ABSTRACT. In software development, testing is the crucial and integral process to produce a reliable and high quality system. Model-based testing plays a signicant role in practice and a lot of researches on it has been investigated in recent years. The paper proposes an automated test data generation method from the information embedded in model elements such as UML sequence diagrams, class diagrams and Object Constraint Language (OCL). The method supports UML 2.0 sequence diagrams including 8 kinds of combined fragments describing control flow. Comparing with some approaches by using DFS or BFS search algorithms, our proposed method generates all possible test scenarios that error uncover capability is higher. Test data is also generated in testing loop fragment. Therefore, it helps to detect errors in testing loop and the concurrency errors such as safety and liveness property of the system can be found.

ABSTRACT. Software maintenance is an activity which includes enhancements, error corrections, optimization and deletion of existing features. These modifications may cause the system to work incorrectly. Therefore, regression testing becomes necessary. Regression testing is any type of software testing that seeks to uncover new software bugs, or regressions, in existing functional and non-functional areas of a system after changes, such as enhancements, patches or configuration changes, have been made to them. During the software maintenance phase, regression testing is certainly an expensive but necessary activity to make sure the new versions of the system do not “regress”. As software evolves, changes are that not only the implementation changes, but that the specification of the system changes too. We argue that guiding regression testing by the system specifications will be more accurate and cost effective. Lustre is a formal synchronous declarative language widely used for modeling and specifying safety critical applications in the fields of avionics, transportation, and energy production. In such applications, the testing activity to ensure correctness of the system plays a crucial role. During the development process, Lustre programs are often upgraded, so regression testing should be performed to detect bugs. In this paper, we present an approach to generate test cases in regression testing of Lustre programs. In this approach, a Lustre program is modeled by an operator network, then we determine the set of paths and compute symbolically the path activation conditions for each version. Test cases for regression are generated by comparing paths between versions. We apply this approach on a case study for regression testing of a Heater Controller System.

ABSTRACT. Activity Theory has had a long history when its research had been initiated by three researchers of the cultural historical school of Russian psychology, L. S. Vygotsky, A. N. Leont'ev and A. R. Luria, in the period from 1920s to 1930s. Recently, this theory has again drawn much attention in some fields of information technology such as human-computer interaction, Computer Supported Collaborative Work (CSCW). However, the theory still seems to lack of a suitable formalism. This causes some real problems related to the clarity of key concepts and representation of activities. That issues have limited the development of the theory. Therefore, our research aims to resolve that issues by presenting a formalism for the Activity Theory which is called plan model.

ABSTRACT. Designing software applications for multiple tenants is challenging. The task is even harder when designing pure multi-tenancy applications that must support different customers using a single codebase and data store. One of the most common problems when developing these systems is to support different graphical user interface not only for different users but also tenants. This critical requirement also applies to context-aware applications in which different graphical user interface should be presented to each user according to the user's context or software components and platforms that should allow developers easily to create different looks and feel for their applications. In this paper, we propose an architecture pattern for modeling graphical user interface that support different customizations and configurations. We evaluate the modularity, the reusability and the maintainability of our architecture pattern by qualitative analysis based on well-known patterns used in our proposed architecture pattern. We built real world systems to validate the applicability, the correctness, the security and the performance of our pattern. We believe that our pattern will be useful for software providers as well as normal organizations when building software components or software systems for different customers or creating multi-tenancy applications in a cloud-based environment or building context-aware applications.

ABSTRACT. Stateflow is a visual tool that is used extensively in industry for designing the reactive behaviour of embedded systems. Stateflow relies on techniques like simulation to aid the user in finding flaws in the model. However, simulation is inadequate as a means of detecting inconsistencies and incompleteness in the model. Tabular Expressions (function tables) have been used successfully in software development for more than thirty years. Tabular expressions are also visual representations of functions, but include the important properties of completeness and disjointness. In other words, a tabular expression is well-formed only when the input domain is covered completely (completeness), and when there is no ambiguity in the behaviour described by the tabular expression (disjointness). The goal of our work is to use the completeness and disjointness properties of well-formed tabular expressions to aid us in establishing those properties in Stateflow models. From the Stateflow models, we generate a new kind of tabular expression that includes extended output options. We use the informal Stateflow semantics from MathWorks documentation as the basis for generating our tabular expressions. The generated tabular expressions are then used to guarantee completeness and disjointness. We provide a transformation algorithm that we are implementing in a tool to automatically generate tabular expressions from Stateflow models.

