Chuan Feng

A Hybrid View in a Laparoscopic Surgery Training System

In this paper, a hybrid view application is proposed - a subsystem of a computerized laparoscopic surgery training system. To minimize the potential hazards of laparoscopic surgery, an assistive training system is being developed. A digital camera and magnetic position sensors are used to detect laparoscopic instruments in the system. The hybrid view is a component of this system which overlays the positions of organs and objects with the path history of the instruments. This method could help confirm erroneous movements made by surgeons and provide more useful information than separate sensors. This may minimize the cognitive overload on the surgeons. Initial experimental results are presented to show the feasibility of the proposed method

Alert Fusion for a Computer Host Based Intrusion Detection System

Intrusions impose tremendous threats to todays computer hosts. Intrusions using security breaches to achieve unauthorized access or misuse of critical information can have catastrophic consequences. To protect computer hosts from the increasing threat of intrusion, various kinds of intrusion detection systems (IDSs) have been developed. The main disadvantages of current IDSs are a high false detection rate and the lack of post-intrusion decision support capability. To minimize these drawbacks, we propose an event-driven intrusion detection architecture which integrates subject-verb-object (SVO) multi-point monitors and an impact analysis engine. Alert fusion and verification models are implemented to provide more reasonable intrusion information from incomplete, inconsistent or imprecise alerts acquired by SVO monitors. DEVS formalism is used to describe the model based design approach. Finally we use the DEVS-JAVA simulation tool to show the feasibility of the proposed system

A hybrid intrusion detection and visualization system

Network attacks have become the fundamental threat to todays largely interconnected computer systems. Unauthorized activities and unauthorized access account for a large proportion of these networks. Unauthorized accesses and misuse of critical data can be catastrophic to businesses, emergency services, and even threaten the defense and security of a nation. Intrusion detection system (IDS) is indispensable to defend the system in the face of increasing vulnerabilities. This paper proposes a hybrid intrusion detection and visualization system that leverages the advantages of current signature-based and anomaly detection methods. The hybrid instruction detection system deploys these two methods in a two-staged manner to identify both known and novel attacks. When intrusion is detected, autonomous agents that reside on the system automatically take actions against misuse and abuse of computer system, thus protecting the system from internal and external attacks

Computer-Guided Laparoscopic Training with Application of a Fuzzy Expert System

Laparoscopic surgery is a widely accepted operating technique which continues to spread into different areas of medicine. Because of its differences to open surgery (like limited perception) it demands a different training program than the traditional surgical training programs. Since its introduction in 1980s several training curriculums for laparoscopic surgeons have been deployed and a set of skills that need to be mastered during the training has been defined. The training system proposed in this paper uses a knowledge base to guide the trainee trough the process of acquiring the necessary skills, based on the trainees measured performance in several areas. The systems guidance allows for better understanding of areas that need additional work and for faster acquisition of those, without the need for extra attention from the tutoring staff.

Design of a Wireless Sensor Network Based Automatic Light Controller in Theater Arts

In this paper, an automatic lighting controller designed and built using a wireless sensor network indoor positioning technology is described. This controller can autonomously track actors during a real-time theatrical performance. Kalman filter and 3D trilateration technologies were used with Cricket wireless sensors to implement this system. In addition, an entertainment industry standard protocol DMX-512 and an efficient calibration method were applied to realize remote computerized fixture control. As far as we know, this controller is the first application of wireless sensor networks in the theater arts area. A successful public performance concert at the University of Arizona validated the performance of the system

Fuzzy Logic-Based Performance Assessment in the Virtual, Assistive Surgical Trainer (VAST)

The Virtual Assistive Surgical Trainer (VAST) is an approach developed to train surgeons in minimally invasive procedures. It uses surgical instruments augmented with micro-sensors, and knowledge-based inference techniques to provide objective, data-driven feedback and performance assessment for complex exercises. The assessment is typically based on the expertise of senior surgeons and, thus, a single objective standard is difficult to define. To formulate such a standard, and to provide an accurate scoring method, a fuzzy logic method is proposed in this paper. This makes it easier to mimic tasks that are already successfully performed by human experts. A multi-level fuzzy inference engine and new performance metrics are implemented. Experimental results demonstrate the feasibility of this method and the efficacy of the new performance metrics.

A Non-numerical Predictive Model for Asymmetric Analysis

Predicting asymmetric threats (e.g., terrorist events) is becoming ever more important. Prior works have focused on tactical, statistical, and data-fusion systems. The thrust of our work has been the development of a non-numerical predictive model for amplifying intelligence analysts’ recognition of emergent threats. The intelligence community uses a Template schema for assessing courses of action. Our predictive model processes non-numerical data to arrive at automated assessment and confidence scores for these Templates. The predictive model is traceable, transparent, and utilizes Human-in-the-Loop data-fusion. For future work, this predictive model will be further enhanced with behavioral filtering. Behavioral filtering adjusts the assessment and confidence of the predictions by intelligently evaluating characteristic behavioral data. This non-numerical predictive model has been tested and verified in the Asymmetric Threat Response and Analysis Program (ATRAP).

Behavior Analysis-Based Learning Framework for Host Level Intrusion Detection

Machine learning has great utility within the context of network intrusion detection systems. In this paper, a behavior analysis-based learning framework for host level network intrusion detection is proposed, consisting of two parts, anomaly detection and alert verification. The anomaly detection module processes unlabeled data using a clustering algorithm to detect abnormal behaviors. The alert verification module adopts a novel rule learning based mechanism which analyzes the change of system behavior caused by an intrusion to determine whether an attack succeeded and therefore lower the number of false alarms. In this framework, the host behavior is not represented by a single user or program activity; instead, it is represented by a set of factors, called behavior set, so that the host behavior can be described more accurately and completely

An Event-Driven Architecture for Fine Grained Intrusion Detection and Attack Aftermath Mitigation

In todays computing environment, unauthorized accesses and misuse of critical data can be catastrophic to personal users, businesses, emergency services, and even national defense and security. To protect computers from the ever-increasing threat of intrusion, we propose an event-driven architecture that provides fine grained intrusion detection and decision support capability. Within this architecture, an incoming event is scrutinized by the subject-verb-object multipoint monitors. Deviations from normal behavior detected by SVO monitors will trigger different alarms, which are sent to subsequent fusion and verification modules to reduce the false positive rate. The system then performs impact analysis by studying real-time system metrics, collected through the Windows management instrumentation interface. We add to the system the capability to assist the administrator in taking effective actions to mitigate the aftermath of an intrusion