2021 Fiscal Year Research-status Report
Resilience in the Facility Location Problem: Theory and Practice
| Project/Area Number |
20K11947
|
|---|
| Research Institution | National Institute of Advanced Industrial Science and Technology |
|---|
Principal Investigator |
Nicolas Schwind 国立研究開発法人産業技術総合研究所, 情報・人間工学領域, 主任研究員 (60646397)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | Resilience / Multiagent systems / Facility Location / Team Formation / Belief Change |
|---|
| Outline of Annual Research Achievements |
Following our work in FY2020 on the introduction of a novel notion of partial robustness in team formation and facility location, we have pursued our investigation and introduced a novel algorithm to find resilient team deployment solutions. Our new algorithm is anytime, which means that a sub-optimal solution is found first and improved over time. Our empirical evaluation shows that a near-optimal solution can be found very quickly, which shows the practicability of our new approach. We have also pursued the development of a software that artificially generates facility location instances. Our software allows for a high degree of flexibility, allowing the user to generate various types of instances according to many different parameters. We have also included the solving algorithms in the software and a solution visualization tool. This work is currently under review for a journal submission (Autonomous Agents and Multiagent Systems) and is expected to be published in FY2022. The final version of our software will also be publicly release together with the journal publication. Lastly, we have conducted some theoretical and fundamental aspects of resilience and change and published two papers in top tiers international conference on the topic.
|
| Current Status of Research Progress |
Current Status of Research Progress
1: Research has progressed more than it was originally planned.
Reason
Our new anytime algorithm for finding resilient team deployment solutions outperforms our previous algorithm published in FY2021 by far: near-optimal resilient solutions can now be found on all our facility location instances in a few seconds, while our previous complete algorithm failed to find any solution within one hour. As planned, all of our results have been reported and submitted to a journal, and our software is about to be finalized and publicly released. In addition, we have investigated theoretical aspects of resilience and change and published our works in AAAI 2022, the premier international conference in Artificial Intelligence, and KR 2022, one of the premier international conference in theoretical aspects of Artificial Intelligence. Both venues are top-tiers (A*).
|
| Strategy for Future Research Activity |
In the short term, we plan to publish our full report to the journal of Autonomous Agents and Multiagent Systems, and publicly release the final version of our software. To make it accessible to the community, we plan to present our software and facility location instances to the SAT community, to make them directly useful to the research community. We will submit our results to the workshop POS 2022 (Pragmatics of SAT), and prepare a set of benchmark for the SAT 2022 competition. As we did in FY2021, we will also pursue our theoretical works on resilience in terms of the system's accommodation to change.
|
| Causes of Carryover |
Our plan was to use the fundings for traveling purpose to international conferences and to visit the collaborators of this project to CRIL, University of Artois, France. The coronavirus situation made this plan impossible due to travel restrictions, so only a small amount of the budget was used to register to an online conference (AAAI 2022). The remaining budget will be used in FY2022 for traveling purpose to get physical meetings with the project collaborators and to attend international conferences.
|
