How machine learning helped me understand some aspects of the early development of children
When my first son was only two, he already loved cars, he knew all the brands and models (even more than I, thanks to my friends), he could recognize them on a small part of the image. Everyone said: genius. Although they noted the complete uselessness of this knowledge. And the son in the meantime slept with them, rolled them, arranged exactly in a row or a square.
When he was 4 he learned to count, and at 5 he could multiply and add up to 1000. We even played in Math Workout (a game like that on Android - I liked to do it in the subway after work), and at some point he became me do only so. And in his spare time, he counted up to a million, which freezed those around him. Genius! - they said, but we somehow suspected that not quite.
By the way, in the market he helped his mom well - he calculated the total amount faster than the sellers on the calculator.
However, he never played on the court, did not communicate with peers, did not get along very well with the children and caregivers in the garden. In general, I was a little bit of a closed child.
The next stage was geography - we tried to channel the love for numbers somewhere, and handed the old Soviet atlas to our son. He immersed in it for a month, and after that he began to ask us tricky questions in the style:
- Dad, do you think which country has a large area: Pakistan or Mozambique?
"Probably Mozambique," I answered.
- And here not! The area of Pakistan is as much as 2350 km2 more, - the son answered happily.
At the same time, he was not at all interested in the peoples inhabiting these countries, nor their languages, nor their clothes, nor folk music. Only naked figures: the area, the population, the volume of mineral resources, etc.
Everyone again admired. "I'm not smart for years," they said, but I was worried again. I understood that these are completely useless knowledge, not tied to life experience, and which are difficult to continue to develop. The best application from all that I found was the suggestion to calculate how many cars would fit in the parking lot, if any particular country was rolled up with asphalt (without taking into account the mountainous terrain), but quickly grew stilted, tk. it smacks of genocide.
Interestingly, the theme of cars left at that moment, the son did not even remember the name of his favorite cars from his huge collection, which we began to give away with loss of interest. And then he began to count slowly in his mind and soon forgot the area of the countries. At the same time he began to communicate more with peers, became more contact. Genius passed, friends stopped admiring, the son became just a good student with a penchant for mathematics and exact sciences.
Repetition is the mother of learning
It would seem, why all this. This is observed in many children. Their parents all claim that their children are brilliant, grandmothers admire and praise children for their "knowledge". And then they grow out of ordinary just smart children, not more brilliant than the son of my mother's friend.
Studying neural networks, I came across a similar phenomenon, and it seems to me that from this analogy we can draw certain conclusions. I'm not a biologist or a neuroscientist. All further - my guesses without claiming special scientific character. I would be happy to comment on the professionals.
When I tried to understand how my son learned to count faster than me (he passed the level in Math Workout in 20.4 seconds, while my record was 21.9), I realized that he does not think at all. He learned that when you get 55 + 17 you need to press 72. At 45 + 38 you need to click on 8? and so on. The first time he certainly believed, but a spurt in speed occurred at a time when he was able to remember all the combinations. And quickly enough he began to remember not specific inscriptions, but combinations of symbols. This is exactly what they teach in school, by studying the multiplication table - remember the correspondence table MxN -> P.
It turned out that most of the information he perceived as a link between input data and output, and that very general algorithm, which we used to scroll to get the answer, did not simply reduce to a very well sharpened highly specialized algorithm for calculating two-digit numbers. He did slightly different tasks, but much slower. Those. something that everyone seemed super cool, in fact was simply simulated by a well-trained neural network for a specific task.
Why do some children have the ability to remember it, but others do not?
Let's imagine the field of interests of the child (here we approach the question qualitatively, without any measurements). On the left is the field of interests of an ordinary child, and on the right is the field of interests of a "gifted" child. As expected, the main interest is concentrated in areas to which special propensities. But the everyday things and communication with peers focus is not enough. This he considers superfluous.
Interests of an ordinary child 5 years old
The interests of the "brilliant" child are 5 years old
These children brain analyzes and conducts training only on selected topics. By training the neural network in the brain must learn to successfully classify incoming data. But the brain has many, many neurons. Stronger than necessary for normal work with such simple tasks. Usually in life children solve many different tasks, but here all the same resources are thrown on a narrower range of tasks. And training in this mode easily leads to what professionals in ML call overfitting. The network, using an abundance of coefficients (neurons), has learned in such a way that it always gives the right answers (but can produce complete nonsense on the intermediate input data, but nobody sees it). Thus, learning led not to the fact that the brain identified the main characteristics and remembered them, but to the fact that it adjusted many coefficients to produce an accurate result on already known data (as in the picture on the right). And on other topics the brain has learned itself so badly, badly trained (as in the picture on the left).
It seems to me that it is the child's penchant for a particular topic (fixation) and total disregard for other topics that leads to the fact that too many "coefficients" are given to these same subjects.
Given that the network is configured for specific input data and did not allocate "features", but stupidly "remembered" the input data, it can not be used with a few other input data. The applicability of such a network is very narrow. With age, the horizon widens, the focus is blurred, and the ability to assign the same number of neurons to the same task is no longer there - they begin to be used in new tasks more necessary for the child. "Settings" of that overfitted-network collapse, the child becomes "normal", the genius disappears.
Of course, if a child has a skill that is in itself useful and can be developed (for example, music or sports), then its "genius" can be maintained for a long time, and even bring these skills to a professional level. But in most cases this does not work, and from the old skills and the track will not be left to 8-10 years.
if you have a child "genius" and a little more withdrawn than peers, then develop these same skills carefully, actively developing a parallel horizon, and do not focus on these "cool", but usually useless skills.
At you the ingenious child? it will pass;) just develop its other sides, and this "genius" is an effect of too strong training, and not genius at all.
It may be interesting
Thankyou for sharing the data which is beneficial for me and others likewise to see.Gulf Coast Western Reviews