ABSTRACT. Today, many modern devices create digital images, such as digital cameras, X-Ray scanners, and so on. Noise reduces image quality and result of the processing. For example, biomedical images are a type of digital images. In these images, there is a combination of two types of noises: Gaussian noise and Poisson noise. In this paper, we propose a method to remove these noises. This method is based on the total variation of an image brightness function. We combine two famous denoising models to use this combination of noises.

ABSTRACT. Dynamic hand gesture recognition is a challenge field evenly this topic has been studied for a long time because of lack of feasible techniques deployed for Human-Computer Interaction (HCI) applications. In this paper, we propose a new type of gestures which presents a cyclic pattern of hand shapes during a movement. Through mapping of commands (e.g., turn devices on/off; increasing volume/channel) as output of a gesture recognition system, main purposes of the proposed gestures are to provide a natural and feasible way in control alliances in a smart home such as television, light, fan, door, so on. The proposed gestures are represented by both hand shapes and directions. Thanks to cyclic pattern of the hand shapes during performing a command, hand gestures are more easily segmented from video stream. We then focus on several challenges of the proposed gestures such as: non-synchronization phase of the gestures, change of hand shapes along temporal dimension and direction of hand movements. Such issues are addressed using combinations of spatial and temporal features extracted from consecutive frames of a gesture. The proposed algorithms are evaluated on several subjects. Evaluation results confirm that the proposed method obtains accuracy rates at 96% for segmenting a hand gesture and 95% for recognizing a command, averagely.

ABSTRACT. The ability to identify maritime vessels and their type is an important component of modern maritime safety and security. In this work, we present the application of deep convolutional neural networks to the classification of maritime vessels images. We use the AlexNet deep convolutional neural network as our base model and propose a new model that is twice smaller than the AlexNet. We conduct experiments on both models on commodity hardware. We comparatively evaluate and analyse the performance of the models. We measure the top-1 and top-5 accuracy rate. The contribution of this work is the implementation, tuning and evaluation of automatic image classifier for the specific domain of maritime vessels with deep convolutional neural networks under the constraints imposed by commodity hardware and size of the image collection.

ABSTRACT. Basically, the task of Content Based Image Retrieval (CBIR) is, given an image from an user, to find the most similar images among images in the database. It is not an easy task due to the fact that the definition of similarity is very different between human and computer. This difference is called semantic gaps. To filling this gaps, many methods try to take into account the user feedback, i.e., running the system in cycle, at each loop to ask the feedback from users on the output, then using it to refine the searching procedure. However, focusing too much on user feedback may be harmful as it shadows the importance of feature design on this task, especially in earlier loops when less information from users are available. Regarding many literatures since years on feature extraction for CBIR, it is natural to raise a question: "Is the feature design is at the top of its performance?". Recently, the study on Searching Images with MPEG-7 (& MPEG-7-like) Powered Localized dEscriptors (SIMPLE) has shown that the answer is "No". Convinced by this work but without ignoring its limitations, in this article, we aim to show that, by carefully fine-tuning a SIMPLE based system based on deep analyses, the performance can still be improved significantly. At the end, using Gaussian distribution to detect interest regions and using Jensen-Shannon Divergence as the distance measure seems to be the key points to bootstrap the system performance. The significant relative Mean Average Precision improvements on UCID (5-7%) and UKBench (24-36%), the two well-known datasets for CBIR, make this conclusion soundness.

ABSTRACT. The fuzzy co-clustering algorithms are used to solve the problem clustering of large data, multi-dimensional, multi-feature. When the dimensionality and size of data increases, the size of the membership function matrix increases, the amount of calculation increases, computational problems become a burden for the computer. In fuzzy co-clustering algorithms, multiple rules and sub-algorithms may be executed in parallel by the GPU to help speed overall for the whole system. This paper studies the structure and fuzzy co-clustering algorithms and proposed a new solution how to implement fuzzy co-clustering algorithms on GPU. GPU-based calculations are solutions to improve computational efficiency and independent operation for the CPU. The experimental results are conducted for color image classification shows performed on the GPU performance is much higher than executing on the CPU.

