28.01.18 - 02.02.18, Seminar 18052

Genetic Improvement of Software

The following text appeared on our web pages prior to the seminar, and was included as part of the invitation.

Motivation

Recent work on Genetic Improvement (GI) has covered automatic bug repair and improving both functional and non-functional properties of existing software code.

Non-functional improvements have included radical speeds ups and (particularly for low resource computational motes and mobile computing) reducing energy consumption, and memory footprint.

In addition to automatic bug fixing, functional improvements have included growing and grafting in new functionality, automatic porting to new hardware (often parallel hardware such as GPU and SSE vector instructions), automatic tuning and transplanting functionality from existing often open source repositories like GitHub.

These are exciting times but as highlighted by the recent Dagstuhl Seminar on Automated Program Repair (17022), there is a risk that lessons learnt in one area will only be exploited by that area. Therefore this Dagstuhl Seminar will draw participants from all corners of GI to contribute their thoughts on experiences, tools, datasets and benchmarks, validation, theoretical analysis and the ways forward.

What will programming look like in ten years’ time? How will GI in 2018 be thought of in 2038 or in 2050?

The seminar will focus on various genetic improvement approaches and related areas where software has been reused for purpose of automated software improvement. The proposed topics of discussion include:

  • software mutational robustness
  • program repair
  • non-functional software property improvement
  • search-based approaches for software improvement
  • data mining for GI

License
Creative Commons BY 3.0 Unported license
Stephanie Forrest, William B. Langdon, Claire Le Goues, and Justyna Petke