This post is my live blog from KCDC. For more, see the 2021 KCDC live blog TOC

Speaker: Randall Koutnik

Twitter @rkoutnik

————————-

Genetic Algorithm

• Machine learning
• Represent problem as an array of things
• Create a weighting to represent the problem as a number
• Showed scale of evolving solutions and when processing exceeds laptop processing capacity
• Keep changing best on best one until that point
• Fiinite state machines – also like an array of booleans representing choices. Generate possibilities and see which one works. Use fitness function to see which are most successful
• Get jitter in high score as average goes up because randomness
• Determine if need perfect answer vs good enough answer

Examples

• Generating counterfeit art
• Eater game demo where they learn to eat plants
• Calcu-lords game to determin which cards are best

My take

I learned what a genetic algorithm is And was nice to see some code. The examples were al fun. Hard to take notes on but I enoyed the presentation.