What machine learning really is! (own realization)

Hey, How are you? Hope all is well.
Machine Learning
There are too many formal definitions of ML and all of them are available on google. So, if you want to know about it, you may make a google search that’s it. :grinning:

Now it’s time to share my own realization:
Consider the computer as a dumb human being who doesn’t know anything. Unlike a human being, it just can not do anything else without the explicit instruction of the human. The instruction/program can be compared to the brain of a human. If somehow we can train the dumb machine to act rationally, we can argue that the machine is learning and we people are the successful teacher. Right?

When a new baby born, it doesn’t know anything else about the world except crying. With the passage of time, it learns about the environment and takes the decision according to the previous experience. With growing up, the baby becomes a rational being.
The machine learning algorithms also follow the same process to learn. We feed the machine with previous data as previous experience. ML algorithms ensure the best decision to best fit with future data.
And we the people who will use the algorithm properly to learn the Machine/Computer.

2 Likes

Hello @Zubair,
You have a good point. But, when we dive deeper we will see that ML is nothing more than different types of gradient descent functions trying to find such a function that minimizes the errors of the dataset.
Initially, Machine Learning may look like a baby trying to learn more stuff with more datasets. But, under the hood, each of the ML techniques is trying to find parameters of an unknown function that best represents the data. The model is as good as its dataset. The model will not be able to generalize from what it learned to make predictions on unknown categories of data. We could say ML or current AI techniques still use conventional calculus to find a better function that describes any given dataset.
Image we have the following dataset of house prices: 1 room - $105k, 2 rooms - $205k, 3 rooms - $305k. We immediately see pattern for this small dataset, but we can not see the pattern if the dataset has millions of observations. Then, when we use an ML model and it comes up with a function price = 5000 + 100,000*rooms. This function is a very good representation of the dataset. In general, any ML technique tries to find such a function (e.g. price function) that minimizes some error function whether be it an image, voice or other types of datasets. Sure, the function can get very complex and sometimes very hard to understand (like Neural Networks), but all of the models are still doing similar calculations to obtain a complex function. As of today, these techniques are very powerful because we have access to tons of data. However, we might need a very different approach to reach the Artificial General Intelligence (AGI) level that can think like an average human being.
I recommend you to watch videos by Lex Fridman who has interviewed many top scientists in AI, Computer Science, and Neuroscience fields. This specific lecture describes the current state of AI: https://youtu.be/53YvP6gdD7U

5 Likes

hey @saidakbarp,
thanks for sharing your valuable information. I know it. But I have written the article to share my thoughts, nothing else. These thoughts came across my mind when I came to know about Machine Learning, probably a year back.

1 Like

This is very helpful, thank you!

1 Like

Hi,

Machine studying is a program of artificial intelligence (AI) which supplies systems the capability to automatically learn and improve in expertise without being explicitly programmed. Machine studying concentrates on the evolution of computer programs which could get information and use it in order to learn for themselves.

Machine Learning is a comprehensive region of AI concentrated within the design and production of an algorithm that defines and finds patterns which exist in data supplied as input. This invention will specify a replacement strategy of regulating and managing a farther, particularly businesses.