ABSTRACT. This paper investigates the energy-aware virtual machine (VM) allocation problems in clouds along characteristics: multiple resources, fixed interval time and non-preemption of virtual machines. Many previous works have been proposed to use a minimum number of physical machines; however, this is not necessarily a good solution to minimize total energy consumption in the VM placement with multiple resources, fixed interval time and non-preemption. We observed that minimizing the sum of total busy time of all physical machines implies minimizing total energy consumption of physical machines. In addition to, if mapping of a VM onto physical machines have same total busy time then the best mapping has physical machine’s remaining available resource minimizing. Based on these observations, we proposed EM algorithm to solve the energy-aware VM allocation with fixed starting time and duration time. In addition, this work studies some heuristics for sorting the list of virtual machines (e.g., sorting by the earliest starting time, or the longest duration time first, etc.) to allocate VM. Using realistic log-traces in the Feitelson’s Parallel Workloads Archive, our simulation results show that the EM-ST, EM-LFT, EMLDTF with/without swapping algorithms could reduce from 21% to 39% total energy consumption compared with a state-of-the-art power-aware VM allocation algorithms that are the power-aware best-fit decreasing (PABFD) [1]), and another modified best-fit decreasing (BFD-ST) with sorting the list of VMs by their starting time.

ABSTRACT. Cloud computing and storage solutions enable end users and entreprises to store their data and process their services in third party data centers. The fast growth of data center in size and in number makes them huge energy consumption points. Some researches claim that 70% energy for a data center focus on server and cooling. That is why, recent researches proposed to turn off certain switches in data center where little traffic flowing through. However, when those switches are turned off, the structure of the data center becomes vulnerable to failures due to low connectivities between servers. This paper aims to overcome this weakness by proposing a way to build a data center where unnecessary switches can still be turn off while the data center remains survivable upon any single failure. The simulation results show that the proposed solution makes data center survivable while still saving a lot of energy.

ABSTRACT. In this paper, we consider the problem of placing a set of transmitters, of bounded total cost, in small area wireless networks with a set of receivers each of which of a predefined non-negative weight. The problem asks to determine and place a set of transmitters in suitable candidate sites such that a set of conflicting objectives are met under cost constraints on transmitters. Namely, we look for transmitter placement that maximizes the total weights of covered receivers, minimizes the total capacitance, and minimizes the overlap in the network. This problem is solved as multi-objective optimization problem by applying a Multi-Objective Evolutionary Algorithm Based on Decomposition Algorithm. Our experimental results show that the proposed algorithm has a good performance in a reasonable running time.

ABSTRACT. In an OBS network, each core node must use a certain scheduling scheme to allocate resources for an arriving burst. In the case of failure, rescheduling and burst segmentation approaches could be invoked to reduce data loss. This paper proposes an improved composite scheduling approach (iCSA), which is based on scheduling, rescheduling and burst segmentation. Analysis and simulation results demonstrate that the iCSA is better than the existing ones.

ABSTRACT. TimeTabling is a classical problem in combinatorial optimization. There are many variants of the problem depending on specific requirements of the given institution. In this paper, we describe a new variant of the timetabling problem arising in FPT university called FUTimeTabling. We propose an algorithm for solving FUTimeTabling problem. We conduct expriments on a real data test which show the feasibility of the proposed technique.

ABSTRACT. This paper introduces a practical hybrid transportation model for Tokyo city that allows a passenger and parcels are handled in a same taxi. We inherit the recent model given by Li et al. in 2014 and make it much more realistic by adding some constraints related to the real-life case. We propose the time-dependent model to facilitate formulating constraints. The feasibility and the efficacy of the model are proved by two proposed heuristic algorithms. Especially, we use the real-case experimental data set recorded by Tokyo-Musen Taxi company. The data set includes more than 20,000 requests per day, more than 4,500 served taxis per day and more than 130,000 crossing points on the Tokyo map. The experimental results are analyzed on various factors such as the total benefit, the accumulating distances during the day and the number of shared taxis.

