This document provides an overview of key concepts in linear algebra that are relevant for deep learning, including:
- Vectors are 1-D arrays of numbers that can be represented as points in space. Matrices are 2-D arrays where each element is identified by two indices. Tensors generalize this to arrays with more than two axes.
- Operations like matrix multiplication and transposition are defined. The dot product of two vectors or matrices is also introduced.
- Systems of linear equations can be represented using matrix-vector notation. Matrix inversion allows solving such systems, though it is numerically unstable.
- Norms are functions that measure the "size" of vectors and are useful in machine learning,