Fast Community Detection Algorithm with GPUs and Multicore Architectures

Soman, Jyothish; Narang, Ankur

doi:10.1109/ipdps.2011.61

Cited by 53 publications

(16 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A deeper analysis of this behavior is necessary, but difficult due to non-determinism. Additional techniques might help, such as counteracting "epidemic spread" of labels as examined in [19].…”

Section: Discussionmentioning

confidence: 99%

“…More related to our work is a variant of label propagation by Soman and Narang [19] for multicore and GPU architectures, which seeks to improve quality by re-weighting the graph, and has been shown to cluster a graph with about 100 million edges in a few minutes on an IBM Power6 system. Moreover, for an R-MAT graph similar to the one we call kron-g500-simple-logn20 in our experiments, they report running times of 60s with 50 threads for their label propagation algorithm on their system.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Engineering High-Performance Community Detection Heuristics for Massive Graphs

Staudt

Meyerhenke

2013

2013 42nd International Conference on Parallel Processing

View full text Add to dashboard Cite

The amount of graph-structured data has recently experienced an enormous growth in many applications. To transform such data into useful information, high-performance analytics algorithms and software tools are necessary. One common graph analytics kernel is community detection (or graph clustering). Despite extensive research on heuristic solvers for this task, only few parallel codes exist, although parallelism is often necessary to scale to the data volume of real-world applications.We address the deficit in computing capability by a flexible and extensible clustering algorithm framework with shared-memory parallelism. Within this framework we implement our parallel variations of known sequential algorithms and combine them by an ensemble approach. In extensive experiments driven by the algorithm engineering paradigm, we identify the most successful parameters and combinations of these algorithms. The processing rate of our fastest algorithm exceeds 10M edges/second for many large graphs, making it suitable for massive data streams. Moreover, the strongest algorithm we developed yields a very good tradeoff between quality and speed.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Engineering High-Performance Community Detection Heuristics for Massive Graphs

Staudt

Meyerhenke

2013

2013 42nd International Conference on Parallel Processing

View full text Add to dashboard Cite

show abstract

“…To reduce the computational complexity and to improve the modularity measure several algorithms are proposed [2][3][4][5][6][7][8][9][10][11][12]. These algorithms are measured the modularity by substractithe expected number of community edges from the actual number of intra-community edges [3], which mathematically given as:…”

Section: Preliminariesmentioning

confidence: 99%

Community Detection Using Modularity

2017

IJAERD

View full text Add to dashboard Cite

-In this paper, consider an algorithm of a community detection taking the diameter of the community as 1.INTRODUCTIONThe term community structure in a network means natural division of the nodes into densely connected subgroups within and between them connection is sparse. By identifying and analysing the community structure of a network, we can understand and characterised the related network and its properties. It is evident that in graph or network, communities are loosely defined and it is not possible to compare the quality of two communities getting by different algorithm of the same graph [1]. By modularity metric, the quality of community structure is measured by Newman[1] , which is based on expected connectivity between nodes .A good high modularity nodes gives good quality community structure. The problem of community detection is that how to define the modularity metric that it may compare the quality of two communities and the partitioning of the nodes that miximizes modularity. Based on the topology of the given network itself not on the expected connectivity, this paper consider a new approach to compute modularity of a community structure. The modularity of a community structure is the average node modularity within it,where the node modularity is defined as the ratio of the number of neighbours of a node v within the same community to the total degree of the node v. So we can say that the modularity value varies between 0 and 1 . By our approach, by joining a node into a community ,it always attempts to maximize its modularity. A node can be addedto a community till that the diameter of the community structure is less than or equal to the predetermined constant k<

show abstract

“…In our previous work [38] we presented an optimization where the compiler identifies iterations over common neighbors of two Algorithm MB DS PM DP CI Adamic Adar [17] Betweenness-Centrality [32] Closeness-Centrality [22] Dijkstra [19] Fattest-Path [28] Kosaraju [19] PageRank [36] Soman and Narang [39] Tarjan [42] Triangle Counting [41] vertices and uses a specialized algorithm to perform this iteration. Depending on the results of a program analysis, it also generates code to prune the search space for these common neighbors which significantly reduces the work to be done.…”

Section: Common Neighbor Iterationmentioning

confidence: 99%

Using domain-specific languages for analytic graph databases

Sevenich

Rest

et al. 2016

Proc. VLDB Endow.

View full text Add to dashboard Cite

Recently graph has been drawing lots of attention both as a natural data model that captures fine-grained relationships between data entities and as a tool for powerful data analysis that considers such relationships. In this paper, we present a new graph database system that integrates a robust graph storage with an efficient graph analytics engine. Primarily, our system adopts two domain-specific languages (DSLs), one for describing graph analysis algorithms and the other for graph pattern matching queries. Compared to the API-based approaches in conventional graph processing systems, the DSL-based approach provides users with more flexible and intuitive ways of expressing algorithms and queries. Moreover, the DSL-based approach has significant performance benefits as well, (1) by skipping (remote) API invocation overhead and (2) by applying high-level optimization from the compiler.

show abstract

Fast Community Detection Algorithm with GPUs and Multicore Architectures

Cited by 53 publications

References 23 publications

Engineering High-Performance Community Detection Heuristics for Massive Graphs

Engineering High-Performance Community Detection Heuristics for Massive Graphs

Community Detection Using Modularity

Using domain-specific languages for analytic graph databases

Contact Info

Product

Resources

About