ADVISER+: Toward a usable web-based algorithm portfolio deviser

• Main Authors: LAU, Hoong Chuin, MISIR, Mustafa, LI, Xiang LI, JIANG, Lingxiao
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2017
• Subjects: Artificial Intelligence and Robotics; Software Engineering
• Online Access: https://ink.library.smu.edu.sg/sis_research/3814; https://ink.library.smu.edu.sg/context/sis_research/article/4816/viewcontent/MIC2017_paper_153.pdf
• Institution: Singapore Management University

The present study offers a more user-friendly and parallelized version of a web-based algorithm portfolio generator, called ADVISER. ADVISER is a portfolio generation tool to deliver a group of configurations for a given set of algorithms targeting a particular problem. The resulting configurations are expected to be diverse such that each can perform well on a certain type of problem instances. One issue with ADVISER is that it performs portfolio generation on a single-core which results in long waiting times for the users. Besides that, it lacks of a reporting system with visualizations to tell more about the generated portfolios. ADVISER is extended as ADVISER+ to overcome both of these issues while introducing new tuning based portfolio generation capabilities.