Machine Learning
Learning to Spot a Bad Mango
6 min read
Learning to Spot a Bad Mango
Your grandmother can tell if a mango is sweet just by smelling it — and she never read a chemistry textbook.
She's smelled and tasted thousands of mangoes over 40 years. Her brain built a model: this smell + that colour + this firmness = sweet mango. No one taught her the formula. She discovered it from experience. Machine Learning is exactly this — giving a computer thousands of examples so it builds its own formula.
In Plain English
Machine Learning is a way of building AI where instead of programming every rule, you feed the system thousands of examples and let it find the patterns itself. The more examples, the better it gets.
The Technical Picture
Machine Learning is a subset of AI where algorithms improve their performance on tasks through experience (training data) without being explicitly programmed. The model learns a mapping from inputs to outputs by minimising prediction error across a training dataset.
Real-World Examples
- Spam filters learning which emails are junk from millions of examples
- Netflix recommending shows based on your watch history
- Swiggy estimating delivery time from thousands of past orders
Machine Learning = giving computers examples instead of instructions.