Search

An important challenge in the cyber-space is the effective identification of network anomalies, often caused by malicious activities. With the remarkable advances, machine learning algorithms have widely been studied for network intrusion and anomaly detection. In particular, deep learning based on neural network structures has recently been given a greater attention to deal with the growing complexity of data with higher dimensions and non-linearity. Convolutional Neural Networks (CNNs) is one of the widely employed deep learning methods. In this work, we introduce a new encoding technique that enhances the performance for the identification of anomalous events using a CNN structure. To evaluate, we utilize three different datasets for the extensive analysis. The experimental results show that our method consistently outperforms the gray-scale encoding technique previously proposed over the datasets employed in the evaluation.

Deep learning has been widely applied to network anomaly detection to improve performance. In our past research, we empirically evaluated a set of deep learning models, including Fully Connected Network (FCN), Variational Auto Encoder (VAE), and Sequence to Sequence model with Long Short-Term Memory (Seq2Seq-LSTM), for network anomaly detection. Additionally, we evaluate Convolution Neural Networks (CNNs) for network anomaly detection in this paper. We set up three simple CNN models with different internal depths (shallow CNN, moderate CNN, and deep CNN) to see the impact of the depth to the performance. We evaluate the models using three different types of traffic datasets. Our experimental results show that deeper structures do not make any performance improvement. In addition, we observed that the evaluated CNN models occasionally outperform the VAE models, but do not work better than the other deep learning models based on FCN and Seq2Seq-LSTM.

Identifying anomalous events in the network is one of the vital functions in enterprises, ISPs, and datacenters to protect the internal resources. With its importance, there has been a substantial body of work for network anomaly detection using supervised and unsupervised machine learning techniques with their own strengths and weaknesses. In this work, we take advantage of the both worlds of unsupervised and supervised learning methods. The basic process model we present in this paper includes (i) clustering the training data set to create referential labels, (ii) building a supervised learning model with the automatically produced labels, and (iii) testing individual data points in question using the established learning model. By doing so, it is possible to construct a supervised learning model without the provision of the associated labels, which are often not available in practice. To attain this process, we set up a new property defining anomalies in the context of clustering, based on our observations from anomalous events in network, by which the referential labels can be obtained. Through our extensive experiments with a public data set (NSL-KDD), we will show that the presented method perform very well, yielding fairly comparable performance to the traditional method running with the original labels provided in the data set, with respect to the accuracy for anomaly detection.

Biological network inference is of importance to understand underlying biological mechanisms. Gene regulatory networks describe molecular interactions of complex biological processes. Graph models are mainly used for gene regulatory networks, where nodes and edges represent genes and their regulations respectively. In the most research, the molecular interactions (edges) of gene regulatory networks are inferred from a single type of genomic data, e.g., gene expression data. However, gene expression is a product of sequential interactions of DNA sequence variations, single nucleotide polymorphism, copy number variation, histone modifications, transcription factor, DNA methylation, and many other factors. There are high-throughput genomic data that measure the various biological processes. We call the multiple types of genomics data as ‘multi-omics data’. In this paper, we propose an Integrative Gene Regulatory Network inference method (iGRN) that can incorporate multi-omics data and their interactions in the graph model of gene regulatory network. Copy number variation and DNA methylation were considered for multi-omics data in this paper. The proposed method, iGRN, was applied to the human brain data of psychiatric disorder. Through the experiments, iGRN showed its better performance on model representation and interpretation than other integrative methods in gene regulatory network inference.

Cyber-attacks have been evolved in a way to be more sophisticated by employing combinations of attack methodologies with greater impacts. For instance, Advanced Persistent Threats (APTs) employ a set of stealthy hacking processes running over a long period of time, making it much hard to detect. With this trend, the importance of big-data security analytics has taken greater attention since identifying such latest attacks requires large-scale data processing and analysis. In this paper, we present SEAS-MR (Security Event Aggregation System over MapReduce) that facilitates scalable security event aggregation for comprehensive situation analysis. The introduced system provides the following three core functions: (i) periodic aggregation, (ii) on-demand aggregation, and (iii) query support for effective analysis. We describe our design and implementation of the system over MapReduce and high-level query languages, and report our experimental results collected through extensive settings on a Hadoop cluster for performance evaluation and design impacts.

Robots have come up as an emerging technology to assist humans in performing those repetitive and dangerous tasks which humans prefer not to do, or are unable to do due to size limitations and extreme environmental conditions. So the design of robots is required to be stringent to be able to operate at any environmental conditions. A wheeled robot is the one mostly associated in mobile robot operations. But a wheeled robot can travel only on smooth surface and not at irregular surface conditions. To overcome this, the robot's mobile platform equipped with an efficient structure and function is required. Another issue with mobile robot is the power source which is generally battery. The disadvantage of battery is its low efficiency and low energy density. To override the above two issues, A competent robot platform suitable for irregular areas is designed and the conventional power source is replaced with a fuel cell to develop an efficient and environmental independent robot.

We propose a computational model based on OWL/SWRL enabled ontologies, which can shape the development of an automated software tool for the purpose of providing patient-specific reminders, advice and action-items in preventing the development of diabetic foot in diabetic patients. The tool is aimed at both: (i) patients who would like to manage their illness efficiently by being informed and alerted to the significance of any change(s) they detect in their feet and (ii) healthcare professionals who can disseminate their knowledge to patients more effectively, and thus prevent the development of diabetic foot, which may cause the premature death of diabetic patients. The advantages of using OWL/SWRL enabled ontologies in our computational model are numerous. They range from the power to store, manage and reason effectively upon knowledge and information related to diabetic foot problems and their prevention through OWL/SWRL computations, to the feasibility of including such computations into software applications, which may run as a set of Apps on Android devices or on personalized healthcare iClouds. Consequently in the core of our proposal are (a) the OWL ontological model and its constraints which define and store the semantics of symptoms and observations of the changes in diabetic patient's feet and (b) a reasoning process which uses the semantics and the power of ontological matching through SWRL for the purpose of delivering functionalities of the tool.

Proton Exchange Membrane Fuel Cells (PEMFC) are the most appropriate energy source for small robot applications. PEMFC is superior in power density and thermodynamic efficiency as compared with the Direct Methanol Fuel Cell (DMFC). Furthermore, PEMFC is lighter weight and smaller in size than DMFC, which are very important factors for small robot power systems. Operating conditions and fabricating methods are of key importance to fuel cell performance and efficiency. In this study, analysis of pressure drop, concentration variation, deformation of GDL and electrolyte, and stack clamping pressure were completed for a 500 watt PEM fuel cell stack. Several novel designs were discussed and will be continuously studied.

In mobile multi-robot, estimation of robot location is a fundamental and critical issue to provide efficient management of given tasks for the robot systems. We propose the ideal method to estimate location information of multiple mobile robots without stationary anchors. According to our simulation result, the proposed localization algorithm decreases maximal 34.55% of computational overhead compared with the traditional trilateration based localization.

Proton exchange membrane fuel cell (PEMFC) is a key of solving recent energy issues such as increasing carbon monoxide emission and energy consumption; however, the system of PEMFC is complex and unclear. Thus, the proper method for understanding the PEMFC system is urgently required. The purposes of this study are that design reliable control based and lumped model for PEMFC system and utilize the designed model for systematic analysis of PEMFC stack and variety operating logic. The model was designed using MathWork® Matlab/Simulink. The model for the PEMFC stack consists four sub models which are the cathode, the anode, the membrane hydration, and the voltage model. Each model is based on the proper fluid dynamic and the thermodynamic principal. The I-V curves at varying pressure and relative humidity, which are the results of this model, show the reliable results. The results of this study' control based PEMFC modeling, represent that they are a key of systematic approach of PEMFC.

This paper presents an unmanned underwater vehicle (UUV) with high maneuverability to be controlled in strong tidal currents. A radio buoy is connected to the UUV with an intermediate cable, and a wireless RF modem is installed to communicate with a support vessel. This paper presents the design and development of the UUV and the buoy systems. The vehicle equips with two main thrusters and four vertical/ horizontal tunnel thrusters in the middle of the vehicle. This paper discusses the hydrodynamic characteristics of the multiple thrusters, and strategies to maneuver the UUV with multiple thruster. Finally this paper presents effectiveness of the UUV with the buoy in conducting a simulated object installation.

Recently, agent-based technology is actively being considered in the field of defense Modeling and Simulation (M&S). An existing defense M&S technology has limitations to modeling and simulation for complex combat system effectively. And for various weapon systems, mission effectiveness analysis also includes subjective and qualitative factors. In other words, a large number of complex weapon systems, or integration factors are modeled as a single object to perform the simulation, are highly vulnerable in terms of scalability, reusability etc. In this paper, we perform an agent-based modeling to quickly assemble and disassemble object just as you want, and ensure a seamless interface between all the objects using one of the commercial software ‘AnyLogic’ that includes an agent function. As a result, we will overcome limitations of existing defense M&S.

The Differentiated Services architecture represents a highly scalable architecture for deployment of Quality of Services across the next-generation Internet. The Differentiated Services enable the packets that pass through network devices operating on layer 3 information, such as routers, to have their relative priority differentiated from one another. With Differentiated Services marking, Layer 3 devices can establish aggregated precedence-based queues and provide better service (when packet service is subject to queuing, as is the case under significant traffic loads) to packets that have higher relative priority. Though the Differentiated Services filled in the Internet Protocol header serves the purpose of identification of different kinds of traffic and aggregating traffic with similar Quality of Services requirements into classes, it may practically pose challenges when getting deployed on a large scale network. These challenges include: a) Differentiated services handling large amount of traffic focused on differentiating different kinds of traffic rather than analyzing who generated the traffic. b) Differentiated Service based switching is with the management of actual service response, which an end node receives. The primary purpose of this paper is to identify these shortcomings related to prioritization of traffic based on the Internet Protocol header's Differentiated Services field at the routers in a large scale network and to propose a unique solution by using the MPLS and some rule sets to help overcome these shortcomings.

Distributed Interactive Simulation of Modeling and Simulation in the field of national defense has a problem of interoperability among heterogeneous simulators. The problem is caused of simulators are limited due to environmental, and it is obstacle to interchange the data. Therefore, intermediate standard data format was required to interchange environmental data to solve the problem. SEDRIS were intent to be developed the standard intermediate format. SEDRIS provide a standard interchange mechanism to distribute environmental data and promotes reusability of the data. SEDRIS are broad enough to represent all environmental things, but also complicated to use for constructing environmental structure. Interchange via SEDRIS require converters between SEDRIS transmittal format(STF) and others. This study addresses converting and verifying process from NetCDF, which has time-dependent grid environmental structure to STF. In addition, this study suggests efficient DRM structure for time-dependent grid environmental data.

Location sensing of physical objects is one of critical issues in many applications. Passive UHF Radio Frequency Identification (RFID) technique provides an efficient solution because of its low cost for installation and easy identification of the tagged physical objects. In this paper, we research on the localization problem using passive UHF RFID systems. We discuss theoretical and practical characteristics of a passive UHF RFID system. We propose novel algorithm to minimize the number of ambiguous tag points against a single RSSI value from a target tag and increase accuracy of location estimation in 2D grid space. Because of a single Received Signal Strength Indicator (RSSI) can be related to multiple points in the monitoring area due to signal propagation when we use the RSSI based localization technique. According to the analysis of our experiment results, our proposed approach shows over 50% improvement compared with the conventional k-Nearest Neighbor algorithm in a test frame, and 23.24cm of estimation error with a high granularity for localization of box level items with 4 different positions in the immobile applied application such as a smart shelf.

Many global enterprises are now operating their production facilities in foreign countries in an effort to capitalize on the merits that each country provides. In such environment, where design, production, supply chains, and other business functions are geographically separated, great challenges exist in terms of how to effectively measure, control and monitor the product quality on remotely situated production lines. In this context, this paper presents a novel, holistic, network-based quality control strategy, which is called EQM (short for e-quality for manufacture). With EQM, remotely situated designers can access real-time quality data and update quality control processes as changes in production occur. EQM also allows the quality data to be integrated within the company's information network with other related business functions. Such integration enables instant generation of corrective actions, such as automatic notification to supply chains/vendors for non-conformities. EQM relieves human operators from laborious, tedious, and error-prone tasks of quality monitoring and sampling practices, hence greatly improves the overall efficiency of production operations. The proposed remote-quality control strategy has been tested using the networked-robotic cell, which produces automotive starter gear housing assemblies. The results vindicate the notion of EQM, manifesting many potential benefits for industry.

This paper proposes an simplification algorithm of a range image. One of the merits of the proposed approach is that it is not necessary to convert the range image into a triangular mesh (initial triangular mesh), a process that requires a large amount of memory, because the proposed algorithm works directly on a range image. The key idea of the paper is to employ an additional data structure, called a ‘flag map’, for the support of the irregular topology changes that happen during the simplification procedure. As the flag map is able to maintain all the history of the topology changes, an easy extraction of the simplified triangular mesh from the flag map is possible.

This Paper presents three different ways to describe the notion concept. The first one uses the idea of hierarchy and employs a graph to define the connections between attributes and concepts. To enable concepts generation, manipulation or measurement a matrix model is developed. Thus, the entire space of terms could be generated by a set of (linearly) independent terms over a numerical field. The total number of concepts to be generated over a numerical field using a given number of attributes is estimated in this paper. A memory efficient spherical presentation is developed on the above models. An algorithmic implementation of Hierarchy of concepts and attributes (HAC) is developed and experimental results are presented using concepts from the field of education.

Advent of recent high-throughput next generation sequencing technologies has fostered the demand of high performance sequence data analysis. Next generation sequence analysis is a very important but very challenging task in bioinformatics due to extremely large-scale datasets. A variety of searching methods have been proposed. Recent emerging multi-core computing technology makes it possible to speed up sequence analysis. In this paper, we discuss the problem of parallelizing next generation sequence analysis queries on multi-core technology. We develop a systematic method to solve this problem by using genetic algorithms. The proposed method provides us a globally optimal solution of this parallelization problem. The resulting efficiency and accuracy is superior to the ones obtained by using those approaches in which this problem is divided into the two parts: location and allocation, respectively.

Clustering enhances the value of existing databases by revealing rules in the data. These rules are useful for understanding trends, making predictions of future events from historical data, or synthesizing data records into meaningful clusters. Through clustering are similar data items grouped together to form clusters. Clustering algorithms usually employ a distance metric based (e.g., Euclidean) similarity measure in order to partition the database such that data points in the same partition are more similar than points in different partitions. In this paper, we study clustering algorithms for data with categorical attributes. Instead of using traditional clustering algorithms that use distances between points for clustering which is not an appropriate concept for Boolean and categorical attributes, we propose a novel concept of HAC (hierarchy of attributes and concepts) to measure the similarity/proximity between a pair of data points. In this study, HAC will be used as an aid to represent medical domain knowledge substructures to simplify the generation process of the databases through clustering. As a result, the research will identify interesting relationships and patterns among the data, and represent them in the form of association rules.

We present a 3D tracking method of multi-objects by color and stereo. The results are applied to navigation and manipulation in a virtual environment. We choose a human face and hands as tracking targets for HCI. To extract the area of face and hands in a complex background, we transform an input color image using the GSCD (generalized skin color distribution). Based on the transformed image, we detect the area of face and hands by WUPC. Furthermore, the face and hands candidate region are estimated by a Kalman filter in the following frame. Then we can extract the depth information about the only corresponding region, respectively, by stereo matching. Finally, we can navigate and manipulate objects based on the extracted 3D information of face and hands naturally in the virtual environment without any additional device.