Program (tentative)

June 4, 2025

All oral presentations will be 25 minutes in duration, which includes a 20-minute presentation followed by a 5-minute question and discussion period.

Chair: Koji Nakano, Hiroshima University

Speaker: Ryo Kuroiwa, National Institute of Informatics

Title: Massively Parallel and Distributed State Space Search Algorithms for Domain-Independent Dynamic Programming

Abstract: This talk focuses on our recent work on developing general-purpose solvers for combinatorial optimization in a massively parallel and distributed environment. Our work is based on domain-independent dynamic programming (DIDP), where a problem is modeled as a dynamic programming model and solved by state space search. After introducing DIDP, we show that naive extensions of existing solvers result in poor scalability. Then, we introduce our new parallel and distributed state space search algorithms to overcome such issues. We demonstrate the scalability of the developed solvers using up to 93.2 TB of RAM and 24576 CPU cores.

Chair: Victor Parque, Hiroshima University

Parallel Processing for Distributed Machine Learning: Taxonomy of Techniques and Cloud Deployment Implications, Abdulfatah Bahbouh, Ishfaq Ahmad, Hansheng Lei and Saif Ul Islam

Efficient Intra-node Hierarchical Parallelisms And Dynamic Load Balancing Strategies On Heterogeneous Systems, Dikshant Pratap Singh, Mathialakan Thavappiragasam and Brice Videau

Chair: Victor Parque, Hiroshima University

Extending Microservices Performance Optimization Through Horizontal Pod Autoscaling: A Comprehensive Study, Fernando Buzato and Alfredo Goldman

Enhancing Productivity and Performance of HClib-Actor with Efficient Task Termination, Youssef Elmougy, Nirjhar Deb, Akihiro Hayashi and Vivek Sarkar

Dynatune: Dynamic Optimization of Raft Election Parameters Using Network Measurement, Kohya Shiozaki and Junya Nakamura

Chair: Ryota Yasudo, Kyoto University

Pairbot: Enhancing Computational Capabilities by Pairing of Autonomous Mobile Robots, Yonghwan Kim, Yoshiaki Katayama and Koichi Wada

Towards a Fast and Generalizable Neural Inference Scheme for Tabular Data, Victor Parque

Performance Modeling of Non-Uniform Heterogeneous Platforms, Steven Harris, Roger Chamberlain and Chris Gill

Chair: Ryo Kuroiwa, National Institute of Informatics

RL-assisted Annealing for QUBO on a Multi-GPU System, Reo Gakumi and Ryota Yasudo

CUBO-to-QUBO Conversion: Reducing Cubic Formulations to Quadratic Formulations, Shunsuke Tsukiyama, Xiaotian Li, Koji Nakano, Victor Parque, Yasuaki Ito, Takumi Kato, Yuya Kawamata and Kaiki Ii

QUBO++: A C++ Library for Developing and Solving QUBO Problems, Koji Nakano, Shunsuke Tsukiyama, Xiaotian Li, Yasuaki Ito, Victor Parque, Takumi Kato, Yuya Kawamata and Kaiki Ii