Looked at Siraj Raval 3 months learning machine learning plan of video, feel very good, address: https://www.youtube.com/watch?v=Cr6VqTRO1v0 combined with some we study the experience in a Hybrid machine learning self-study program.

According to Siraj’s suggestion: machine learning involves knowledge proportionally distributed

  1. 35% linear algebra
  2. 25% probability theory and statistics
  3. 15% of calculus
  4. 15% algorithm and its complexity
  5. 10% is data preprocessing knowledge

Highly recommended subscribing: Siraj Raval’s YouTube

To watch his video very comfortable, and a very unique way of learning and useful, the address is: https://www.youtube.com/channel/UCWN3xxRkmTPmbKwht9FuE5A

You may not be familiar with Reddit, but it’s already the fourth most visited site in the United States, behind Google, YouTube and Facebook. It has very high quality and dedicated content. https://www.reddit.com/r/MachineLearning/

Month 1: Math

Linear algebra:

Professor Gillbert Strang’s tutorial is enough: Why not recommend https://www.youtube.com/playlist?list=PL49CF3715CB9EF31D mathematics courses at the university of China, in fact, netease open class with linear algebra courses in college are basically identical although is Chinese but is difficult to learn, Not much interaction. If it’s for the test, that’s fine. Professor Gillbert Strang talked more about thinking methods and principles and various images of metaphor, this way is more suitable for our on-the-job learning, strengthen understanding and thinking. Note: always take notes, not just listen or watch, always take notes, points, questions, thoughts, etc., this is very important, is the key to determine whether you can learn well. The worst thing in the world is when someone who works harder than you and is better than you takes notes that look better than you. https://www.slideshare.net/TessFerrandez/notes-from-coursera-deep-learning-courses-by-andrew-ng attach a picture, look:

Calculus:

The nature of calculus of 3Blue1Brown, the teacher was watching this video to understand calculus, the teacher stupid, watch about 8 times, some videos watched more than 15 times, there is no problem is true, because each video is not long, suitable for repeated watching, but also can improve English ability. https://www.youtube.com/playlist?list=PLZHQObOWTQDMsr9K-rj53DwVRMYO3t5Yr

Probability and statistics:

EdX (MIT and Harvard University jointly create open online classroom platform) has a very good course called “scientific uncertainty” https://www.edx.org/course/introduction-probability-science-mitx-6-041x-2

Month 2: Machine learning

Here we follow Siraj’s advice for The first week: Python, The Math of Intelligence, Tensorflow

Week 2: Machine learning at Udacity

Week 3: Practice machine learning projects

The relevant address is as follows:

python https://www.youtube.com/watch?v=T5pRlIbr6gg

The Math of Intelligence https://www.youtube.com/watch?v=xRJCOz3AfYY

Tensorflow https://www.youtube.com/watch?v=2FmcHiLCwTU

Udacity https://eu.udacity.com/course/intro-to-machine-learning–ud120

Machine learning open source projects at https://github.com/NirantK/awesome-project-ideas

The third month is deep learning

Deep learning is computation-intensive and requires a GPU, even if you’re just getting started. An NVIDIA Tesla K80 GPU costs $2,500. But very lucky is that Google provides us with a free GPU available: Google account registration, landing, visit: https://colab.research.google.com and then to use. The video tutorial is recommended for Siraj himself: https://www.youtube.com/watch?v=vOppzHpvTiQ another all over the world say good is Fast. The AI course, http://course.fast.ai/ finally attached to the open source code of some deep learning, also can realize, To https://github.com/llSourcell?tab=repositories on one’s own making

conclusion

This paper introduces a self-study plan of machine learning and related resources, and ensures 2 hours of dedicated learning time every day. The focus is to think more and find patterns to solve problems. Instead of treating your brain as a solid state disk to store data, you should treat your brain as a CPU or GPU to calculate.