Think Together · Start with a Short Story
The girl is teaching a robot to tell fruits apart. Saying “Be a little smarter” is not enough. She shows it examples so the system has something to learn from.
Lay Out the Process
- Prepare examplesData such as photos
- TrainCompare and adjust
- Keep the modelSave the parameters it learned
- Try new examplesCheck whether it can use what it learned
Machine learning uses patterns in data to improve how well a model performs a task.
Data can be pictures, text, sounds, or numbers. Different tasks need different kinds of data. For a classification task with labels, the labels provide the expected answers.
Learning is not just putting all the materials unchanged into a drawer. Training usually changes the model's parameters, though a model may also memorize some examples.
Now You Try
If a fruit classifier is trained only with red apples, can it be expected to recognize bananas?
Ready? Open this to see the explanation
No. The training examples do not include bananas and do not cover the variety of fruits. It needs more suitable examples and independent testing.
One Idea to Take Away
Machine learning uses patterns in data to improve how well a model performs a task.
Try covering the diagram above and explaining each step to someone in your family. If one step is hard to explain, return to that card and look again.