Theory of Intelligence
MACHINE LEARNING•Supervised Learning•Classification•1.1 Training a Binary Classifier

Introduction

Meet the MNIST handwritten-digit dataset, train a first yes/no classifier, and understand the difference between predicting labels and predicting probabilities.

Introduction

In the Foundations tab we predicted numbers (house prices) — that's regression. Now we switch to classification: predicting a category.

The simplest kind is binary classification: there are only two possible answers.

  • Is this email spam? yes / no
  • Will this customer cancel? yes / no
  • Is this transaction fraud? yes / no
  • Is this picture the digit 5? yes / no

We'll learn with the most famous practice dataset in machine learning — handwritten digits — and build a "5-detector": a model that looks at an image and answers "is it a 5 or not?"

Discuss theoretical concepts, mathematical formulas, and architectural foundations.

Supports inline code with `backticks`
No comments yet. Be the first to start the discussion!