Vol. 36 · No. 4

In Description Logics, subsumption is regarded as one of the most prominent reasoning services. It checks, relative to the logical definitions in the ontology, whether one concept is more general/specific than another. When no subsumption relationship is identified, however, no information about the two concepts can be given. In several realistic Semantic Web applications, knowing the level of similarity between two concepts, though lacking the subsumption relationship, is beneficial. This work introduces a new method for measuring the degree of similarity between two concept descriptions in the DL ELH, despite not being in a subsumption relation. Two algorithms are devised based on the known homomorphism-based structural subsumption characterization. The first algorithm employs the top-down approach, whereas the second is carried out in the reverse direction. A bottom-up algorithm has better efficiency, making it more suitable to large-scale ontologies developed using an inexpressive DL in the EL family, such as the renowned medical ontology Snomed ct. The computational performance of the proposed measure is intensively studied, and interesting findings in Snomed ct are reported.

In this paper, we are interested in energy-efficient concurrency control for real-time tasks on a non-ideal DVS processor. Based on well-known ceiling-based concurrency control protocols (such as priority ceiling protocol (PCP) and stack resource policy (SRP)), researchers have proposed energy-efficient approaches to mange concurrent accesses to shared resources so that the energy consumption can be reduced. However, ceiling-based protocols have a problem of ceiling blocking which imposes a great impact on the performance of real-time systems. In order to achieve sufficient performance, we propose a new protocol, called conditional abortable stack resource policy (CA-SRP), to resolve the ceiling blocking problem for dynamic-priority real-time tasks by incorporating a conditional abort rule into SRP. Based on the schedulability analysis of CA-SRP, we also propose a method, called dynamic speed assignment (DSA), to dynamically calculate and assign proper processor speeds for task execution so that the energy consumption can be reduced further. The capabilities of our proposed CA-SRP and DSA have been evaluated by a series of experiments, for which we have encouraging results.

This paper proposes a real-time traffic light detection and recognition algorithm that would allow for the recognition of traffic signals in intelligent vehicles. This algorithm is based on C-HOG features (Color and HOG features) and Support Vector Machine (SVM). The algorithm extracted red and green areas in the video accurately, and then screened the eligible area. Thereafter, the C-HOG features of all kinds of lights could be extracted. Finally, this work used SVM to build a classifier of corresponding category lights. This algorithm obtained accurate real-time information based on the judgment of the decision function. Furthermore, experimental results show that this algorithm demonstrated accuracy and good real-time performance.

Recently feature extraction methods have commonly been used as a principled approach to understand the intrinsic structure hidden in high-dimensional data. In this paper, a novel supervised learning method, called Supervised Sparsity Preserving Projections (SSPP), is proposed. SSPP attempts to preserve the sparse representation structure of the data when identifying an efficient discriminant subspace. First, SSPP creates a concatenated dictionary by class-wise PCA decompositions and learns the sparse representation structure of each sample under the constructed dictionary using the least squares method. Second, by maximizing the ratio of non-local scatter to local scatter, a Laplacian discriminant function is defined to characterize the separability of the samples in the different sub-manifolds. Then, to achieve improved recognition results, SSPP integrates the learned sparse representation structure as a regular term into the Laplacian discriminant function. Finally, the proposed method is converted into a generalized eigenvalue problem. The extensive and promising experimental results on several popular face databases validate the feasibility and effectiveness of the proposed approach.

Cast shadow affects computer vision tasks such as image segmentation, object detection and tracking since objects and shadows share the same visual motion characteristics. This unavoidable problem decreases video surveillance system performance. The basic idea of this paper is to exploit the evidence that shadows darken the surface which they are cast upon. For this reason, we propose a simple and accurate method for detection of moving cast shadows based on chromatic properties in RGB, HSV and YUV color spaces. The method requires no a priori assumptions regarding the scene or lighting source. Starting from a normalization step, we apply canny filter to detect the boundary between self-shadow and cast shadow. This treatment is devoted only for the first sequence. Then, we separate between background and moving objects using an improved version of Gaussian mixture model. In order to remove these unwanted shadows completely, we use three change estimators calculated according to the intensity ratio in HSV color space, chromaticity properties in RGB color space, and brightness ratio in YUV color space. Only pixels that satisfy threshold of the three estimators are labeled as shadow and will be removed. Experiments carried out on various video databases prove that the proposed system is robust and efficient and can precisely remove shadows for a wide class of environment and without any assumptions. Experimental results also show that our approach outperforms existing methods and can run in real-time systems.

