Approximating the Nondominated Front Using the Pareto Archived Evolution Strategy
Evolutionary Computation
Multiobjective genetic algorithms for multiscaling excited state direct dynamics in photochemistry
Proceedings of the 8th annual conference on Genetic and evolutionary computation
An ontology for a Robot Scientist
Bioinformatics
Improving PSO-Based multi-objective optimization using crowding, mutation and ∈-dominance
EMO'05 Proceedings of the Third international conference on Evolutionary Multi-Criterion Optimization
Closed-loop evolutionary multiobjective optimization
IEEE Computational Intelligence Magazine
| Hi-index | 0.00 |
A Robot Scientist is a physically implemented system that applies artificial intelligence to autonomously discover new knowledge through cycles of scientific experimentation. Additionally, our Robot Scientist is able to execute experiments that have been requested by human biologists. There arises a multi-objective problem in the selection of batches of trials to be run together on the robot hardware. We describe the use of the jMetal framework to assess the suitability of a number of multi-objective metaheuristics to optimise the flow of experiments run on a Robot Scientist. Experiments are selected in batches, chosen in order to maximise the information gain and minimise the use of resources. The evolutionary multi-objective algorithms evaluated here perform well in finding solutions to this problem, either finding a long, fairly efficient Pareto optimal front, or a shorter, highly efficient Pareto optimal front.