en
0.25
0.5
0.75
1.25
1.5
1.75
2
CatchSync: Catching Synchronized Behavior in Large Directed Graphs
Published on Oct 07, 20142721 Views
Given a directed graph of millions of nodes, how can we automatically spot anomalous, suspicious nodes, judging only from their connectivity patterns? Suspicious graph patterns show up in many applica
Related categories
Chapter list
CatchSync00:00
Fraud Detection: Graph Analysis Problem - 100:19
Fraud Detection: Graph Analysis Problem - 201:00
Our Goals01:14
Outline - 101:37
Anomalies in Degree Distributions - 101:51
Anomalies in Degree Distributions - 202:28
Linear Classifier with “Degree”: Fail02:44
Graph Structure Distorted03:01
Traditional Fraud Detection03:12
Empty Profile?03:43
Few Followers?04:14
Many Followings?04:32
Content: Unavailable? Look Normal?04:48
Behavior is the Nature04:56
Outline - 205:19
Behavior-based Features05:21
Behavior-based Feature Space05:59
Fraudulent Behavior Patterns - 106:28
Fraudulent Behavior Patterns - 206:59
Fraudulent Behavior Patterns - 307:03
Fraudulent Behavior Patterns - 407:12
Fraudulent Behavior Patterns - 507:30
Fraudulent Behavior Patterns - 607:45
Outline - 308:12
Synchronicity and Normality - 108:18
Synchronicity and Normality - 208:32
Synchronicity-Normality Plot08:40
Theorem09:05
CatchSync Algorithm09:21
Outline - 409:33
Experiments09:37
Q1: Does CatchSync Remove Anomalies? - 109:56
Q1: Does CatchSync Remove Anomalies? - 210:23
Before CatchSync10:30
After CatchSync10:38
Q2: Is CatchSync Catching Actually Fraudulent Users? - 110:50
Q2: Is CatchSync Catching Actually Fraudulent Users? - 211:20
Q2: Is CatchSync Catching Actually Fraudulent Users? - 312:14
Q2: Is CatchSync Catching Actually Fraudulent Users? - 412:22
Q3: Is CatchSync Robust to Camouflage? - 112:33
Q3: Is CatchSync Robust to Camouflage? - 212:55
Q3: Is CatchSync Robust to Camouflage? - 313:16
Q3: Is CatchSync Robust to Camouflage? - 413:36
Conclusion13:50
Questions14:22