Samsung SDS’s Paper on Reinforcement Learning Adopted by NeurIPS

Samsung SDS has published Artificial Intelligence in Neural Information Processing Systems (NeurIPS) for three consecutive years.

Samsung SDS has demonstrated its technological strength in artificial intelligence by publishing papers in Neural Information Processing Systems (NeurIPS), the world’s most prestigious AI society, for three consecutive years.

NeurIPS is the world’s largest and most prominent academic community in the field of artificial intelligence and machine learning. It is known for its strict standards of selecting and publishing only the top 25% of articles in the field. In addition, research papers published in society have a very strong technological impact.

A Samsung SDS paper published this year titled “Simulation-Guided Beam Search for Neural Combinatorial Optimization” drew attention to reinforcement learning, which allows neural networks to self-learn optimal problem solving methods. This technology quickly and accurately finds optimal answers to various problems that may arise in industrial areas, such as cloud computing resource (GPU) allocation and process optimization in manufacturing and logistics.

Using this technology, Samsung SDS joined the NeurIPS 2022 vehicle routing competition and placed third among 54 companies and universities worldwide.

The degree of competition is determined by how quickly and accurately decision makers find the optimal path to deliver the ordered product to the customer in the presence of various variables and constraints.

Samsung SDS also won the competition’s judges’ award, which is given to the team that uses artificial intelligence technology to solve problems in the most innovative and effective way.

This year, Samsung SDS has published a total of 11 papers in world-renowned societies such as NeurIPS.

For example, Empirical Methods in Natural Language Processing (EMNLP) 2022, a global language processing society, published a Samsung SDS paper on optimizing language models using meta-learning.

Meta-learning is a self-learning artificial intelligence system that can learn on its own from only given data.

In addition, the International Conference on Pattern Recognition (ICPR) 2022, a well-known conference in computer vision, adopted two papers on Samsung SDS self-learning based image classification and automated document classification.

Self-supervised learning is an AI technology that discovers and analyzes rules on its own using minimal data. It recognizes the subject and assigns meaning without human guidance.

Gwon Young-jun, Executive Vice President and Head of Technology Research at Samsung SDS, said, “This achievement is the result of Samsung SDS’s efforts to advance research in the medium and long term. We will continue to actively invest in R&D to innovate our technology. . to expand its use in a production environment.”


Also Read :  Mobile operators must pay Tk2,413cr owed to govt: BTRC

Leave a Reply

Your email address will not be published.

Related Articles

Back to top button