a discrete krill herd optimization algorithm for community detection
TRANSCRIPT
A Discrete Krill Herd Optimization Algorithm forCommunity Detection
11th International Computer Engineering Conference , Cairo university 2015.
Khaled Ahmed
http://www.egyptscience.net
Overview
Introduction Motivation Problem Definition
Related Work Proposed Approach Results and Discussion Conclusion and Future Works
11th International Computer Engineering Conference , Cairo university 2015.)
2
Introduction3
11th International Computer Engineering Conference , Cairo university 2015.)
The rapid increase on the social networks present an urgent need for identifying the community structure.social network is a set of nodes which represent users or profiles and a set of edges, which represents the interaction between nodes it could be internal or external connectivity.
Motivation4
11th International Computer Engineering Conference , Cairo university 2015.
Can we get Valuable insights for these social networks ?One of these Valuable insights is ‘Community detection ‘ .
Why Community detection ?
- Its help in many analysis the social networks: citation network might represent related papers on a single topic , represent pages of related topics, easy to visualize and understand. �
Problem Definition Finding Communities in complex network such as Online Social
network is a difficult task, many challenges . Scalability: such networks can be huge, often in a scale of millions
of actors and hundreds of millions of connections, Existing Community detection techniques might fail when applied directly to networks of this size.
Heterogeneity: In reality, multiple relationships can exist between individuals, and multiple types of entities can also be involved in one network.
Evolution: Social network emphasizes timeliness, which makes the network dynamic and changes over time.
Evaluation: the task of comparison and evaluation different work in community detection is also a challenging process, because of the lack of ground truth for many social computing tasks.
11th International Computer Engineering Conference , Cairo university 2015.
5
Related work There are common algorithms used for
community detection : Discrete BAT Algorithm. Artificial fish swarm algorithm. Infomap . Fast Greedy. label
propagation. Multilevel . Walktrap.11th International Computer Engineering Conference , Cairo university 2015.
6
Contribution of the paper Change the krill algorithm domain Algorithm redesign Present discrete krill algorithm Comparing our results with 7 popular
algorithm over four benchmarks
7
Proposed Approach Krills [21] move in herd searching for food and the
food makes the herd is divided into groups with minimum distance between krill , food and calculate an objective function with highest density krills groupsbasic steps :
Induced Motion (N). Foraging Motion (F). Random diffusion (D).11th International Computer Engineering Conference , Cairo university 2015.
8
Proposed Approach
11th International Computer Engineering Conference , Cairo university 2015.
9
Results and discussion We used four social networks which are The
Zachary Karate , Club,The Bottlenose Dolphin ,American College football anda Facebook data set.
We compared our results againts Discrete BAT Algorithm . Artificial fish swarm algorithm .
Infomap . Fast Greedy . label propagation. Multilevel . Walktrap .
11th International Computer Engineering Conference , Cairo university 2015.
10
Results and discussion
11th International Computer Engineering Conference , Cairo university 2015.
11
Results and discussion
11th International Computer Engineering Conference , Cairo university 2015.
12
Results and discussion
SRGE workshop in Cairo University Conference Hall (12-September-2015)
13
Results and discussion
11th International Computer Engineering Conference , Cairo university 2015.
14
AKHSO’s community detection for Facebook data set.
Results and discussion
11th International Computer Engineering Conference , Cairo university 2015.
15
AKHSOs community detection for Bottlenose Dolphin data set.
Results and discussion
11th International Computer Engineering Conference , Cairo university 2015.
16
AKHSOs community detection for Zachary Karate data set.
Results and discussion
11th International Computer Engineering Conference , Cairo university 2015.
17
AKHSOs community detection for college Football data set.
Conclusion discrete Krill herd algorithm for complex social network’s community
detection. The locus-based adjacency scheme is used for two reasons first for encoding and decoding tasks , the second for representing a community structure. A discrete Krill herd uses Modularity as an objective function in the optimization process.
This research presents a quite promising accuracy and high Modularity results for community structure. It presents good results for small and medium size of data sets, although it presents medium results over big data sets .
11th International Computer Engineering Conference , Cairo university 2015.
18
Future Works Future Works of this research is to
enhance it’s performance over big data sets, presents
hybrid krill herd with other optimization algorithm. Presents a comparative study for more than Swarm optimization algorithm over social networks based on experimental results.11th International Computer Engineering Conference , Cairo university 2015.
19
Thanks and Acknowledgement20
11th International Computer Engineering Conference , Cairo university 2015.