Accelerating Maximal Biclique Enumeration on GPUs

2024-01-10

arXiv.org

AI Summary

The maximal biclique enumeration (MBE) problem on large-scale graphs suffers from high computational complexity and poor scalability. To address this, we propose cuMBE—the first efficient GPU-parallel algorithm for MBE. Our method eliminates recursion entirely via a compact array-based data structure and synergistically combines coarse-grained task partitioning, fine-grained intra-thread optimizations, and dynamic work-stealing to mitigate load imbalance and memory overhead bottlenecks. By leveraging GPU parallelism and thread-block-level cooperative optimization, cuMBE achieves geometric mean speedups of 4.02× and 4.13× over the best sequential and multicore CPU algorithms, respectively, on both synthetic and real-world datasets. This advancement significantly enhances the scalability and practical applicability of MBE computation.

Technology Category

• Data Mining & Knowledge Management: Scalability, Parallel & Distributed Systems
• Machine Learning: Graph-based Machine Learning
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines

Abstract

Maximal Biclique Enumeration (MBE) holds critical importance in graph theory with applications extending across fields such as bioinformatics, social networks, and recommendation systems. However, its computational complexity presents barriers for efficiently scaling to large graphs. To address these challenges, we introduce cuMBE, a GPU-optimized parallel algorithm for MBE. Utilizing a unique data structure, called compact array, cuMBE eradicates the need for recursion, thereby significantly minimizing dynamic memory requirements and computational overhead. The algorithm utilizes a hybrid parallelism approach, in which GPU thread blocks handle coarse-grained tasks associated with part of the search process. Besides, we implement three fine-grained optimizations within each thread block to enhance performance. Further, we integrate a work-stealing mechanism to mitigate workload imbalances among thread blocks. Our experiments reveal that cuMBE achieves an geometric mean speedup of 4.02x and 4.13x compared to the state-of-the-art serial algorithm and parallel CPU-based algorithm on both common and real-world datasets, respectively.

Problem

• Accelerating Maximal Biclique Enumeration for large graphs
• Reducing computational complexity with GPU-optimized parallel algorithm
• Addressing workload imbalances in parallel processing for MBE

Innovation

• GPU-optimized parallel algorithm for MBE
• Compact array eliminates recursion and memory overhead
• Hybrid parallelism with work-stealing for load balance