Machine Learning Course on Coursera

For the two months that I have sacrificed my spare time to do this :-)

A bit summery for main topics I’ve learned from this machine learning course:

supervised learning:

• linear regression
• logistic regression
• neural networks
• SVMs

unsupervised learning:

• kmeans
• PCA
• anomaly detection

special topics:

• recommender system
• large scale machine leaning(mapReduce, parallel computing, CUDA)

Insight:

• bias/variance, regularization
• deciding what to do next: evaluation of leaning algorithms
• learning curves, error analysis, ceiling analysis.

What’s Next?

Next step is deciding what you would like to focus on. Someone just gave a few that are interesting to pursue but they’re by no means encompassing all that is going on in the field right now.

1. Computer Vision

For building vision system, whether for photos or videos.

Course:

• CS231n Convolutional Neural Networks for Visual Recognition

What you will learn in the course:

• the state-of-art convolutional neural networks
• set up GPU instance in AWS

Follow up would be like:

• scene labeling by combining with Recurrent Neural networks such as LSTM.

2. Natural Language Processing

For machine translation, question and answering, sentiment analysis, etc.

Course:

• CS224d: Deep Learning for Natural Language Processing

What you will learn in the course:

• word embedding (word2vec, GLove)
• bi-directional LSTMs for translation, etc.

3. Deep Reinforcement Learning

• Deepmind’s publication section: Publications - Google DeepMind
• Professor Rich Sutton’s book is must go-to place to start learning RL
• and supplement it with David Silver’s (prominant researcher at Google Deepmind) UCL video lectures: Advanced Topics: RL
• LEARNING REINFORCEMENT LEARNING (WITH CODE, EXERCISES AND SOLUTIONS)

Others You Could Do

Suggested by サービスをつくるエンジニアが機械学習を学ぶべき3つの理由, you could probably read the following resources.

• Rules of Machine Learning: Best Practices for ML Engineering
• Kaggle Data Science Use Cases
• DataQuest
• Hands-On Machine Learning with Scikit-Learn and TensorFlow
• Python機械学習プログラミング (監訳者による読み方が参考になる)
• 戦略的データサイエンス入門

See What Other Engineers Read

• Machine Learning Reading resources
  • 9 Lessons Learned from a Decade Spent Developing Machine Learning Products
  • CNN(卷积神经网络)、RNN(循环神经网络)、DNN(深度神经网络)的内部网络结构有什么区别？
• ML Materials Collection (by a self-driving car engineer)
• Lessons Learned from Deploying Deep Learning at Scale
• CS 20SI: Tensorflow for Deep Learning Research
• Automatic Colorization
• Online graduate-level machine learning course from CMU’s Tom Mitchell via Hacker News
• Udacity: Deep Learning by Google
• See GCP’ Solutions for Machine Learning with Financial Time Series Data
• 6 Ways Companies Leverage Machine Learning Algorithms
• Android 之父 Andy Rubin 新挑戰：讓 AI 無所不在

Reference

• Stanford University’s Machine Learning course on Coursera
• Stanford Unsupervised Feature Learning and Deep Learning Tutorial
• 7 Steps to Mastering Machine Learning With Python
• Neural Networks and Deep Learning free online book
• I have completed Andrew Ng’s Coursera class on Machine Learning. What should I do next? What can I do next?
• サービスをつくるエンジニアが機械学習を学ぶべき3つの理由