ABSTRACT. Constraint-Based Local Search (CBLS) is an architecture for local search that uses constraints and objectives to control the local search. CBLS has many advantages in designing and implementing local search programs such as compositionality, modularity and reusability. We implement in this paper a Java library for CBLS. The implemented library will then be applied to the resolution of a combinatorial optimization problem arising in the education management in most of Vietnamese universities: the master thesis defense timetabling problem. Experimental results show the modelling flexibility and the efficiency of the constructed library.

ABSTRACT. Learner-centered learning methods have been considered for school education settings to help improve problem-solving skills of learners. Collaborative learning (CL), an approach in which two or more participants attempt to solve a problem together, has proven to be effective. Ambient support, which does not interrupt face-to-face interaction, is necessary in computer-supported collaborative learning or computer-supported cooperative work. Particular displays for such support are used in numerous recently released support systems. Much information, such as the number of utterances of learners and the degree of contribution to a discussion during the learning process, is represented on the displays by these systems; however, such support is not ambient because some participating learners watched the displays instead of focusing on other learners. In addition, support content for each learner is broadcast by these systems. According to research results, such broadcasting displeases some students; therefore, such support is not appropriate. In our research, we have aimed to develop an ambient conversation support system (ACSS) to solve these problems. First, we compared support methods of recently released systems (i.e., public conditions) with our developed approach (i.e., private conditions). Our experimental results indicate that (1) the average speech time per utterance under private conditions was statistically shorter than that of public conditions and (2) this average speech time per utterance under public conditions was statistically shorter than that of conditions in which participants were not supported by an ACSS (i.e., the no-support conditions). Experimental results and our discussion suggest that (1) the ACSS had a positive effect in support of increasing the number of participant utterances and (2) private conditions might pressure each participant against our expectations.

ABSTRACT. This paper addresses a challenging single-channel speech enhancement problem in real-world environment where speech signal is corrupted by high level background noise. While most state-of-the-art algorithms tries to estimate noise spectral power and filter it from the observed one to obtain enhanced speech, the paper discloses another approach inspired from audio source separation technique. In the considered method, generic spectral characteristics of speech and noise are first learned from various training signals by non-negative matrix factorization (NMF). They are then used to guide the similar factorization of the observed power spectrogram into speech part and noise part. Additionally, we propose to combine two existing group sparsity-inducing penalties in the optimization process and adapt the corresponding algorithm for parameter estimation based on multiplicative update (MU) rule. Experiment result over different settings confirm the effectiveness of the proposed approach.

ABSTRACT. We propose a new technique for removing shadow regions based on a learning-based approach. We extract two different types of the examined shadow regions, such as chromacity-based features and physical properties. Two likelihoods or shadow-matching scores are calculated from corresponding features. However, it is different from existing techniques, we take into account a density-based score fusion scheme. A likelihood ratio of shadow per nonshadow score is calculated. Probabilities of shadow and nonshadow are estimated based on approximating distributions of the shadow-matching scores using Gaussian Mixture Models. The experimental results confirm that the proposed fusion scheme outperforms existing techniques which utilize the separated features. We then deploy the proposed technique in framework of person localization system in indoor environments. Averagely, positioning error is reduced from 44.4+/-33.1(cm) (without shadow removal) to 13.5+/-18.8(cm) (with the proposed shadow removal technique). Consequently, this work contributes an effective preprocessing step in order to deploy vision-based localization services in surveillance environments.

ABSTRACT. Holes in sensor network are regions without operating nodes which may occur due to several reasons, including cases caused by natural obstacles or disaster suffered areas. Determining the location and shape of holes can help to monitor these disaster events (such as volcano, tsunami, etc.) or help to make smart, early routing decisions for circumventing a hole. There are many hole determination algorithms have been proposed in the literature. However, all of these algorithms only consider the networks with static holes. Moreover, most of these protocols are conducted in the centralized manner which is not suitable with the resource constraint of the sensor nodes. In this paper, we propose algorithms which can not only determine the hole but also quickly update the hole when it is enlarged. The algorithms are conducted in a distributed manner. We also conduct simulation to evaluate performance of proposed protocols.

ABSTRACT. The paper proposes a new approach to model the insects surveillance network as synchronous network systems, systems consist of components running simultaneously. In the network, insects and surrounding environments compose a physical system of which executions proceed concurrently in synchronous rounds. This system is synchronized with a synchronous wireless sensor network, the observation network. The simulation of the case study Brown Planthoppers surveillance network in the Mekong Delta provides some useful information in managing pest insects as well as emerges the roles of wireless sensor network in monitoring environments.

