Downloaded 76 times
![Distance Metric Learning
Bryan Perozzi
Focused Clustering and Outlier Detection in Large Attributed Graphs
11
nodes
attributes
Feature Matrix
PS and PD intermixed
nodes
attributes
[Xing, et al 2002]
Focused Attribute Vector
PS closer together](https://image.slidesharecdn.com/14-kdd-focusedclustersandoutliers-perozzi-141201172459-conversion-gate01/85/Focused-Clustering-and-Outlier-Detection-in-Large-Attributed-Graphs-11-320.jpg)
![Experiment set up
Synthetic and Real World Graphs
Performance measures:
Cluster quality: NMI
Outlier accuracy: precision, F1
Compared to:
CODA [Gao+’10]
METIS (no outlier detection) [Karypis+’98]
Bryan Perozzi
Focused Clustering and Outlier Detection in Large Attributed Graphs
16](https://image.slidesharecdn.com/14-kdd-focusedclustersandoutliers-perozzi-141201172459-conversion-gate01/85/Focused-Clustering-and-Outlier-Detection-in-Large-Attributed-Graphs-16-320.jpg)
Uploaded byBryan Perozzi
Focused Clustering and Outlier Detection in Large Attributed Graphs
The document presents a new approach called FocusCO for focused clustering and outlier detection in large attributed graphs. FocusCO infers the most relevant attributes (called the "focus") based on examples provided by the user, and uses this focus to extract dense clusters that are coherent in the focused attributes. It can also detect outliers that deviate from the inferred focus. The experiments on synthetic and real-world graphs show that FocusCO outperforms other methods in identifying clusters aligned with different user-specified focuses and detecting outliers corresponding to each focus.
More Related Content
Similar to Focused Clustering and Outlier Detection in Large Attributed Graphs
Focused Clustering and Outlier Detection in Large Attributed Graphs
- 1. Focused Clustering andOutlier Detection in Large Attributed Graphs ACM SIG-KDD August 26, 2014 Patricia Iglesias Sánchez*, Emmanuel Müller*† *Karlsruhe Institute of Technology †University of Antwerp Bryan Perozzi, Leman Akoglu Stony Brook University
- 2. Attributed Graphs Attributedgraph: each node has 1+ properties Examples: Age School Relationship Status Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 2
- 3. Focused Mining ofAttributed Graphs Numerous attributes (ex: Facebook profiles) Many irrelevant for most queries Ex: When trying to sell mortgages Useful: Income, Credit Score, Employer Not Useful: Hair Color, # Apps Installed Ex: When trying to sell make up Useful: Hair Color, Skin Tone, Gender Not Useful: Shoe Size Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 3 Users have a Focus Algorithms need a Focus too! Focus Focus
- 4. Adding Focus toAlgorithms Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 4 Users provide examples of the kind of similarity they are interested in. We infer the similarity function that matters to them. examples ! user infer focus attributes focus Task
- 5. Outline Introduction NewProblem: Focused Clustering & Outliers Our Approach: FocusCO Evaluation Conclusion Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 5
- 6. Focused Clusters andOutliers: Problem Given 1) a graph w/ node attributes, 2) exemplar nodes by the user Infer attribute weights/relevance Extract focused clusters: 1) dense in structure, 2) coherent in “heavy” attributes (called the “focus”) Detect focused outliers: *) nodes deviating in focus attribute values Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 6
- 7. An Example BryanPerozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 7 Users provide examples of nodes they consider similar. Ex: ‘Yann LeCun’ and ‘Foster Provost’ We learn a focus Education Level Location We extract clusters which agree with the focus We detect outliers which don’t agree with focus
- 8. Related Work Graph Clustering Attributed Graphs Attribute Subspace User Preference Outlier Detection METIS, Spectral ✓ Parallel Nibble, BigClam ✓ CoPaM, Gamer ✓ ✓ ✓ CODA ✓ ✓ ✓ GOutRank, ConSub ✓ ✓ ✓ FocusCO ✓ ✓ ✓ ✓ ✓ Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 8
- 9. examples FocusCO: sketch Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 9 … … age gender location 1 2 3 infer “focus” attribute(s) 4 detect focused clusters & outliers
- 10. Focus attribute inference Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 10 1. Construct a set of similar pairs, PS 2. Construct a set of dissimilar pairs, PD Randomly sample pairs (u,v) 3. Learn a distance metric between PS and PD Input: Set of similar nodes, Cex Pair user examples together Cex
- 11. Distance Metric Learning Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 11 nodes attributes Feature Matrix PS and PD intermixed nodes attributes [Xing, et al 2002] Focused Attribute Vector PS closer together
- 12. examples FocusCO: sketch Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 12 … … age gender location 1 2 3 infer “focus” attribute(s) 4 detect focused clusters & outliers
- 13. FocusCO: Cluster Extraction Local clustering algorithm Not cluster whole graph Expands a cluster around a starting set Two procedures: 1.Finding good candidate sets to start at 2.Growing clusters Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 13
- 14. Finding nodes tocluster around Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 14 1.) We reweigh the graph using the focus 2.) We keep only highly weighted edges 3.) The connected components are our seeds A seed set
- 15. Growing a FocusedCluster Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 15 2. At each step in cluster expansion: 2.1 - Examine boundary nodes 2.2 - Add node with best Δ 2.3 - Record best structural node 1. Clustering objective: conductance weighted by focus Cluster Member Focused Outlier 3. Focused Outliers: left out best structural nodes
- 16. Experiment set up Synthetic and Real World Graphs Performance measures: Cluster quality: NMI Outlier accuracy: precision, F1 Compared to: CODA [Gao+’10] METIS (no outlier detection) [Karypis+’98] Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 16
- 17. Focused clustering performance Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 17 9 clusters (3 focus1 + 3 focus2 +3 unfocused). 5 focus attributes.
- 18. Focused clustering performance Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 18
- 19. Outlier detection performance Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 19 # deflated focus attributes increased (easier) from left to right
- 20. Disney: Amazon co-purchasegraph Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 20 Images are Focused Outliers
- 21. DBLP co-authorship graph Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 21 Focused Outlier publishes in IR
- 22. Political blogs citationgraph Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 22 Focused Outlier did not mention Waas.
- 23. Summary Bryan Perozzi Focused Clustering and Outlier Detection in Large Attributed Graphs 23 A new graph mining paradigm where the focus steers graph mining according to user preference. A new problem formulation Focused Clustering & Outlier detection Thanks! Any questions? Bryan Perozzi (bperozzi@cs.stonybrook.edu) examples ! user infer focus attributes focus Clustering