Math for ML
Vectors, dot products, cosine similarity, derivatives and gradients, only the parts the rest of the track uses.
How much math you need
Less than people fear, more than zero. For this track you need four ideas:
- Vectors: lists of numbers that represent things.
- Dot product and cosine similarity: how aligned two vectors are.
- Matrices: a table of numbers that transforms vectors.
- Gradients: which direction to nudge numbers to reduce an error.
Everything else (transformers, embeddings, training) is these four, repeated at scale.
1 of 7