ABSTRACT. In recent years, Wireless Sensor Networks (WSNs) have proved their power, but they also faced many practical challenges. One of such challenges is covering issue. This paper considers the maximum coverage deployment problem in WSNs. Particularly, with the given number of sensors having various sensing ranges, how to deploy these sensors in a specified domain so that their coverage on which is maximum. This is a NP-hard problem. We propose a new genetic algorithm with some improvements compared to an existing genetic one. These improvements include the definition of a new concept – the overlapping – for the fitness function, using a heuristic technique to initialize population and a dynamic mutation. Our algorithm is experimented on 15 instances constructed for this problem. The experimental results show that our proposed algorithm is effective in all terms of the computational complexity, quality of solutions and stability

ABSTRACT. Nowadays, lung diseases, such as asthma, chronic obstructive pulmonary disease (COPD), and cystic fibrosis, are increasingly attracting the attention of the world medical community. The main reason is due to the increasing environmental pollution problem. Therefore, detecting pulmonary exacerbations and improving outcomes of chronic lung ailments are in urgent need. Traditionally, patients have to go to pulmonary hospital and use conventional spirometers. However, the challenge of this solution is the cost and the loss of time. Patients may take several hours to queue to see a doctor and have to pay for medical expenses. In order to solve this inconvenience, we proposed an approach, called BKSpiro, for measuring lung function using Android phone, a very popular device for everyone today. BKSpiro has the potential for more rapid recovery, earlier treatment of exacerbations, and reduce health care costs.

ABSTRACT. Many learning techniques have been applied to identify disease-associated genes. At the early, they usually approached this problem as a binary classification, where training set is comprised of positive and negative samples. In which, positive samples are constructed from known disease genes, whereas negative samples are the remaining which are not known to be associated with diseases. This is the limitation of the binary classification-based solutions, since the negative training set should be actual non-disease genes; however, construction of this set is nearly impossible in biomedical researches. Therefore, to reduce this uncertainty, more realistic classification-based methods have been proposed. For instance, unary classification technique based on one-class SVM method was proposed by learning from only positive samples. In addition, the remaining set may contain unknown disease genes; therefore, semi-supervised methods such as binary semi-supervised and positive and unlabeled (PU) learning classifications have been proposed. In particular, PU learning methods, which learn from both known disease genes and the remaining genes, were shown to outperform others. In these studies, data sources are usually represented by vectorial form for binary classifiers, while they are in kernel matrices for unary and PU learning ones. The kernel-based data fusion is only suitable for data with different types and it seems unfair for the comparison based on different data representations. Therefore, in this study, we compared different classification techniques for the disease gene prediction based on vectorial representation of samples. The simulation result showed that the unary classification technique, which combines both density and class probability estimation strategies, achieved the best performance, whereas it is worst for the one-class SVM-based method. Interestingly, performance of the best binary classification technique is comparable with that of biased SVM-based PU learning and binary semi-supervised classification methods. And, they are all better than the multi-level SVM-based one.

ABSTRACT. Our main purpose is to build a simulation describing the persons movements in a public area during a crisis situation. Our running example is about the evacuation process of a supermarket during a fire. The main difficulty of such works comes from the fact that it is generally impossible to obtain precise descriptions of persons behaviors. After several works on individual emotions we are convinced that emotion is well adapted to explain such actions in a situation crisis. It concerns particularly fear but not only. Moreover, an increasing number of other works are already concerned by emotion in situation crisis. In the aim to test our hypothesis, we present here a first simulation (without any emotion management) that is a first step of our main purpose. This simulation aims to describe the well-known fact that, in crisis situation, humans tend toward to help each others. Thus, we test the impact of group constitution on the number of survival. The simulation has been implemented with the platform GAMA.

ABSTRACT. In this paper, we propose a framework for logo spotting by using local features. We present key-point matching methods to match local features of logo images to those of document images. For segmentation, a density-based clustering method is used to group matches and remove the outliers. Then, a two-stage algorithm based on homography with RANSAC is used to verify and localize the spotting results. In our experiments, many kinds of local features are employed and compared for their effectiveness. The results show that SIFT and SURF outperform in terms of accuracy.

