Insights on transfer optimization : because experience is the best teacher

• Main Authors: Gupta, Abhishek, Ong, Yew-Soon, Feng, Liang
• Other Authors: School of Computer Science and Engineering
• Format: Article
• Language: English
• Published: 2021
• Subjects: Engineering::Computer science and engineering; Transfer; Multitasking
• Online Access: https://hdl.handle.net/10356/147980
• Institution: Nanyang Technological University

Traditional optimization solvers tend to start the search from scratch by assuming zero prior knowledge about the task at hand. Generally speaking, the capabilities of solvers do not automatically grow with experience. In contrast, however, humans routinely make use of a pool of knowledge drawn from past experiences whenever faced with a new task. This is often an effective approach in practice as real-world problems seldom exist in isolation. Similarly, practically useful artificial systems are expected to face a large number of problems in their lifetime, many of which will either be repetitive or share domain-specific similarities. This view naturally motivates advanced optimizers that mimic human cognitive capabilities; leveraging on what has been seen before to accelerate the search toward optimal solutions of never before seen tasks. With this in mind, this paper sheds light on recent research advances in the field of global black-box optimization that champion the theme of automatic knowledge transfer across problems. We introduce a general formalization of transfer optimization, based on which the conceptual realizations of the paradigm are classified into three distinct categories, namely sequential transfer, multitasking, and multiform optimization. In addition, we carry out a survey of different methodological perspectives spanning Bayesian optimization and nature-inspired computational intelligence procedures for efficient encoding and transfer of knowledge building blocks. Finally, real-world applications of the techniques are identified, demonstrating the future impact of optimization engines that evolve as better problem-solvers over time by learning from the past and from one another.