Support poster for the paper "Resilient Bioinspired Algorithms: A Computer System Design Perspective", presented at EvoStar 2022, https://doi.org/10.1007/978-3-031-02462-7_39
Resilient Bioinspired Algorithms: A Computer System Design Perspective
1. Resilient Bioinspired Algorithms
A Computer System Design Perspective
C. CottaUMA, ITIS, G. OlagueCICESE
The normal operation of the
system should not result in
damage in its own state
Diversity loss
Premature convergence
Malicious agents
The system can perform changes on itself
or adapt any aspect of its functioning in
order to ensure appropriate performance
• Self-tuning
• Meta-optimization
The system must perform above specific
functioning requirements, and deliver
correct service conditions beyond the
typical domain of operation
Fault tolerance
Repairability
Intrinsic
Extrinsic
Self-healing
The system must provide
continuous service and
maximize its readiness to it.
• Service continuation in the presence of failures
• Service continuation in changing environments
Dynamic
optimization
The system must maintain
its operation in the long run.
Green AI
Sustainability
Systems with a low systemic risk build-up increasingly
fragile, ultimately undermining sustainability
Volatility Paradox
Resilience is an intrinsic feature of bioinspired
optimization techniques that deserves further analysis.
Resilience should be boosted to improve the
performance and usefulness of bioinspired optimization
techniques.
Position
Population(s)
Algorithmic add-ons,
checkpointing