Raising a child vs Machine learning: compares a young mother
Elena Gerasimova, Head of Data Science at Netology, translated an article by Lai Queffelec about what the processes of raising children and teaching AI are similar to.
If you, like me, raised children and at the same time taught the algorithm, then most likely you compared these two processes. And even if you are not fond of artificial intelligence, but know a lot about children, welcome to the wonderful world of educating machines ... oops, machine learning.
When writing this article, no child was hurt. I just, like any parent, spend many hours watching my child learn about the world, and am surprised at his behavioral patterns. Just as a data scientist does, observing the results of train / test samples ( training sample of data for learning the algorithm / result of the algorithm on the new data - ed. ).
This is a quote from Jim Stern , author of Artificial Intelligence for Marketing: Practical Applications, from a lecture on machine learning — not about children (I love children!).
')
The essence of machine learning is to actually teach a machine to perform a specific task - just as parents dream of teaching children to collect dirty laundry and put it in the washing machine while mom and dad are relaxing on the couch (admit, tried?).
However, the main difference is that when a child is asked to wash clothes, he already knows what the clothes look like; able to walk, grab, pull and fold - these actions he learned through other events in his young life.
So where is the key that ultimately opens up the opportunity to be lazy on the couch while laundry is magically erased? Context. We give children examples: we show how to perform each step and thank them when they do everything right - because we love them.
Machine learning is pretty much the same thing, except that the “virtual child” still has the experience of a newborn with the abilities of an older baby. Therefore, it is necessary to begin an explanation from scratch: these five pieces, like sausages, which stick out at the end of a long, sausage-like stick, are fingers, a hand and a palm. Only then should you be shown how to perform the necessary actions with their help - grab and pull. The data set that you give to the machine is all you need to get started, but also everything that exists in the world for it. What it does not yet possess is ...
Usually, people successfully distinguish between men and women. Liam, my son, also copes well with this — and I did not give him a large set of marked data at the entrance. I didn’t sit with him in the park and didn’t point at people, saying “man, man, woman, man, woman” - because, let's be honest, it would be strange. Yes, and not necessary. The car lacks the luxury of common sense, which the child has, and which is used already at the first collision with a new concept.
By common sense, I mean:
The ability to make the right decisions and make the right assumptions based on logical thinking and accumulated experience - Wiktionary
Of course, when a child decides to jump his head into the ground from a height, we quite reasonably doubt his common sense. Nevertheless, it exists and allows children to learn from all their experiences. At the same time, no one is clearly broadcasting to them how to learn to distinguish between men and women.
Explaining the topic of AI to non-dascientists, I like to use an analogy. The child only needs a little observation, a few examples and a couple of corrections to learn how to say "Mr." or "Mrs." And to train the machine to do the same, you need to give her thousands of images. The lack of common sense is probably the number one reason why cars are not yet ready to take over the world.
Liam does strange things, like eating a hot dog, holding his ends and biting in the middle. The standard reaction is to tell him: “Liam! They don’t do that! ” But then I hold back and think that the out-of-box solution is not the best that I can give it. Although when he tries to hold a spoon with his nostrils, one really has to set the limits for acceptable behavior at the table.
This is the great similarity between kids and cars - they are free from social norms and prejudices (or Bayes - from the English. Bias). And this is also the difference between the parents and the date of the Scientists. Babies need to be given a set of values ​​and social norms from which they will build their experience. "Good boundaries", let's call them that. As a data scientist, you are most likely playing the opposite role. The machine should be free from your own norms and prejudices. Bias or bias in algorithms is very dangerous.
Everyone loves gossip and HYIP. For example, Amazon's AI recruiter is a sexist ( Amazon's recruiting AI is sexist ), or the "enhancement" filter FaceApp is a racist ( FaceApp's "hot" filter is racist ). This is a good way to explain to non-data scientists that the role of the scientist and the scientist's date largely boils down to preventing bias and creating as ethical an algorithm as possible.
Image source xkcd
Correlation does not imply a causal relationship. And Nicolas Cage is not a monster who provokes drowning in a pool ( read about it at your leisure ). Nevertheless, I learned that this rule is not obvious to the child.
