ScaleMine: Scalable Parallel Frequent Subgraph Mining in a Single Large Graph

Abdelhamid, Ehab; Abdelaziz, Ibrahim; Kalnis, Panos; Khayyat, Zuhair; Jamour, Fuad

doi:10.1109/sc.2016.60

Cited by 60 publications

(55 citation statements)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, in Fig. 2, the subgraph containing vertices 1 and 2 occurs in two places, i.e., [1,2] and [2,1]. These identical subgraphs are called automorphisms, i.e., they are automorphic with each other.…”

Section: Subgraph Trees and Vertex/edge Extensionmentioning

confidence: 99%

See 1 more Smart Citation

Sandslash

Chen

Dathathri

Gill

et al. 2021

Proceedings of the ACM International Conference on Supercomputing

View full text Add to dashboard Cite

Graph pattern mining (GPM) is a key building block in diverse applications, including bioinformatics, chemical engineering, social network analysis, recommender systems and security. Existing GPM frameworks either provide high-level interfaces for productivity at the cost of expressiveness or provide expressive low-level interfaces at the cost of increased programming complexity. They also lack the flexibility to explore combinations of optimizations to achieve performance competitive with hand-optimized applications.We present Sandslash, an in-memory graph pattern mining framework that uses a novel programming interface to support productive, expressive, and efficient GPM on large graphs. Sandslash provides a high-level API that only needs a specification of the GPM problem from the user, and the system implements fast subgraph enumeration, provides efficient data structures, and applies high-level optimizations automatically. To achieve performance competitive with expert-optimized implementations, Sandslash also provides a low-level API that allows users to express algorithm-specific optimizations. This enables Sandslash to support both high-productivity and high-efficiency without losing expressiveness. We evaluate Sandslash using five GPM applications and a wide range of real-world graphs. Experimental results demonstrate that applications written using Sandslash's high-level or low-level API outperform those in state-of-the-art GPM systems Au-toMine, Pangolin, and Peregrine on average by 13.8×, 7.9×, and 5.4×, respectively. We also show that these Sandslash applications outperform expert-optimized GPM implementations by 2.3× on average with less programming effort. CCS Concepts• Software and its engineering → Application specific development environments.

show abstract

Section: Subgraph Trees and Vertex/edge Extensionmentioning

confidence: 99%

“…2, lightly colored subgraphs are removed by symmetry breaking, leaving a unique canonical subgraph for each set of automorphisms. Symmetry breaking can significantly prune the search tree: e.g., the subgraph [2,1] is not extended in Fig. 2 because it is automorphic to the subgraph [1,2].…”

Section: Pattern-aware Gpm Solutionsmentioning

confidence: 99%

Sandslash

Chen

Dathathri

Gill

et al. 2021

Proceedings of the ACM International Conference on Supercomputing

View full text Add to dashboard Cite

show abstract

“…It leverages pruning techniques to reduce the search space and provides optimized graph partitioning and collective communication operations. ScaleMine [3] is an MPI-based system to perform FSM, which uses approximation to optimize load balancing, prune the search space, and guide intra-task parallelism. ASAP [23] accelerates graph mining by sampling subgraph patterns but can only produce approximate results.…”

Section: Related Workmentioning

confidence: 99%

AutoMine

Mawhirter

2019

Proceedings of the 27th ACM Symposium on Operating Systems Principles

View full text Add to dashboard Cite

Graph mining algorithms that aim at identifying structural patterns of graphs are typically more complex than graph computation algorithms such as breadth first search. Researchers have implemented several systems with high-level and flexible interfaces customized for tackling graph mining problems. However, we find that for triangle counting, one of the simplest graph mining problems, such systems can be several times slower than a single-threaded implementation of a straightforward algorithm. In this paper, we reveal the root causes of the severe inefficiencies of state-of-the-art graph mining systems and the challenges to address the performance problems. We build AutoMine, a single-machine system to provide both highlevel interfaces and high performance for large-scale graph mining applications. The novelty of AutoMine comes from 1) a new representation of subgraph patterns and 2) compilation techniques that automatically generate efficient mining code with minimized memory consumption from a highlevel abstraction. We have extensively evaluated AutoMine against 3 graph mining systems on 8 real-world graphs of different scales. Our experimental results show that AutoMine often produces several orders of magnitude better performance and can process very large graphs existing systems cannot handle. CCS Concepts • Computing methodologies → Shared memory algorithms; • Software and its engineering → Compilers.

show abstract

“…The basic frequent subgraph detection problem involves finding subgraphs having frequency higher than a threshold. Parallel approaches for this problem involve a "bottom-up" candidate generation approach, combined with careful pruning, which builds embeddings of larger subgraphs using all possible embeddings of smaller subgraphs [23], [24]. While these results allow scaling to very large networks with millions of nodes, they give no guarantees on the performance.…”

Section: Related Workmentioning

confidence: 99%

MIDAS: Multilinear Detection at Scale

Ekanayake

Cadena

Wickramasinghe

et al. 2018

EasyChair Preprints

View full text Add to dashboard Cite

We focus on two classes of problems in graph mininghere: (1) finding trees and (2) anomaly detection using networkscan statistics in complex networks. These are fundamentalproblems in a broad class of applications. Most of the parallelalgorithms for such problems are either based on heuristics,which do not scale very well, or use techniques like colorcoding, which have a high memory overhead. In this paper, wedevelop a novel approach for parallelizing both these classesof problems, using an algebraic representation of subgraphsas monomials—this methodology involves detecting multilinearterms in multivariate polynomials. Our algorithms show goodscaling over a large regime, and they run on networks with closeto half a billion edges. The resulting parallel algorithm for treesis able to scale to subgraphs of size18, which has not beenshown before, and it significantly outperforms the best priorcolor coding based method (FASCIA) by more than two ordersof magnitude. Our algorithm for network scan statistics is thefirst such parallelization, and it is able to handle a broad class ofscan statistics functions (both parametric and non-parametric),with the same approach.

show abstract

ScaleMine: Scalable Parallel Frequent Subgraph Mining in a Single Large Graph

Cited by 60 publications

References 33 publications

Sandslash

Sandslash

AutoMine

MIDAS: Multilinear Detection at Scale

Contact Info

Product

Resources

About