Dōmo arigatō, Mr. Roboto: Machine Learning with Ruby
summary
A Ruby implementation trains a neural network on the MNIST dataset using the RubyFann gem and serves a React front end with a Sinatra backend; the model maps 24×24 images into a 576 inputs / 300 hidden / 10 outputs network, trains on ~60k examples and tests on 10k, and reaches about 93.3% test accuracy while often misclassifying ambiguous digits in the live demo; training uses feed‑forward + backprop, early stopping and MSE thresholding to avoid overfitting, and the code is available as a runnable Ruby MNIST app.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:06Introduo
- 2:27Eric Weinstein
- 4:08Show of Hands
- 5:38Pattern Recognition
- 6:46Classification and Regression
- 7:57Features and Labels
- 9:56Vector of Intensity
- 10:17Neural Network
- 10:44Perceptron
- 13:28Neural Networks
- 15:48Data
- 17:20Sample Data
- 22:32Ruby Fan
- 24:47Demo
- 29:23Summary
- 31:51Inspirational
- 33:29Questions
related talks
51mMachine Learning with Ruby
29mDiscover Machine Learning in Ruby
28mNeural Networks with RubyFANN
29mTest Driven Neural Networks with Ruby
29mMachine learning for front-end developers
35mIntelligent Ruby: Getting Started with Machine Learning
29mHow I entered the machine learning world
43mA Gentle Introduction to Neural Networks