Overview
Explore innovative optimization techniques and machine learning concepts in this engaging ACCU 2018 conference talk. Dive into the world of hill climbing algorithms, turtle graphics, and paper bag programming challenges. Learn how to navigate rectangular and crumpled paper bags using various methods, including gradient descent, stochastic gradient descent, and simulated annealing. Discover the connections between these techniques and neural networks, deep learning, and other machine learning applications. Gain insights into particle swarm optimization, gradient-based approaches, and their potential pitfalls. Whether you're a curious beginner or an expert seeking a fresh perspective, this talk provides a unique and accessible introduction to fundamental concepts underpinning many areas of artificial intelligence and optimization.
Syllabus
Intro
Overview
Context: Optimisation
Code your way out of a paper bag
Dragons! T-shirt!
Spirangle code
How did the turtle get into the paper bag?
What shaped paper bag?
Hill walking
Crumpled bag
Try random stuff
Actual annealing
Simulated annealing
Exponential curve
Swarm!
Particle Swarm Optimisation, PSO
PSO Move
PSO Swarm
Update velocity
What's a gradient?
Gradients c.f. hill climbing
Gradients: problems
Neural Networks
Backpropagation
What did we learn?
Taught by
ACCU Conference