Evolutionary Computation
The rapid advancement of digital technology has led to a notable expansion in the complexity of optimization problems encountered in practice. Consequently, this presents significant challenges to existing optimization methodologies. Among the plethora of available optimizers, evolutionary computation (EC) has emerged as a subject of growing research interest. This is primarily attributed to EC’s outstanding global search capabilities and its feature of being agnostic to problem-specific information.
Our research aims to develop evolutionary computation algorithms that are efficient, adaptive, learnable, and scalable, thereby enabling them to effectively tackle a wide range of problems that are of high dimensionality, multi-level coupling, and computationally expensive. These encompass various endeavors, including but not limited to the creation of learnable generators capable of generating promising solutions within expansive solution spaces, the design of evolutionary multitasking to concurrently manage multiple distinct tasks, and the application of multiform optimization techniques that leverage auxiliary tasks with innovative formulations to address complex problems. Furthermore, we work with industry collaborators to apply our developed algorithms to resolve complex real-world problems, such as logistics, neural architecture search, and machine health prognostics, among many others. Finally, we develop high-performance computing libraries that can take advantage of modern computing infrastructures to accelerate our algorithms, enabling timely decision-making across various application scenarios.
Publications
[1] Tan, K. C., Feng, L., and Jiang, M. (2021). Evolutionary transfer optimization-a new frontier in evolutionary computation research. IEEE Computational Intelligence Magazine, 16(1), 22-33. (IEEE CIM Outstanding Paper Award 2024)
[2] Zhang, C., Lim, P., Qin, A. K., and Tan, K. C. (2016). Multi-objective deep belief networks ensemble for remaining useful life estimation in prognostics. IEEE Transactions on Neural Networks and Learning Systems, 28(10), 2306-2318.
[3] Feng, L., Zhou, L., Zhong, J., Gupta, A., Ong, Y. S., Tan, K. C., and Qin, A. K. (2018). Evolutionary multitasking via explicit autoencoding. IEEE Transactions on Cybernetics, 49(9), 3457-3470.
[4] Xue, X., Zhang, K., Tan, K.C., Feng, L., Wang, J., Chen, G., Zhao, X., Zhang, L. and Yao, J. (2020). Affine transformation-enhanced multifactorial optimization for heterogeneous problems. IEEE Transactions on Cybernetics, 52(7), 6217-6231.
[5] Tian, Y., Lu, C., Zhang, X., Tan, K. C., and Jin, Y. (2020). Solving large-scale multi-objective optimization problems with sparse optimal solutions via unsupervised neural networks. IEEE Transactions on Cybernetics, 51(6), 3115-3128.