How to learn machine learning algorithms

Learning machine learning algorithms is really a headache, we have so many papers, books, websites can be consulted, they are either refined mathematical description (mathematically), or a step-by-Step text Introduction (textually). If you're lucky enough, you might find some pseudo-code. If the character breaks out, you will even be told how to install it. However, all rely on the character is not a permanent, exhaustive algorithm cultivation cheats is also a few. How good is this dilemma?
A chevalier integration of many years of income, refers to a clear road: from "a few words" in the cobwebs, multi-reference, can be the canon.
This "word" is not random pick, but refers to the total of a certain faction algorithm. The original source of the algorithm (ie, the article), as well as from the review and the classics of the two exposition. In these places, code implementations of algorithms are often hidden. If you study diligently, will be more effective.
The algorithm can be widely used in the knowledge of the algorithm, in order to peep the whole picture, then the code implementation. This is a method, but the algorithm is the source, the code is supported. After the chase, the algorithm heart is to be. However, the heart through the practice, pay attention to the accumulation of accumulated patience, samadhi. The matter is very difficult. The Chevalier also raised an auxiliary cultivation method--structured algorithm learning.
Before practicing an algorithm, list a number of issues to focus on, including:

• What are the criteria and abbreviations for the algorithm? (What's the abbreviations used for the algorithm?)
• What is the information processing strategy? (What is the information processing strategy of the algorithm? ）
• What is the target of the algorithm? (What's the objective or goal for the algorithm?)
• What are the derivative practices of the algorithm? (What metaphors or analogies is commonly used to describe the behavior of the algorithm?)
• Pseudo-code implementation (what is the pseudocode or flowchart description of the algorithm?)
• What are the techniques and precautions for using the algorithm? (What is the heuristics or rules of thumb for using the algorithm?)
• What kinds of problems can the algorithm solve? (What classes of problem are the algorithm well suited?)
• What are the resources that describe the relevant algorithms? (What is useful resources for learning more about the algorithm?)
• Where is the source of the algorithm? (What is the primary references or resources in which the algorithm is first described?)
  In daily learning, the discussion of the same algorithm is grouped into the upper frame.
  The subtlety of this approach is that you don't need to be a part of the world beforehand. Just every encounter an algorithm, it will fall into the framework, the accumulation of accumulated, the natural and increasing.

Reference:

1 How to learn a machine learning algorithm.

How to learn machine learning algorithms