Not long ago, on vacation with the whole family, I told the child that I was going to eat, and began to put food in the plate. It was at that moment that he burst into tears, shouting at me (“Don't eat, mom !!!”), clapping my hand and knocking the fork out of my hand.
When I managed to pick my jaw from the floor, I tried to understand whether my child was a monster who does not want his mother to eat and only two days later, putting him to bed, I understood where it all came from.
Our daily routine was this: I came back from work, fed the child, bathed, put to bed, and then, finally, ate. As a result, each time, putting the baby in bed and reading him a book, I ended the evening with the words: "Mom is going to eat." And after that, she left the child alone for the next 10–12 hours of sleep. Thanks to this correlation, his mind created a causal link: “if mom was going to eat, she would soon leave me alone.” Oh…
Here, my mother's task is to change this pattern so that the son does not understand the connection between food and separation. In Data Scientist, if the machine chooses the wrong sign or reason, the main task is to admit the error.
Let us return to the unsuccessful use of AI by Amazon as a recruitment tool. The 10-year sample of data that they used to evaluate candidates chose men more readily because "most resumes were historically received from men, which reflects male dominance in the technology industry."
And Amazon AI says: “Hey, guys, most of the applicants are men, so you have to hire men, and if the resume was sent by a woman, then I throw it away, because this is an anomaly.”
No, AI. It just makes you a sexist. And it is here that kids have an advantage (and adults, let's be optimistic): it's never too late to learn not to be a sexist.
There is not a single parent who would call the upbringing of children an exceptionally pleasant and easy matter (and if someone says so, he will lie impudently). Each parent must constantly ask themselves the question “what does the baby learn?” And adapt to its constantly evolving neural network.
Data Scientists to some extent bear the same responsibility.
It is impossible, hiring or studying Data Scientist, to expect that all work will be associated only with programming. This is tantamount to the expectation that a happy adult can be raised from a child, training him like a dog, with orders to "sit" and "roll over" all of his childhood. By experience, it works until the child is 6 months old - and as soon as he learned to roll himself, it’s time to teach him human things.
I'll just leave a grinning smiley here. If you are a parent, you already know everything.
If you, like me, raised children and at the same time taught the algorithm, then most likely you compared these two processes. And even if you are not fond of artificial intelligence, but know a lot about children, welcome to the wonderful world of educating machines ... oops, machine learning.
When writing this article, no child was hurt. I just, like any parent, spend many hours watching my child learn about the world, and am surprised at his behavioral patterns. Just as a data scientist does, observing the results of train / test samples ( training sample of data for learning the algorithm / result of the algorithm on the new data - ed. ).
"At first he is stupid like a cork"
This is a quote from Jim Stern , author of Artificial Intelligence for Marketing: Practical Applications, from a lecture on machine learning — not about children (I love children!).
')
The essence of machine learning is to actually teach a machine to perform a specific task - just as parents dream of teaching children to collect dirty laundry and put it in the washing machine while mom and dad are relaxing on the couch (admit, tried?).
However, the main difference is that when a child is asked to wash clothes, he already knows what the clothes look like; able to walk, grab, pull and fold - these actions he learned through other events in his young life.
So where is the key that ultimately opens up the opportunity to be lazy on the couch while laundry is magically erased? Context. We give children examples: we show how to perform each step and thank them when they do everything right - because we love them.
Machine learning is pretty much the same thing, except that the “virtual child” still has the experience of a newborn with the abilities of an older baby. Therefore, it is necessary to begin an explanation from scratch: these five pieces, like sausages, which stick out at the end of a long, sausage-like stick, are fingers, a hand and a palm. Only then should you be shown how to perform the necessary actions with their help - grab and pull. The data set that you give to the machine is all you need to get started, but also everything that exists in the world for it. What it does not yet possess is ...
… common sense
Usually, people successfully distinguish between men and women. Liam, my son, also copes well with this — and I did not give him a large set of marked data at the entrance. I didn’t sit with him in the park and didn’t point at people, saying “man, man, woman, man, woman” - because, let's be honest, it would be strange. Yes, and not necessary. The car lacks the luxury of common sense, which the child has, and which is used already at the first collision with a new concept.