A novel binary decision architecture (BDA) for broadcast news audio classification task is presented in this paper. The idea of developing such architecture came from the fact that the appropriate combination of multiple binary classifiers for two-class discrimination problem can reduce a miss-classification error without rapid increase in computational complexity. The core element of classification architecture is represented by a binary decision (BD) algorithm that performs discrimination between each pair of acoustic classes, utilizing two types of decision functions. The first one is represented by a simple rule-based approach in which the final decision is made according to the value of selected discrimination parameter. The main advantage of this solution is relatively low processing time needed for classification of all acoustic classes. The cost for that is low classification accuracy. The second one employs support vector machine (SVM) classifier. In this case, the overall classification accuracy is conditioned by finding the optimal parameters for decision function resulting in higher computational complexity and better classification performance. The final form of proposed BDA is created by combining four BD discriminators supplemented by decision table. The effectiveness of proposed BDA, utilizing rule-based approach and the SVM classifier, is compared with two most popular strategies for multiclass classification, namely the binary decision trees (BDT) and the One-Against-One SVM (OAOSVM). Experimental results show that the proposed classification architecture can decrease the overall classification error in comparison with the BDT architecture. On the contrary, an optimization technique for selecting the optimal set of training data is needed in order to overcome the OAOSVM.

This paper presents a study of an n-gram-based document comparison method. The method is intended to build a large-scale plagiarism detection system. The work focuses not only on an efficiency of the text similarity extraction but also on the execution performance of the implemented algorithms. We took notice of detection performance, storage requirements and execution time of the proposed approach. The obtained results show the trade-offs between detection quality and computational requirements. The GPGPU and multi-CPU platforms were considered to implement the algorithms and to achieve good execution speed. The method consists of two main algorithms: a document's feature extraction and fast text comparison. The winnowing algorithm is used to generate a compressed representation of the analyzed documents. The authors designed and implemented a dedicated test framework for the algorithm. That allowed for the tuning, evaluation, and optimization of the parameters. Well-known metrics (e.g. precision, recall) were used to evaluate detection performance. The authors conducted the tests to determine the performance of the winnowing algorithm for obfuscated and unobfuscated texts for a different window and n-gram size. Also, a simplified version of the text comparison algorithm was proposed and evaluated to reduce the computational complexity of the text comparison process. The paper also presents GPGPU and multi-CPU implementations of the algorithms for different data structures. The implementation speed was tested for different algorithms' parameters and the size of data. The scalability of the algorithm on multi-CPU platforms was verified. The authors of the paper provide the repository of software tools and programs used to perform the conducted experiments.

The quality enhancement of infocommunication systems based on integration of semantically different information transmitted over different communication channels is discussed. The increase of the informational efficiency of the multimodal communication systems is provided by using multimodal subscriber interfaces, compared to traditional telecommunications. The proposed approach is new in the subject area and requires a variety of research studies to develop innovative elements of the theory of multimodal information transmission. A notion of information value is introduced in order to estimate and chose functional structures and parameters that possess the most favorable informational characteristics during development of multimodal communication systems.

Memory allocation is a critical aspect of any modern operating system kernel because it must run continuously for long periods of time, therefore memory leaks and inefficiency must be eliminated. This paper presents different memory management algorithms and their aplicability to an L4-based microkernel. We aim to reduce memory usage and increase the performance of allocation and deallocation of memory.

An efficient allocation of tasks to the processors is a crucial problem in heterogeneous computing systems. Finding an optimal schedule for such an environment is an NP-complete problem. Near optimal solutions are obtained within a finite duration using heuristics/meta-heuristics are used instead of exact optimization methods. Heuristics and meta-heuristics are the efficient technologies for scheduling tasks in distributed environment because of their ability to deliver high quality solutions in a reasonable time. Discrete Particle Swarm Optimization (DPSO) is a newly developed meta-heuristic computation technique. To enhance the final accuracy and improve the convergence speed of DPSO, this paper presents a modified DPSO algorithm by adjusting its inertia weight based on Hamming distance and also makes a dependency between the two random parameters r 1 and r 2 to control the balance of individual's and collective information in the velocity updating equation. Three criteria such as make span, mean flow time and reliability cost are used to assess the efficiency of the proposed DPSO algorithm for scheduling independent tasks on heterogeneous computing systems. Computational simulations are performed based on a set of benchmark instances to evaluate the performance of the proposed DPSO algorithm compared to existing methods.

Currently known fire models are capable to describe fire dynamics in complex environments incorporating a wide variety of fire-related physical and chemical phenomena and utilizing large computational power of contemporary computers. In this paper, some issues related to realization of the simulation of fire in a cinema hall with sloping floor and curved ceiling furnished by upholstered seats modelled by FDS (Fire Dynamics Simulator) are discussed. The paper concentrates particularly on the impact of a computational meshes choice on resolving flow field and turbulence in the simulation and indicates problems related to parallelization of the calculation illustrated comparing sequential and parallel MPI calculation using 6 CPU cores. Results of the simulation described and their discussion demonstrate the ability of FDS simulation to capture main tendencies of smoke spread and to forecast the related safety risks realistically.