Applied Intelligence & Data Analysis

Projects

/ EphemeCH

Bioinspired Algorithms in Complex Ephemeral Environments (EphemeCH)

ACRONYM OF THE COORDINATED PROJECT: EphemeCH

Funded by: Spanish Ministry of Economy and Competitivity and European Regional Development Fund FEDER (TIN2014-56494-C4-4-P)

Dates: 2015-2018

Research Groups: GISUM–Coordinator– (Universidad de Málaga), Geneura (Unversidad de Granada), Grupo de Evolución Artificial (Universidad Autónoma de Madrid), AIDA (Universidad Autónoma de Madrid)

SUMMARY

This coordinated project focuses on the ephemeral properties -those with a transitory nature- that may affect the functioning of distributed versions of bioinspired algorithms. The availability of highly-volatile heterogeneous computer resources capable of running software agents requires an appropriate design and implementation of algorithms. This will allow to make a proper use of the available resources while circumventing the potential problems that may produce such non-reliable systems. Among the desired features for the algorithms under consideration -that will potentially be run on non-dedicated local computers, remote devices, cloud systems, ubiquituous systems, etc.- we look for ephemerality-awareness, which is related to self-capability for understanding the underlying systems where the algorithm is run as well as taking decisions on how to proceed taking into account the non-reliable nature of the system.

Some of the relevant features that the so-adapted versions of bioinspired algorithm should feature are: (i) Inclusion (all nodes should have a meaningful contribution to the final result, and they should be incorporated to the distributed system in such a way that they do), (ii) Asynchrony (nodes communicate with the others without a fixed schedule due to their different performance), (iii) Resilience (the sudden disappearance of computing nodes must not destabilize the functioning of the algorithm), (iv) Emergence (the nature of the computational environment does not allow a centralized control and requires decentralized, emergent behavior), and (v) Self-adaptation (the algorithm should adapt itself to the changing computational landscape).

We aim to develop models and algorithms featuring these desired properties, while approaching specific problems where they can be tested. The goals of the project are therefore: (1) Use bioinspired algorithms to analyze, model and optimize ephemeral computation environments, and (2) Deploy bioinspired systems on ephemeral computing environments to solve complex problems. We will pursue the previous goals using an Open Science philosophy, thus leading to an additional added-value objective, namely (3) Bridge the gap between Science and Society. More precisely, the above goals substantiate in the following objectives of the project: (i) Design bioinspired algorithms adapted to complex ephemeral environments. (ii) Performance study and prediction in ephemeral environments. (iii) Develop game AI and procedural game-content generators in ephemeral environments. (iv) Study computational creativity in ephemeral computing environments. (v) Create workflows and raise awareness on Open Science.

The two application areas considered, namely computational creativity (where overall goals include analyzing and evolving the underlying social patterns of use and interactions, as well as making use of ephemeral properties of agents involved) and content generation in games (focusing on the automatic generation of diverse content, ranging from fixed components such as maps or levels, to emergent ingredients -e.g., narrative, or secondary goals- and the development of human-like non-player characters), constitute proofs-of-concept for the technologies developed and will provide results of stand-alone value. They are specific targets of the European research strategy and open important avenues for the technological transfer to industry.

KEY WORDS OF THE COORDINATED PROJECT: ephemeral computing, bioinspired algorithms, optimization, self-*, computational creativity, videogames, open science