A5064322994- Advanced Multi-Objective Optimization Algorithms
- Metaheuristic Optimization Algorithms Research
- Evolutionary Algorithms and Applications
- Machine Learning and Data Classification
- Data Stream Mining Techniques
- Simulation and Modeling Applications
- Reinforcement Learning in Robotics
- Topic Modeling
- Advanced Neural Network Applications
- Machine Learning and Algorithms
- Gaussian Processes and Bayesian Inference
- Aerospace and Aviation Technology
- Explainable Artificial Intelligence (XAI)
- Air Traffic Management and Optimization
- Advanced Bandit Algorithms Research
- Machine Learning and ELM
Nanyang Technological University
2019-2024
Agency for Science, Technology and Research
2021-2022
Singapore Institute of Manufacturing Technology
2021
North Carolina State University
2001
This paper presents a first study on solution representation learning for inducing greater alignment and hence positive transfers between distinct multi-objective optimization tasks that bear discrepancies in their original search spaces. We establish novel probabilistic model-based transfer evolutionary (TrEO) framework with learning, capable of activating while simultaneously curbing the threat negative transfers. In particular, well-aligned representations are learned via spatial...
For deep learning, size is power. Massive neural nets trained on broad data for a spectrum of tasks are at the forefront artificial intelligence. These large pre-trained models or "Jacks All Trades" (JATs), when fine-tuned downstream tasks, gaining importance in driving learning advancements. However, environments with tight resource constraints, changing objectives and intentions, varied task requirements, could limit real-world utility singular JAT. Hence, tandem current trends towards...
Transfer evolutionary optimization (TrEO) has emerged as a computational paradigm to leverage related problem-solving information from various source tasks boost convergence rates in target task. State-of-the-art Tr EO algorithms have utilized source-target similarity capture method with probabilistic priors that grants the ability reduce negative transfers. A recent work makes use of an additional solution representation learning module induce high ordinal correlation between and objective...
We aim to showcase the benefit of transfer optimization for route planning problems by illustrating how solution accuracy travelling salesman problem instances can be enhanced via autonomous and positive knowledge from related source that have been encountered previously. Our approach is able achieve better exploiting useful past experiences at runtime, based on a source-target similarity measure learned online.
For deep learning, size is power. Massive neural nets trained on broad data for a spectrum of tasks are at the forefront artificial intelligence. These large pre-trained models or Jacks All Trades (JATs), when fine-tuned downstream tasks, gaining importance in driving learning advancements. However, environments with tight resource constraints, changing objectives and intentions, varied task requirements, could limit real-world utility singular JAT. Hence, tandem current trends towards...
Recent theoretical results have shown that instilling knowledge transfer into black-box optimization with Gaussian process surrogates, aka Bayesian optimization, tightens cumulative regret bounds compared to the no-transfer case. Faster convergence under strict function evaluation budgets - often in order of a hundred or fewer evaluations is thus expected, overcoming cold start problem conventional algorithms. In this short paper, we prove can be further tightened when extending method...
A flight simulator was developed for studying the behavior of pilots in power-off aircraft landing situations. The simulation environment includes a 5-meter hemispherical dome which authors have installed cockpit from Cessna aircraft. manufacturers provided their version OPEN GL 1.1. graphics rendering software has undergone constant modification because computer and projection hardware changes lack knowledge understanding manufacturer's undocumented GL. development team led to believe that...