L3S Researchers have received the Best Paper Award at this year’s IEEE International Conference on Data Mining (ICDM) for their paper “A Deep Reinforcement Learning Approach to Configuration Sampling Problem”.
“Configuration sampling is a complex problem involving finding a minimal subset of configurations that includes all possible t-wise combinations of features. We introduced a new method called RLSampler that uses deep reinforcement learning to achieve this, ensuring 100% coverage with fewer configurations compared to state-of-the art heuristic methods”, says Amir Abolfazli, one of the authors of the paper. “Our tests on eight real-world feature models of software product lines demonstrated that our approach performs much better than heuristic-based methods, meaning we need fewer configurations to test. This reduction in testing size leads to cost savings by requiring less effort, resources, and time for testing” explains Amir Abolfazli further.
The ICDM took place from Dec. 1 – Dec. 4, 2023. It is widely regarded as the leading research conference in the field. The ICDM is a forum for presenting original research, sharing innovative development experiences, and facilitating the exchange of ideas in the field of data mining. The conference covers a wide range of topics, including algorithms, software, systems, and applications. Big data, deep learning, pattern recognition, statistical and machine learning, databases, data warehousing, data visualization, knowledge-based systems, and high-performance computing are all represented at ICDM. The conference aims to advance the field’s state-of-the-art by showcasing novel research and creative solutions to data mining challenges.
Many congratulations to our colleagues for this recognition!
