19-07-2013, 02:01 PM
Creating Evolving User Behavior Profiles Automatically
Creating Evolving User.pptx (Size: 69.74 KB / Downloads: 40)
Automatically Sharing Web Experiences through a Hyperdocument Recommender System
The use of computer-based tools to augment human intellect and improve our overall ability.
In his work at the Bootstrap Institute, Engelbart coined the term “Collective IQ” to describe how a group can “leverage its collective memory, perception, planning, reasoning, foresight, and experience into applicable knowledge” to solve problems of users
Our WebMemex continuously captures users’ Web surfing history and uses this history to provide the users and their friends with suggestions of related Web pages to what they are currently viewing.
This system acts as an instantiation of an architecture for capturing and asynchronously sharing experiences for the automated recommendation of related information.
DEFEATING MASQUERADE DETECTION
In this report we generated four distinct attacks and compared their results with two very well known masquerade detection techniques.
Scattered Attack, Scattered Sorted Attack, Consecutive Attack and Consecutive Random User Attack.
Scattered Attack can successfully evade HiddenMarkovModel-based masquerade detection technique for attack length 10 to 50 attack commands
In case of Scattered Sorted Attack, it can evade both HMM-based masquerade detection and One-Class Na¨ıve Bayes till attack length of 60 commands are inserted in the block
Consecutive Attack can successfully evade HMM-based masquerade detection even when we insert 80 percentage of the block with attack commands
Combining Baiting and User Search Profiling Techniques for
Masquerade Detection
Masquerade attacks are characterized by an adversary stealing a legitimate user’s credentials and using them to impersonate the victim and perform malicious activities.
Like any anomaly-detection based techniques, detecting masquerade attacks by profiling user behavior suffers from a significant number of false positives
We combine a user behavior profiling technique with a baiting technique in order to more accurately detect masquerade activity.