2018-3-24 paper (Grey Wolf Optimizer) note1----------No free lunch theorem_2018 year goal and daily completion record

Like our primary and secondary school learning function, the teacher will always say to us:

"Be sure to write the definition fields, do not write no points, do not write definition domain who knows is that kind of situation"

So is the analogy to the algorithm, which is today's NFL theory:

"Machine Learning" the first Chapter 1.4 summary, has proved:

Http://www.tup.tsinghua.edu.cn/upload/books/yz/064027-01.pdf

The final text of machine learning: So, the most important implication of the NFL theorem is that it makes clear to us that it is meaningless to talk about "what is better" than the specific problem, because if all the potential problems in the oven, all learning algorithms are just as good. To talk about the relative advantages and disadvantages of the algorithm, it is necessary to focus on specific learning problems, the performance of a good learning algorithm on some issues, may not be satisfactory on other issues. Learning algorithm's own inductive preference and problem time match, often play a decisive role.

Additional Resources:

Some Comments on Gradient-free optimization

A qualified study should strictly define the scope of the application of its own algorithm