Genetic Improvement of Source Code
- Speaker: Dr. William B. Langdon, Department of Computer Science, University College London.
- Date: Wednesday, 8 October 2014 from 16:30 to 17:30
- Location: Room 151, Birkbeck Main Building
Genetic programming can optimise programs including evolving test benchmarks, search meta-heuristics, protocols, composing web services, improving hashing and memory allocation, redundant programming and even automatically fixing bugs. There are many ways to balance functionality with resource consumption (eg time, memory, energy). But a human programmer cannot try them all. Similarly the easy to write software may not give the best floating point accuracy or give the best trade off between performance and solution quality. Also the Pareto optimal tradeoff may be different on each hardware platform and it may be dynamic, e.g. as usage changes. Possibly GP could automatically suggest a different balance between multiple objectives for each new market. Recent results include substantial speed up by generating a new version of a program for a special case.
Dr. Langdon gained his PhD at UCL after a career in real-time industrial control software and consulting. After positions in universities and research institutes at home and abroad, he has returned to UCL where he is applying genetic programming to optimising software. He has written 3 books on genetic programming, including "A Field Guide to Genetic Programming", which can be downloaded for free. He also maintains the genetic programming bibliography.
Home: http://www.cs.ucl.ac.uk/staff/W.Langdon/ project: