Daihee Park

Genetic-based new fuzzy reasoning models with application to fuzzy control

The successful application of fuzzy reasoning models to fuzzy control systems depends on a number of parameters, such as fuzzy membership functions, that are usually decided upon subjectively. It is shown in this paper that the performance of fuzzy control systems may be improved if the fuzzy reasoning model is supplemented by a genetic-based learning mechanism. The genetic algorithm enables us to generate an optimal set of parameters for the fuzzy reasoning model based either on their initial subjective selection or on a random selection. It is shown that if knowledge of the domain is available, it is exploited by the genetic algorithm leading to an even better performance of the fuzzy controller. >

An efficient algorithm for fuzzy weighted average

In multisensor intelligent systems, the information fusion plays an important role. Several algorithms have been proposed for the purpose of aggregating imprecise sensory information represented by fuzzy numbers. This paper proposes an efficient algorithm to compute fuzzy weighted average, which turned out to be superior to the previous works by reducing the number of comparisons and arithmetic operations to O(n log n).

LMI-based design of stabilizing fuzzy controllers for nonlinear systems described by Takagi-Sugeno fuzzy model

Abstract There have been several recent studies concerning the stability of fuzzy control systems and the synthesis of stabilizing fuzzy controllers. This paper reports on a related study of the Takagi–Sugeno (TS) fuzzy systems, and it is shown that the controller synthesis problems for the nonlinear systems described by the TS fuzzy model can be reduced to convex problems involving linear matrix inequalities (LMIs). After classifying the TS fuzzy systems into three families based on how diverse their input matrices are, a unique controller synthesis procedure is given for each of the families. A numerical example is presented to illustrate the synthesis procedures developed in this paper.

Traffic flooding attack detection with SNMP MIB using SVM

Recently, as network flooding attacks such as DoS/DDoS and Internet Worm have posed devastating threats to network services, rapid detection and proper response mechanisms are the major concern for secure and reliable network services. However, most of the current Intrusion Detection Systems (IDSs) focus on detail analysis of packet data, which results in late detection and a high system burden to cope with high-speed network traffic. Little or no integration exists between IDS and SNMP-based network management, in spite of the extensive monitoring and statistical information provided by SNMP agents implemented on network devices and systems. In this paper we propose a lightweight and fast detection mechanism for traffic flooding attacks. Firstly, we use SNMP MIB statistical data gathered from SNMP agents, instead of raw packet data from network links. The involved SNMP MIB variables are selected by an effective feature selection mechanism and gathered effectively by the MIB update time prediction mechanism. Secondly, we use a machine learning approach based on a Support Vector Machine (SVM) for attack classification. Using MIB and SVM, we achieved fast detection with high accuracy, the minimization of the system burden, and extendibility for system deployment. The proposed mechanism is constructed in a hierarchical structure, which first distinguishes attack traffic from normal traffic and then determines the type of attacks in detail. Using MIB datasets collected from real experiments involving a DDoS attack, we validate the possibility of our approaches. It is shown that network attacks are detected with high efficiency, and classified with low false alarms.

Real-time classification of internet application traffic using a hierarchical multi-class SVM

In this paper, we propose a hierarchical application traffic classification system as an alternative means to overcome the limitations of the port number and payload based methodologies, which are traditionally considered traffic classification methods. The proposed system is a new classification model that hierarchically combines a binary classifier SVM and Support Vector Data Descriptions (SVDDs). The proposed system selects an optimal attribute subset from the bi-directional traffic flows generated by our traffic analysis system (KU-MON) that enables real-time collection and analysis of campus traffic. The system is composed of three layers: The first layer is a binary classifier SVM that performs rapid classification between P2P and non-P2P traffic. The second layer classifies P2P traffic into file-sharing, messenger and TV, based on three SVDDs. The third layer performs specialized classification of all individual application traffic types. Since the proposed system enables both coarse- and fine-grained classification, it can guarantee efficient resource management, such as a stable network environment, seamless bandwidth guarantee and appropriate QoS. Moreover, even when a new application emerges, it can be easily adapted for incremental updating and scaling. Only additional training for the new part of the application traffic is needed instead of retraining the entire system. The performance of the proposed system is validated via experiments which confirm that its recall and precision measures are satisfactory.

A unified scheme of shot boundary detection and anchor shot detection in news video story parsing

In this paper, we propose an efficient one-pass algorithm for shot boundary detection and a cost-effective anchor shot detection method with search space reduction, which are unified scheme in news video story parsing. First, we present the desired requirements for shot boundary detection from the perspective of news video story parsing, and propose a new shot boundary detection method, based on singular value decomposition, and a newly developed algorithm, viz., Kernel-ART, which meets all of these requirements. Second, we propose a new anchor shot detection system, viz., MASD, which is able to detect anchor person cost-effectively by reducing the search space. It consists of skin color detector, face detector, and support vector data descriptions with non-negative matrix factorization sequentially. The experimental results with the qualitative analysis illustrate the efficiency of the proposed method.

Intrusion detection system based on multi-class SVM

In this paper, we propose a new intrusion detection system: MMIDS (Multi-step Multi-class Intrusion Detection System), which alleviates some drawbacks associated with misuse detection and anomaly detection. The MMIDS consists of a hierarchical structure of one-class SVM, novel multi-class SVM, and incremental clustering algorithm: Fuzzy-ART. It is able to detect novel attacks, to give detail informations of attack types, to provide economic system maintenance, and to provide incremental update and extension with a system.

Automatic Recognition of Aggressive Behavior in Pigs Using a Kinect Depth Sensor

Aggression among pigs adversely affects economic returns and animal welfare in intensive pigsties. In this study, we developed a non-invasive, inexpensive, automatic monitoring prototype system that uses a Kinect depth sensor to recognize aggressive behavior in a commercial pigpen. The method begins by extracting activity features from the Kinect depth information obtained in a pigsty. The detection and classification module, which employs two binary-classifier support vector machines in a hierarchical manner, detects aggressive activity, and classifies it into aggressive sub-types such as head-to-head (or body) knocking and chasing. Our experimental results showed that this method is effective for detecting aggressive pig behaviors in terms of both cost-effectiveness (using a low-cost Kinect depth sensor) and accuracy (detection and classification accuracies over 95.7% and 90.2%, respectively), either as a standalone solution or to complement existing methods.

A Cost-Effective Pigsty Monitoring System Based on a Video Sensor

Automated activity monitoring has become important in many applications. In particular, automated monitoring is an important issue in large-scale management of group-housed livestock because it can save a significant part of farm workers’ time or minimize the damage caused by livestock problems. In this paper, we propose an automated solution for measuring the daily-life activities of pigs by using video data in order to manage the group-housed pigs. Especially, we focus on the circadian rhythm of group-housed pigs under windowless and 24-hour light-on conditions. Also, we derive a cost-effective solution within the acceptable range of quality for the activity monitoring application. From the experimental results with the video monitoring data obtained from two pig farms, we believe our method based on circadian rhythm can be applied for detecting management problems of group-housed pigs in a cost-effective way.

Automatic Detection and Recognition of Pig Wasting Diseases Using Sound Data in Audio Surveillance Systems

Automatic detection of pig wasting diseases is an important issue in the management of group-housed pigs. Further, respiratory diseases are one of the main causes of mortality among pigs and loss of productivity in intensive pig farming. In this study, we propose an efficient data mining solution for the detection and recognition of pig wasting diseases using sound data in audio surveillance systems. In this method, we extract the Mel Frequency Cepstrum Coefficients (MFCC) from sound data with an automatic pig sound acquisition process, and use a hierarchical two-level structure: the Support Vector Data Description (SVDD) and the Sparse Representation Classifier (SRC) as an early anomaly detector and a respiratory disease classifier, respectively. Our experimental results show that this new method can be used to detect pig wasting diseases both economically (even a cheap microphone can be used) and accurately (94% detection and 91% classification accuracy), either as a standalone solution or to complement known methods to obtain a more accurate solution.