
视频地址：https://www.youtube.com/watch?v=ftMq5ps503w
集智AI学园年度回顾系列，点击收藏：
本文由集智小仙女编译整理自Mybridge.co和github.com
关注集智AI学园公众号
获取更多更有趣的AI教程吧！
搜索微信公众号：swarmAI
集智AI学园QQ群：426390994
学园网站：campus.swarma.org
商务合作｜zhangqian@swarma.org
投稿转载｜wangjiannan@swarma.org