ABSTRACT. Detection of buildings in aerial images is an important and challenging task in computer vision and aerial image interpretation. This paper presents an efficient approach that combines Random forest (RF) and a fully connected conditional random field (CRF) on various features for the detection and segmentation of buildings at pixel level. RF allows one to learn extremely fast on big aerial image data. The unary potentials given by RF are then combined in a fully connected conditional random field model for pixel-wise classification. The use of high dimensional Gaussian filter for pairwise potentials makes the inference tractable while obtaining high classification accuracy. Experiments have been conducted on a challenging aerial image dataset from a recent ISPRS Semantic Labeling Contest \cite{ISPRSdataset}. We obtained state-of-the-art accuracy with a reasonable computation time.

ABSTRACT. Nowadays, Linked Open Data (LOD) has been grown rapidly to become large open datasets defined by RDF standards. Thanks to development of Data Web, the information on LOD is increasingly deeper, larger and easier to link in multi-domains, constituting the Linked Open Data cloud. Recently end-users applications using linked data sources as background knowledge appeared [1]. Thus, the exploitation of information on LOD effectively brings tremendous values as well as challenges. Meanwhile, Exploratory Search (ES) describes information-seeking processes that are opportunistic, iterative, and multi-tactical [20]. Furthermore, systems based on ES capitalize on new technological capabilities and interface paradigms that facilitate an increased level of interaction with information. In this paper, we present a method on searching and recommending information to empower exploratory search with semantic technologies. Our aim is to use algorithms with incorporation of structured semantics in search to give users the best related-semantic results and enhance users’ interactions. We have exploited the data within LinkedMDB to support users in finding some information in the movie domain.

ABSTRACT. Even though Petri Nets (PNs) are powerful for modelling concurrent systems, they suffer from a high computational cost when verifying properties on the modelled system. Abstraction is one of the efficient approaches to tackle this problem. In this paper, taken into account the fact that most real life systems are made from basic components, we suggest a component-based approach for abstraction and verification of a PN-based model. In our approach, a PN system can be considered as a set of connected components. For instance, a Wireless Sensor Network (WSN) includes a Sensor component and a Channel component. When performing verification on a model, we can abstract the insignificant components and only focus on the remaining ones. We illustrate the advantages of our approach on a case study of congestion detection on a WSN by means of model checking. Experiments demonstrate that our approach is practically feasible when tested with various kinds of congestion occurring in different PN-modelled WSNs.

ABSTRACT. Using the protocol IEEE 802.11, multi-hop wireless ad hoc networks yield only a poor performance, especially in throughput and fairness. When the offered load becomes large, i.e. the system is in saturation state, long-distance flows suffer a high degree of throughput deterioration. These problems not only come from medium contention at the MAC layer but are due to the link layer. In this paper, we propose a method to solve the throughput degradation and unfairness problem by providing fair treatment between flows at the link-layer. In our proposed method, a fair scheduling algorithm using round robin queue and the estimation for the average interval of packet enqueueing is applied to alleviate the unfairness problem at both MAC and link layer. The simulation results reveal that our proposed method is able to achieve better throughput and fairness compared to the standard IEEE 802.11.

ABSTRACT. Multimedia communications over mobile ad hoc networks (MANET) draw much attention from the research community. However, realization of such distributed applications, including session control functionality, still faces numerous hurdles: node dynamism, unstable wireless channels, etc. Conventional SIP is not applicable since using a particular mobile endpoint as a centralized server is unrealistic. To date, there have been reported studies in session control protocols aiming at MANET, but results are mainly presented in form of simulation statistics or theoretical analysis. In this study, we introduce a joint route discovery and distributed session control strategy that has been verified via our testbed experiments. Ad hoc nodes disseminate routing messages that also carry node identity for later session setup, which allows to setup multimedia sessions without querying any centralized server. To lower communication overhead, we also propose a distributed searching scheme in which each node maintains a routing and identity database of partial set of network nodes only. Upon a session request arrives, surrounding nodes help pass the request for searching until the identity of the called node is discovered. The experiments we deployed demonstrate the soundness of the proposed approach.