1

Topic: In what an essence ?

1. Than it in essence differs from the Database, from algorithm. At the heart of its operation any other principles lie? 2. What means "to train "? 3. In what advantage  before algorithm producing data array statistic analysis consists?

2

Re: In what an essence ?

Hello, developer, you wrote: D> 1. Than it in essence differs from the Database, from algorithm. At the heart of its operation any other principles lie? The essence that functioning bases  are programmed only, further it "does all itself". D> 2. What means "to train "? To run through it 100500 dial-ups of the input data with known result therefore it will produce the right answer with a certain share of probability. As it is widely used  at training on the same dial-ups of the input data each time will turn out another  - with other coefficients D> 3. In what advantage  before algorithm producing data array statistic analysis consists? There is no algorithm - it is not necessary to think Programmed , "trained" it and further all plows. Appeared rather recently because it is necessary a lot of computing power that it worked with comprehensible time of training. But now everywhere push - in a fantastic way, any Hindu, and algorithm good it of figs can program invents

3

Re: In what an essence ?

Hello, developer, you wrote: Children saw?

4

Re: In what an essence ?

Hello, sambl4, you wrote: S> There is no algorithm - it is not necessary to think Programmed , "trained" it and further all plows. Appeared rather recently because it is necessary a lot of computing power that it worked with comprehensible time of training. But now everywhere push - in a fantastic way, any Hindu, and algorithm good it of figs can program invents Yes, Hindus . In  even the apple photo is simple so do not use, it is necessary to prepare it specially.

5

Re: In what an essence ?

Hello, developer, you wrote: D> 1. Than it in essence differs from the Database, from algorithm. At the heart of its operation any other principles lie? D> 2. What means "to train "? D> 3. In what advantage  before algorithm producing data array statistic analysis consists? 1, 2 - study a subject, please. 3 - trained  also is as a matter of fact result of operation of such algorithm.

6

Re: In what an essence ?

Hello, developer, you wrote: D> 1. Than it in essence differs from the Database, from algorithm. At the heart of its operation any other principles lie? D> 2. What means "to train "? D> 3. In what advantage  before algorithm producing data array statistic analysis consists? Well roughly speaking, this certain  function and to train they be to pick up its coefficients. Well type, a polynomial and its coefficients.

7

Re: In what an essence ?

Hello, developer, you wrote: D> 1. Than it in essence differs from the Database, from algorithm. At the heart of its operation any other principles lie? D> 2. What means "to train "? D> 3. In what advantage  before algorithm producing data array statistic analysis consists? The Simplex method transited in HIGH SCHOOL? Here, if strongly roughly to simplify,  it something of that type. Only criteria of filling of a matrix the network tries to search itself (therefore it is necessary a pack of the initial data, the more - the better), type as "for a method of branches and boundaries" from the optimization theory. Well and grid functions with convolutions too it would be necessary to recall, for  to a trained grid (model of communications between neurons). Well and further still all on slightly from higher mathematics. Only any it not Machine intelligence as many blow at all corners. Yes,  can search is cool in a fantastic way for raisin in a roll. But she there never adds it, and some powdered sugar on top does not pour. It is not given commands.

8

Re: In what an essence ?

Hello, developer, you wrote: D> 1. Than it in essence differs from the Database, from algorithm. At the heart of its operation any other principles lie? The database is a file. The algorithm is sequence of actions. The artificial neural network is the graph. D> 2. What means "to train "? To apply algorithm which allows to change weight of the graph so that the network approximated a given function as necessary. D> 3. In what advantage  before algorithm producing data array statistic analysis consists? With growth of volume of learning sampling quality of statistical model gradually is decelerated, reaching any ceiling. At networks this ceiling above, therefore current interest to networks in the big level is caused by stored given and new, faster methods of training. On small data volumes neural networks can as nedo-/be retrained, and strongly to concede to other methods in accuracy. In separate areas, for example, in the image understanding, new algorithms of training and new architecture of networks allowed to get rid from manual feature engineering (I find it difficult to translate correctly), raising possibility of total model to generalize.

9

Re: In what an essence ?

Hello, developer, you wrote: D> 1. Than it in essence differs from the Database, from algorithm. At the heart of its operation any other principles lie? D> 2. What means "to train "? Here, as correctly above speak, it is better to read the textbook, than a forum. D> 3. In what advantage  before algorithm producing data array statistic analysis consists? At statistical analysis, the author of algorithm sets the signs which statistican to consider. For example, an average and a dispersion, either the histogram, or an amount of sharp transitions above a certain limit, .  can select demanded statistical signs from the crude data, and already on their basis to learn to classify. For tasks, type of pattern recognition and of some others it gave the big break in accuracy of results.