Using JavaScript to Teach Machines How to Do Cool Things

What's the difference between machine learning and artificial intelligence? Can we use JavaScript to implement it? How easy is it to teach these machines to do cool things?

In this talk, I cover:

• The definition of machine learning and how the concept differs from artificial intelligence
  • "Machine learning paves the path to artificial intelligence"
• Supervised vs unsupervised learning
• Some basic machine learning types:
  • Liner Regression
  • Classification
  • Clustering
• Discrete vs continuous data, and what it tells us about the learning types
• A demo of how to improve a "dumb" Twitter bot using only a standard input (tweets)
• Bias-Variance Tradeoff
• The definition of a neural network and a couple of its types: artificial vs convolutional
• Why use JavaScript to implement these concepts?
  • What are the limitations to using JS (compared to using something more suited to handle large data sets)?

Further Reading

• Step-by-Step Back propagation Example
• Anomaly Detection in JS
• Creating a Neural Network in 30 Lines of JS Code
• Neural Networks and Deep Learning
• Machine Learning in Other Languages: JavaScript
• Synaptic.js: Architecture-free neural network library for node.js and the browser
• Brain.js: GPU-accelerated neural network library for node.js and the browser
• The Tensorflow Playground

Seen At

Zeit Day on April 28th, 2018

• 🎥 Recorded presentation

Slides

→ 📺 Live view

→ 📑 Exported PDF