Classification
Table of Contents
Classification is a task in Machine Learning.
1. Architecture
1.1. Logit -> Softmax
1.2. Embedding -> Probability
Using Cosine similarity with Class Prototype
2. Classification is easier than regression for NNs
The first fact is that neural networks (dense, recurrence, and graph) are more often used as classifiers and not as approximation functions for the ranges of floating point or integer numbers as they are used by neural acceleration. [Neural Acceleration of Graph Based Utility Functions for Sparse Matrices.pdf pg. 13]
