News – Page 2 – UCD Natural Computing Research & Applications Group

NCRA @ ACM GECCO 2021

10 years on from hosting GECCO 2011 in Dublin, Ireland, Mark Connor and Michael O’Neill presented an approach to optimising the Banister Dose-Response Fitness-Fatique model used in athlete training management (arxiv draft) at GECCO 2021

DSSV-ECDA 2021

Michael presented an invited talk on “Grammars, Evolutionary Computation and Intepretability” in a special session on Grammatical Evolution at DSSV-ECDA 2021 (7-9 July, 2021). The session was organised by Andreas Geyer-Schulz who spoke on “Architectural Design of a Unified GA/GP Package for R”, and Franz Rothlauf gave a talk on “Program Synthesis with Grammatical Evolution”.

Grammatical evolution for constraint synthesis for mixed-integer linear programming

Handcrafting mixed-integer linear programming (MILP) models can be a time-consuming and error-prone task. A novel algorithm, Grammatical Evolution for Constraint Synthesis (GECS), has been proposed which produces well-formed MILP models in the ZIMPL modelling language. GECS outperform state-of-the-art algorithms, and appears resistant to the curse of dimensionality. The research collaboration between Dr Tomasz Pawlak (Poznan University of Technology) and Prof Michael O’Neill (UCD Natural Computing Research & Applications Group) has been published in the journal Swarm and Evolutionary Computation.

Pawlak T., O’Neill M. (2021). Grammatical Evolution for Constraint Synthesis for Mixed-Integer Linear Programming. Swarm and Evolutionary Computation, 64:100896. 

Data-driven time series feature extraction

Stefano Mauceri and co-authors (James Sweeney, Miguel Nicolau and James McDermott) have published their latest research on a data-driven approach to feature extraction from time series for one-class classification in the journal Genetic Programming & Evolvable Machines. The approach, which uses Grammatical Evolution to automatically select both the features to extract and the sub-sequences from which to extract them, is demonstrated to lead to problem understanding and improved performance.

Figure 9 from Mauceri et al., 2021 is licensed under CC BY 4.0

Best Paper Award nomination @ PPSN 2020

David Lynch and Michael O’Neill with NCRA alumnus James McDermott have been nominated for the Best Paper Award at the PPSN 2020 conference to be held in Leiden in September. The paper brings together the use of grammars and autoencoders in a novel approach to program synthesis.

Lynch D., McDermott J., O’Neill M. (2020). Program Synthesis in a Continuous Space using Grammars and Variational Autoencoders. The Sixteenth International Conference on Parallel Problem Solving from Nature (PPSN XVI). Springer LNCS.

Better generalisation performance for symbolic regression models

Miguel Nicolau with NCRA alumnus Alexandros Agapitos have published their latest research on function sets, generalisation and symbolic regression in the Genetic Programming & Evolvable Machines journal.

Nicolau, M., Agapitos, A. (2020). Choosing function sets with better generalisation performance for symbolic regression modelsGenetic Programming and Evolvable Machineshttps://doi.org/10.1007/s10710-020-09391-4

20th Anniversary issue of the Genetic Programming & Evolvable Machines journal

NCRA researchers (Róisín Loughran, Tony Brabazon and Michael O’Neill) have published three articles in the 20th Anniversary issue of the journal Genetic Programming & Evolvable Machines. Two articles provide an overview of research in application areas we have been focusing on as a group for sometime, Finance & Economics (https://rdcu.be/b4KkQ), and Computational Creativity (https://rdcu.be/b4KkT). The third article, “Automatic Programming: The Open Issue?” (https://rdcu.be/b4KkV) follows on from an article by O’Neill et al that appeared in the 10th Anniversary issue highlighting Open Issues in the field of GP, and raises a challenge to the community to re-ignite a focus on Automatic Programming, the open issue, which we previously referred to as the “elephant in the room”. 

PHP Code Snippets Powered By : XYZScripts.com