Machine Learning models “learn” kaise karte hain?

Question

Today, ML is everywhere — recommendations, fraud detection, chatbots, automation.
But here’s the interesting part:

Machine Learning doesn’t work like traditional rule-based programming.

We feed it data, it gradually improves over time…
But what does “learning” actually mean for a machine?

Some questions worth thinking about:

Does the model simply memorize patterns?

Or does it really understand anything?

When it predicts wrong, how does it correct itself?

Is its learning similar to the human brain — or completely different?

Many people talk about “training a model”…
but very few can explain how that learning process truly happens.

What’s your take — how do you think ML models learn?
Drop your thoughts below

Answer 1

There are 3 types of machine learning
1) supervised learning - labels
2) unsupervised learning - clustering
3) reinforcement learning - try and error

We choose the models based on the type of dataset and hence while training we use lots of layers accordingly and we choose the best inputs and outputs data by feature engineering of converting dataset into machine understanding like one hot encoding along with we use different optimization, loss functions and metrics and also most importantly activation functions which decides how to outputs will be.

Optimization helps in updating weights which helps increasing accuracy and loss functions helps to reduce the loss to make reduce the error along with activation functions.

In overall neural network try to mimic the working of the human brain but it predicts based on the probability and not like a human brain it's just too good to guess rather then understanding like the human brain.

Answer 2

Are you guys just using A.I. to post questions and answer them?