Machine Learning

"Machine Learning or ML is a class of Artificial Intelligence methods, the characteristic feature of which is not a direct solution to a problem, but training in the application of solutions to many similar problems. To construct such methods, mathematical statistics, numerical methods, optimization methods, probability theory, theory are used graphs, various techniques for working with data in digital form."

"The class of Artificial Intelligence methods (AI or Artificial Intelligence) ..." Of course - although in a different edition for a different audience - there is also a definition of AI. True, in this case the author, time and even place of publication are already indicated (John McCarthy, 1956, AI Conference).

Although at the time of preparing of this material, the WiKi page with a Russian analogue of definition of AI had a postscript from which it became clear that the corresponding text was not checked, I will not criticize the respected author, because it was indirectly done in another publication. Moreover, more than 60 years have passed.

Here I intend to present my own vision of Machine Learning as the "Largest Strokes," because it becomes uncomfortable with "Using tools of mathematical statistics, numerical methods, optimization methods, probability theory, graph theory, various techniques for working with data in digital form." For some reason, there is a feeling that he had touched a science-like solidified in mold, which did not have, does not and will not have a practical significance.

On this, one could leave the claim, if not for the huge number of young minds spending time and effort on designing useful mechanical boobs, which, unfortunately, are not much different from the gears in the gearbox of an ordinary car. Without a really tangible result in the field of AI, inspiration and a young fuse will very quickly supplant disappointment and emotional emptiness, as it already happened in the middle of the last, twentieth century.

We should not to forget that we are Homo Sapiens, and we do not know anything reasonable except Homo really. We have not yet learned how to talk with animals and plants. Various publications of the most talented science fiction writers as an alien person describe a mirror with our own reflection, where we note what we are ready to agree with only. For this reason, to transfer FOR OTHER MIND - even artificial and fashioned with your own hands - your own Self, is probably premature and presumptuous. This is something like prosthetics, moreover, of little intellectual.

And what I understand when we use the "Machine Learning" phrase?

1. Answering the question "Who is the Trainee?", I believe that it is "The Machine itself (a corresponding special hardware and software) and its customers (people and organisms, including other machines)."
2. To the question "What is Learning?", I believe that the answer "Information obtained from various sources, including from own repository and offline" is correct.
3. Answers to the questions "How, What, When, and Where?" depend on the relevant sources.

For example, I believe that it is necessary to stop "Using entities beyond what is necessary" and instead of the obscure and ugly Neuro-Linguistic Programming, return to the good old Interactive Help, which runs on demand, offering context-dependent content accumulated during training. Without discussing various claims, which probably also have the reasons, it is sufficient to indicate that our option technologically indicates what is being implemented and how.

And although the chain of actions for obtaining information already available at the time of the request for transfer to the student can be quite complicated, nevertheless, it is intuitive and to a greater or lesser extent already implemented (for example, Oracle iLearning and Oracle ApexApps). And where will the data come from, which is not yet available? And how then to extract the VALUABLE INFORMATION? For example, what are 1, 2, 3, and 4? How to establish that the first three values are the levels of preparation of the schoolchild in primary school, and 4 is the number of rifling in the barrel of the Mosin Rifle, that there are no mutual relations in the four, but there are more and less strong and weak Dependencies? I tried to give an answer to this question in the publications "The Method of Automated Discovery" and "Intelligence or Prosthesis".

Finally: "What's next, What's abroad Artificial Intelligence?" This is the "Universal Interpreter." In the "Summa Technologiae" book Stanislav Lem gave the answer to the such question. #AlexeyPopovitch #USA #Britain