ANOMALY DETECTION ACROSS MULTI SCALE TEMPORAL DATA STREAMS FOR HUMAN BEHAVIOR MODELING
Loading...
Links to Files
Permanent Link
Author/Creator
Author/Creator ORCID
Date
2020-01-20
Type of Work
Department
Information Systems
Program
Information Systems
Citation of Original Publication
Rights
Distribution Rights granted to UMBC by the author.
This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see http://aok.lib.umbc.edu/specoll/repro.php or contact Special Collections at speccoll(at)umbc.edu
This item may be protected under Title 17 of the U.S. Copyright Law. It is made available by UMBC for non-commercial research and education. For permission to publish or reproduce, please see http://aok.lib.umbc.edu/specoll/repro.php or contact Special Collections at speccoll(at)umbc.edu
Abstract
This research focuses on developing computational models for human behavior from multi scale temporal data to detect anomalous behavior, evaluating car driving behavior as a case study. Human behavioral patterns capture frequent or repeated behaviors of users in the data. Here the usage data can be multi-scale from devices, computer networks or even vehicle driving data. We define novel computational models that represent these patterns in data in order to associate them with the human behaviors for the detection of the anomalous situations. Such behavioral patterns can be associated with identifying anomalies which we believe are precursors or even indicators of impending or ongoing unexpected behavior. Current state of the art in driving behavior has not focused on assessing behavioral models from multiple streams of temporal data which might be complex. This is an important problem in the domain of driving pattern detection at large because human behavior is impacted by the environment in and around the car. Thus, it is important to study any type of this usage data in combination across multiple data streams to understand a human behavioral perspective. Through our research we aim to address the discovery of anomalous human behavioral patterns in the driving domain. We present time series based anomaly detection utilizing car telematics data, eye gaze distraction data and health vital statistics data to provide a comprehensive view of the driver behavioral patterns. We analyze different scales and resolutions of time from seconds to minutes and the anomalous variations and their intensities in the data streams also impacted accordingly from minor fluctuations to major spikes.
We identify anomalies, which might be precursors or even indicators of impending or ongoing unexpected behavior and detect anomalous activities in different settings to understand behaviors in reactive environments such as automobiles. We attribute the anomalous behavior to distraction, driver health, vehicular state or other external factors. Our results indicate that each of the heterogeneous temporal data streams of Telematics data, Eye tracking data, Driver vital health data individually detect anomalies in driving states. However, gaze data is more representative of the anomalies than Health and Telematics data in individual streams. In general, we also found that all data stream combinations are useful, however, presence of anomalies in eye gaze data is more indicative of anomalous behavior. We are also able to supplement the anomalous state information through the combination and overlap of anomalies in the three data streams. We compare results from our methodology with traditional data mining methods and found that some of these overlapping anomalies across the three data streams and unique anomalies in gaze data are missed. Auto industry and the auto insurance companies gather and analyze mostly Telematics data to gage driver'sdriving behavior and categorize the safety of the driver according to their driving speed, abrupt acceleration, abrupt deceleration, sharp turns, etc. Our research helps provide a more comprehensive view of the driving behavior by incorporating multiple heterogeneous data streams.