By common sense, I mean:
The ability to make the right decisions and make the right assumptions based on logical thinking and accumulated experience - Wiktionary
Of course, when a child decides to jump his head into the ground from a height, we quite reasonably doubt his common sense. Nevertheless, it exists and allows children to learn from all their experiences. At the same time, no one is clearly broadcasting to them how to learn to distinguish between men and women.
Explaining the topic of AI to non-dascientists, I like to use an analogy. The child only needs a little observation, a few examples and a couple of corrections to learn how to say "Mr." or "Mrs." And to train the machine to do the same, you need to give her thousands of images. The lack of common sense is probably the number one reason why cars are not yet ready to take over the world.
Rules and oddities
Liam does strange things, like eating a hot dog, holding his ends and biting in the middle. The standard reaction is to tell him: “Liam! They don’t do that! ” But then I hold back and think that the out-of-box solution is not the best that I can give it. Although when he tries to hold a spoon with his nostrils, one really has to set the limits for acceptable behavior at the table.
This is the great similarity between kids and cars - they are free from social norms and prejudices (or Bayes - from the English. Bias). And this is also the difference between the parents and the date of the Scientists. Babies need to be given a set of values ​​and social norms from which they will build their experience. "Good boundaries", let's call them that. As a data scientist, you are most likely playing the opposite role. The machine should be free from your own norms and prejudices. Bias or bias in algorithms is very dangerous.
Everyone loves gossip and HYIP. For example, Amazon's AI recruiter is a sexist ( Amazon's recruiting AI is sexist ), or the "enhancement" filter FaceApp is a racist ( FaceApp's "hot" filter is racist ). This is a good way to explain to non-data scientists that the role of the scientist and the scientist's date largely boils down to preventing bias and creating as ethical an algorithm as possible.
Correlation and causation
Image source xkcd
Correlation does not imply a causal relationship. And Nicolas Cage is not a monster who provokes drowning in a pool ( read about it at your leisure ). Nevertheless, I learned that this rule is not obvious to the child.
Not long ago, on vacation with the whole family, I told the child that I was going to eat, and began to put food in the plate. It was at that moment that he burst into tears, shouting at me (“Don't eat, mom !!!”), clapping my hand and knocking the fork out of my hand.
When I managed to pick my jaw from the floor, I tried to understand whether my child was a monster who does not want his mother to eat and only two days later, putting him to bed, I understood where it all came from.
Our daily routine was this: I came back from work, fed the child, bathed, put to bed, and then, finally, ate. As a result, each time, putting the baby in bed and reading him a book, I ended the evening with the words: "Mom is going to eat." And after that, she left the child alone for the next 10–12 hours of sleep. Thanks to this correlation, his mind created a causal link: “if mom was going to eat, she would soon leave me alone.” Oh…
Here, my mother's task is to change this pattern so that the son does not understand the connection between food and separation. In Data Scientist, if the machine chooses the wrong sign or reason, the main task is to admit the error.
Let us return to the unsuccessful use of AI by Amazon as a recruitment tool. The 10-year sample of data that they used to evaluate candidates chose men more readily because "most resumes were historically received from men, which reflects male dominance in the technology industry."
And Amazon AI says: “Hey, guys, most of the applicants are men, so you have to hire men, and if the resume was sent by a woman, then I throw it away, because this is an anomaly.”
No, AI. It just makes you a sexist. And it is here that kids have an advantage (and adults, let's be optimistic): it's never too late to learn not to be a sexist.
Both parenthood and Data Science are about people.
There is not a single parent who would call the upbringing of children an exceptionally pleasant and easy matter (and if someone says so, he will lie impudently). Each parent must constantly ask themselves the question “what does the baby learn?” And adapt to its constantly evolving neural network.
Data Scientists to some extent bear the same responsibility.
It is impossible, hiring or studying Data Scientist, to expect that all work will be associated only with programming. This is tantamount to the expectation that a happy adult can be raised from a child, training him like a dog, with orders to "sit" and "roll over" all of his childhood. By experience, it works until the child is 6 months old - and as soon as he learned to roll himself, it’s time to teach him human things.
So what is easier - to raise a child or raise a car?
I'll just leave a grinning smiley here. If you are a parent, you already know everything.
From the Editor
- Online course " Machine learning "
- Profession " Data Scientist "
- Online course " Big Data "
Source: https://habr.com/ru/post/457508/
All Articles