Understanding Program Evolution via Data Visualisation
Automatically designing and generating algorithms has long been a dream of computer scientists. Genetic programming (GP) has been one of the powerful approaches to (partially) achieving this goal. Inspired by biological evolution, GP evolves computer programs tailored to specific tasks such as financial trading, credit scoring, production planning, and project management. Although GP has shown its success in many application areas, researchers have not fully understood how the algorithm works due to the lack of analysis tools for studying the emergent complexity of evolutionary dynamics.
This project aims to develop new genetic-based machine learning techniques to efficiently explore powerful programs for complex computational problems and to provide useful insights for the dynamic evolutionary process by means of data visualisation and incremental learning. The new techniques not only allow users and decision makers to understand how programs or models are evolved but also let them intervene and guide the evolutionary process.
- Su Nguyen, Mengjie Zhang, Damminda Alahakoon, Kay Chen Tan, “Visualizing the Evolution of Computer Programs for Genetic Programming”, 2018, accepted by IEEE Computational Intelligence Magazine (supplementary document)
- Su Nguyen, Mengjie Zhang, Kay Chen Tan. “Adaptive Charting Genetic Programming for Dynamic Flexible Job Shop Scheduling”, 2018, Proceedings of the Genetic and Evolutionary Computation Conference. 1159-1166.