Big data driven genetic improvement for maintenance of legacy software systems
Bibliographic record
Abstract
Software is vital to modern life, yet much of it is old and suffers from bit-rot. There are not and never will be enough software experts to keep it all up to date by hand. Instead we suggest combining data driven learning with evolutionary search to maintain computer systems. @RE: <1>N. Alshahwan. Industrial experience of genetic improvement in Facebook. In J. Petke, S. H. Tan, W. B. Langdon, and W. Weimer, editors, GI-2019, ICSE workshops proceedings, page 1, Montreal, 28 May 2019. IEEE. Invited Keynote. <2>W. Banzhaf, P. Nordin, R. E. Keller, and F. D. Francone. Genetic Programming - An Introduction; On the Automatic Evolution of Computer Programs and its Applications. Morgan Kaufmann, San Francisco, CA, USA, Jan. 1998. <3>N. Hansen and A. Ostermeier. Completely derandomized self-adaptation in evolution strategies. Evolutionary Computation, 9(2):159--195, Summer 2001. <4>S. Haraldsson, A. Brownlee, and J. R. Woodward. Computers will soon be able to fix themselves - are IT departments for the chop? The Conversation, page 3.29pm BST, Oct. 12 2017. <5>S. O. Haraldsson, J. R. Woodward, A. E. I. Brownlee, and K. Siggeirsdottir. Fixing bugs in your sleep: How genetic improvement became an overnight success. In J. Petke, D. R. White, W. B. Langdon, and W. Weimer, editors, GI-2017, pages 1513--1520, Berlin, 15--19 July 2017. ACM. Best paper. <6>M. Harman and B. F. Jones. Search based software engineering. Information and Software Technology, 43(14):833--839, Dec. 2001. <7>F. Hutter, H. H. Hoos, K. Leyton-Brown, and T. Stuetzle. ParamILS: An automatic algorithm configuration framework. JAIR, 36:267--306, 2009. <8>G. Kendall. Evolutionary computation has been promising self-programming machines for 60 years - so where are they? The Conversation, page 8.54am BST, Mar. 27 2018. <9>J. R. Koza. Genetic Programming: On the Programming of Computers by Natural Selection. MIT press, 1992. <10>W. B. Langdon. Genetic improvement of programs. In R. Matousek, editor, 18th International Con- ference on Soft Computing, MENDEL 2012, Brno, Czech Republic, 27--29 June 2012. Brno University of Technology. Invited keynote. <11>W. B. Langdon and M. Harman. Optimising existing software with genetic programming. IEEE Transactions on Evolutionary Computation, 19(1):118--135, Feb. 2015. <12>W. B. Langdon, B. Y. H. Lam, J. Petke, and M. Harman. Improving CUDA DNA analysis soft- ware with genetic programming. In S. Silva, A. I. Esparcia-Alcazar, M. Lopez-Ibanez, S. Mostaghim, J. Timmis, C. Zarges, L. Correia, T. Soule, M. Giacobini, R. Urbanowicz, Y. Akimoto, T. Glasmach- ers, F. Fernandez de Vega, A. Hoover, P. Larranaga, M. Soto, C. Cotta, F. B. Pereira, J. Handl, J. Koutnik, A. Gaspar-Cunha, H. Trautmann, J.-B. Mouret, S. Risi, E. Costa, O. Schuetze, K. Kraw- iec, A. Moraglio, J. F. Miller, P. Widera, S. Cagnoni, J. Merelo, E. Hart, L. Trujillo, M. Kessentini, G. Ochoa, F. Chicano, and C. Doerr, editors, GECCO '15: Proceedings of the 2015 Annual Conference on Genetic and Evolutionary Computation, pages 1063--1070, Madrid, 11--15 July 2015. ACM. <13>W. B. Langdon and R. Lorenz. Improving SSE parallel code with grow and graft genetic programming. In J. Petke, D. R. White, W. B. Langdon, and W. Weimer, editors, GI-2017, pages 1537--1538, Berlin, 15--19 July 2017. ACM. <14>W. B. Langdon and J. Petke. Evolving better software parameters. In T. E. Colanzi and P. McMinn, editors, SSBSE 2018 Hot off the Press Track, volume 11036 of LNCS, pages 363--369, Montpellier, France, 8--9 Sept. 2018. Springer. <15>W. B. Langdon, J. Petke, and R. Lorenz. Evolving better RNAfold structure prediction. In M. Castelli, L. Sekanina, and M. Zhang, editors, EuroGP 2018: Proceedings of the 21st European Conference on Genetic Programming, volume 10781 of LNCS, pages 220--236, Parma, Italy, 4--6 Apr. 2018. Springer Verlag. <16>C. Le Goues, M. Pradel, and A. Roychoudhury. Automated program repair. Communications of the ACM. To appear. <17>M. Orlov. Evolving software building blocks with FINCH. In J. Petke, D. R. White, W. B. Langdon, and W. Weimer, editors, GI-2017, pages 1539--1540, Berlin, 15--19 July 2017. ACM. <18>L. Perez Caceres, M. Lopez-Ibanez, H. Hoos, and T. Stuetzle. An experimental study of adaptive capping in irace. In R. Battiti, D. E. Kvasov, and Y. D. Sergeyev, editors, Learning and Intelligent Optimization - 11th International Conference, LION 11, Nizhny Novgorod, Russia, June 19--21, 2017, Revised Selected Papers, volume 10556 of Lecture Notes in Computer Science, pages 235--250. Springer, 2017. <19>J. Petke. Constraints: The future of combinatorial interaction testing. In 2015 IEEE/ACM 8th International Workshop on Search-Based Software Testing, pages 17--18, Florence, May 2015. <20>J. Petke, S. O. Haraldsson, M. Harman, W. B. Langdon, D. R. White, and J. R. Woodward. Genetic improvement of software: a comprehensive survey. IEEE Transactions on Evolutionary Computation, 22(3):415--432, June 2018. <21>J. Petke, M. Harman, W. B. Langdon, and W. Weimer. Specialising software for different downstream applications using genetic improvement and code transplantation. IEEE Transactions on Software Engineering, 44(6):574--594, June 2018. <22>R. Poli, W. B. Langdon, and N. F. McPhee. A field guide to genetic programming. Published via http://lulu.com and freely available at http://www.gp-field-guide.org.uk, 2008. (With contri- butions by J. R. Koza). <23>E. Schulte, S. Forrest, and W. Weimer. Automated program repair through the evolution of assembly code. In Proceedings of the IEEE/ACM International Conference on Automated Software Engineering, pages 313--316, Antwerp, 20--24 Sept. 2010. ACM. <24>E. Schulte, W. Weimer, and S. Forrest. Repairing COTS router firmware without access to source code or test suites: A case study in evolutionary software repair. In W. B. Langdon, J. Petke, and D. R. White, editors, Genetic Improvement 2015 Workshop, pages 847--854, Madrid, 11--15 July 2015. ACM. Best Paper. <25>J. R. Woodward, J. Petke, and W. Langdon. How computers are learning to make human software work more efficiently. The Conversation, page 10.08am BST, June 25 2015. <26>K. Yeboah-Antwi and B. Baudry. Online genetic improvement on the java virtual machine with ECSELR. Genetic Programming and Evolvable Machines, 18(1):83--109, Mar. 2017.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.006 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".