Turtles! Hill climbing! Hammers! Paper bags!
summary
Live demos use Python turtle graphics to implement hill climbing, simulated annealing, particle swarm and gradient-descent variants with visible step-size, cooling schedule and parameter choices; you see one-dimensional quadratic, cosine and crumpled paper‑bag examples that expose local minima, oscillation and convergence failures. Code-level decisions include deterministic vs random moves, learning rate schedules, velocity updates for swarms and simple feed‑forward neural net weight tweaks. Rough working demos sometimes converge, sometimes miss the global minimum — useful if you want practical failure modes and tunable examples.
AI-written summary · about
chapters (16)
- 0:00Introduction to talk
- 0:30Turtle graphics and machine learning
- 4:04Introducing Python turtle
- 5:15Coding out of paper bag
- 8:31Hill climbing basics
- 14:13Problems with hill climbing
- 18:58Simulated annealing explained
- 27:24Implementing simulated annealing
- 34:48Particle swarm optimization
- 45:55Testing machine learning algorithms
- 47:00Gradient descent and neural networks
- 50:31Neural networks overview
- 55:533D turtle graphics challenges
- 56:13Other applications of turtle graphics
- 57:16Book and resources
- 59:40Q&A and closing remarks
related talks
01h 15mAn Introduction to Swarm Intelligence Algorithms
01h 26mHow to Evolve Your Way Out of a Paper Bag
30mEvolutionary Algorithms Perfecting the Art of "Good Enough"
50mOptimal control, no! Than what? (...)
56mAn Introduction to Reinforcement Learning - Snake Your Way Out of a Paper Bag
01h 31mWorkshop on developing Reinforcement Learning
05mGenerative Drawing with Python Turtle Graphics
54mAmazing Algorithms for Solving Problems in Software