Adversarial machine learning

Can you imagine a world where machines learn without seeing examples first?

Imagine you're teaching a child to recognize different fruits without showing them every type. They might struggle to identify a fruit they've never seen before.

Zero-shot learning is like teaching that child to guess what fruit a picture is of without having seen that exact fruit before. It's about making educated guesses based on what they've learned from other fruits.

Example

If you've seen apples, bananas, and oranges, zero-shot learning helps you guess what a picture of a grape might look like.

Remember this

Zero-shot learning enables machines to recognize and categorize new things they've never encountered before, expanding their knowledge beyond pre-labeled examples.

Text adapted from Wikipedia, licensed under CC BY-SA 4.0.

Related concepts

Prompt engineering

The GenAI model learns tasks from examples in the prompt

weight tying does in language models: shares embedding and output projection matrices

Ever wonder how machines understand the sequence of words in a sentence?

soft targets carry more information than hard labels: they encode class similarities

Why do some learning methods need to explore more than others?

classifier-free guidance does: interpolates between conditional and unconditional generation

"Classifies samples as either conditioned or unconditioned, guiding generation towards desired outcomes."

Speech recognition

How do machines understand our spoken words?

[CLS] pooling does: uses the first token's embedding as the sentence representation

CLS pooling: uses the first token's embedding as the sentence representation

Swipe through more Machine Learning concepts

Open Pocket Polymath