NEWS

Rakuten and Hitachi Collaborate at NeurIPS 2024 Workshop

2025.01.17

On December 15, 2024, Pablo Loyola and advisor Toyotaro Suzumura from Rakuten Institute of Technology, in collaboration with researchers from Hitachi, Ltd., presented their joint research at the NeurIPS 2024 workshop, “Machine Learning with New Compute Paradigms.” During the poster session, the team showcased a paper titled “Annealing Machine-assisted Learning of Graph Neural Network for Combinatorial Optimization.”

This collaborative research introduces a novel hybrid approach that leverages the strengths of CMOS annealing machines and graph neural networks (GNNs) to tackle complex combinatorial optimization problems. CMOS annealing machines, known for their ability to find highly accurate solutions to optimization tasks, are used to guide the training of GNNs, which excel at learning patterns in graph-structured data. By combining these technologies, the method achieves both the precision of CMOS annealing machines and the scalability and flexibility of GNNs, making it suitable for large-scale, real-world applications.

The proposed approach has significant potential in industries such as logistics, telecommunications, and manufacturing, where solving large and complex optimization problems are critical for improving efficiency and performance.

External link

Hitachi R&D Page (Japanese only): 日立と楽天が、CMOS アニーリング技術とグラフニューラルネットワーク技術を組み合わせることで、大規模な組合せ最適化問題を高速かつ高精度に解く新たな手法を開発

Copied! instagram