Amazing Algorithms for Solving Problems in Software
summary
Simulations show a 20-node best-path problem solved with bee and ant colony heuristics, a 40-firefly swarm optimizing the Mikhailovich test surface, and an amoeba optimizer (size 13, max epochs ~70) used to train a simple ML model with 16 inputs on ~430 records, reaching 95.7% accuracy. Code is plain C# run in Visual Studio and available on GitHub; live demos sit at amazingalgorithms.azurewebsites.net. Expect nondeterministic results, tunable parameters, and no optimality guarantees from these bio-inspired methods.
AI-written summary · about
chapters (27)
- 0:00Introduction
- 0:40Favorite physicists and mathematicians
- 1:17Open source projects
- 1:33Liquid Victor
- 1:47GiveCamp
- 3:43Agenda
- 4:17Best Path
- 6:15Bee Colony
- 10:06Bee Colony Optimization
- 16:51Reducing Costs
- 17:11Mikhailovich Function
- 17:42Firefly Optimization
- 20:26Difficulties
- 22:49Amoeba
- 24:06Flowchart
- 26:37Amoebas
- 27:35Linear regression
- 30:04Error function
- 31:02Prediction model
- 33:29Sigmoid function
- 34:13C Code
- 39:39Training the Model
- 41:40Predict Method
- 43:11Results
- 45:14Bioinspired algorithms
- 46:48Best path algorithms
- 48:22Resources
related talks
01h 15mAn Introduction to Swarm Intelligence Algorithms
07mSwarm Intelligence Algorithms - Session Interview With Frances Buontempo
50mOptimal control, no! Than what? (...)
30mEvolutionary Algorithms Perfecting the Art of "Good Enough"
57mSolutions That Evolve: Building Self-Improving Systems with Genetic Algorithms
34mAlgorithms to live
57mWhat is Swarm Intelligence? - Swarm AI Explained
46mTight Genes: Intro to Genetic Algorithms