The 18th International Conference on Hybrid Artificial Intelligence Systems (HAIS 2023) took place in the historical city of Salamanca (Spain), from September 5th to September 7th. This is a long running event whose current edition has been co-chaired by Emilio Corchado and Héctor Quintián. The conference was held in the historical building of the Colegio Mayor de Fonseca, a magnificient college founded in 1519, in conjunction with two other related scientific conferences, namely SOCO 2023 and CISIS 2023, as well as with a entrepeneurship event, the Startup OLE 2023. All together, this provided a lively environment to present and discuss scientific matters as well as to get to know people from different areas.
The Bio4Res project was present at the conference with a work entitled “Enhancing Evolutionary Optimization Performance Under Byzantine Fault Conditions” by C. Cotta. This work continues the research line on resilient bioinspired algorithms and provides a more in-depth numerical study of several mechanisms to cope with Byzantine faults in the form of unreliable sources of fitness information. The abstract follows:
We evaluate the performance of panmictic evolutionary algorithms (EAs) in Byzantine environments, where fitness values are unreliable due to the potential presence of malicious agents. We investigate the impact of this phenomenon on the performance of the algorithm considering two different models of malicious behavior of different severity, taking the unreliability rate of the environment as a control parameter. We observe how there can be a significant toll in the quality of the results as the prevalence of cheating behavior increases, even for simple functions. Subsequently, we endow the EA with mechanisms based on redundant computation to cope with this issue, and examine their effectiveness. Our findings indicate that while a mechanism based on statistical averaging can be an effective approach under a relatively benign fault model, more hostile environments are better tackled via an approach based on majority voting.
C. Cotta, Enhancing Evolutionary Optimization Performance Under Byzantine Fault Conditions. In: García Bringas, P., et al. Hybrid Artificial Intelligent Systems. HAIS 2023. Lecture Notes in Computer Science 14001. Springer, Cham, 2023, doi:10.1007/978-3-031-40725-3_29
The presentation was followed by an really interesting discussion with the audience. Overall, the conference was a very nice experience from the scientific and social point of view. Kudos to the local organizers for providing such a good environment for discussion and networking!