Abstract:Recently, ontology learning is emerging as a new hotspot of research in computer science. In this paper the issue of ontology learning is divided into nine sub-issues according to the structured degree (structured, semi-structured, non-structured) of source data and learning objects (concept, relation, axiom) of ontology. The characteristics, major approaches and the latest research progress of the nine sub-issues are summarized. Based on the analysis framework proposed in the paper, existing ontology learning tools are introduced and compared. The problems of current research are discussed, and finally the future directions are pointed out.

Abstract:In recent years, there have been extensive studies and rapid progresses in automatic text categorization, which is one of the hotspots and key techniques in the information retrieval and data mining field. Highlighting the state-of-art challenging issues and research trends for content information processing of Internet and other complex applications, this paper presents a survey on the up-to-date development in text categorization based on machine learning, including model, algorithm and evaluation. It is pointed out that problems such as nonlinearity, skewed data distribution, labeling bottleneck, hierarchical categorization, scalability of algorithms and categorization of Web pages are the key problems to the study of text categorization. Possible solutions to these problems are also discussed respectively. Finally, some future directions of research are given.

Abstract:An efficient digital ink data coding algorithm IWPHSP (integer wavelet packet based hierarchical set partitioned) is proposed in this paper. The algorithm compresses digital ink multi-dimension data losslessly using three approaches: integer wavelet packet transform, hierarchical set partitioned, significant bits combination code and fast adaptive arithmetic code. The experiments show that the IWPHSP algorithm is efficient.

Abstract:Multi-Modality fusion is one of the hottest discussed issues in the current research of medical image processing and it has a deep impact on the cognitive science and clinical treatment. In this paper, an fMRI-constraint equivalent dipole model (FC-ECD) based on ICA is proposed to solve the fusion of fMRI and EEG. The ICA is adopted as a preprocessing step to exclude the noise and select the available ERP components. At the same time, it can provide a prior estimate of the number of dipoles. Then considering the spatial information provided by fMRI, the selected ERP components are localized by FC-ECD model based on an ideal four-sphere head model. Thus it can reduce the computation time dramatically. Finally, the simulation study proves the correctness and validity of the method proposed in the paper and the human study coincides with the physiology fact.

Abstract:Use of structural information and lexicalization are two of the main challenges facing syntactic analysis, and they are investigated in this paper. First, the probabilities of lexical dependencies are obtained by training a large-scale dependency treebank and used to build the lexical model. Second, the governing degree of words is introduced to utilize the structure information. The lexical method overcomes the weakness of POS dependencies in the past work; meanwhile the governing degree of words is helpful to distinguish the syntactic structures so some ill-formed structures are avoided. Finally, the paper shows a good experimental result of around 74% accuracy on the test set that consists of 4000 sentences.

Abstract:Enlightened by the behaviors of gregarious ant colonies, an artificial ant movement (AM) model and an adaptive ant clustering (AAC) algorithm for this model are presented. In the algorithm, each ant is treated as an agent to represent a data object. In the AM model, each ant has two states: sleeping state and active state. In the algorithm AAC, the ant’s state is controlled by both a function of the ant’s fitness to the environment it locates and a probability function for the ants becoming active. By moving dynamically, the ants form different subgroups adaptively, and consequently the whole ant group dynamically self-organizes into distinctive and independent subgroups within which highly similar ants are closely connected. The result of data objects clustering is therefore achieved. This paper also present a method to adaptively update the parameters and the ants’ local movement strategies which greatly improve the speed and the quality of clustering. Experimental results show that the AAC algorithm on the AM model is much superior to other ant clustering methods such as BM and LF in terms of computational cost, speed and quality. It is adaptive, robust and efficient, and achieves high autonomy, simplicity and efficiency. It is suitable for solving high dimensional and complicated clustering problems.

Abstract:The navigation problem of multi-robot movement in a complex and unknown environment is studied in the paper. A new algorithm, ants navigation algorithm, is presented. At the start the method maps the global targets onto the area near the border of the robot’s eyeshot, and takes them as the local targets. Then two groups of ants will be cooperating to complete the search for the local optimal path in the robot’s eyeshot. Based on these configurations, the algorithm can predict possible collision with other robots and execute subsequent avoidance plans. The local search will be executed by the algorithm repetitively whenever the robot progresses a step. So, the path of the robot will be altered dynamically, which makes the robot move on the global optimal path to the ending node. The simulation results indicate that the optimal path, which the robot moves on, can lead the robot to reach the end safely even in complicated geographical environment. The effect is very satisfactory.

Abstract:This paper presents a new framework for spatiotemporal alignment of two video sequences. It proposes Intra-video and inter-video matching strategy for spatial alignment; modifies Dynamic Time Warping for temporal alignment. Intra-video matching tracks feature points and binds them together. Contextual inter-video matching uses track correspondences to provide initial feature correspondences for inter-video frame matching and updates track correspondences using frame-matching results. The proposed matching strategy makes best use of coherency of source videos and improves coherency of aligned video, stability and efficiency of alignment. The Modified Dynamic Time Warping establishes frame correspondences by minimizing global differences between them, keeps temporal order of frames, and handles nonlinear misalignment of videos. The proposed method can successfully align videos viewing different events recorded by independently moving cameras. Experimental results and comparison show that great improvements on stability and efficiency of video matching together with coherency of aligned video are reached.

Abstract:Texture segmentation is a typical difficult problem in image processing. This paper presents a new textural oscillatory feature based on image decomposition. The oscillatory feature together with other textural features based on the structure tensor and nonlinear diffusion constructs a 5 dimensional textural feature space. The last result can be obtained by segmenting the feature space using level set and non-parametric active contours technology. The validity of the method in this paper is proved by different texture segmentation tests.

Abstract:Gradient vector flow (GVF) snake shows high performance at capture-range enlarging and boundary concavity convergence, however, the initial contours encounter a so-called critical point problem (CPP). The initial contour must contain the critical points inside the object and exclude those outside the object, otherwise, the final result would be far from the expected. This paper investigates the CPP of the GVF snake and points out that, serving as an external force field for snake models, gradient vector flow could be effective only under some restrictions. Also, it is proved that the theoretical foundation, the Navier-Stokes equation for viscous fluid flow, for the solution to this CPP in literatures is incorrect. Finally, an empirical solution to the CPP is presented and its performance is validated by experiments.

Abstract:The inter-domain routing system is a complex macrosystem just like the Internet, and the self-organization theory is the efficient utility for studying complex system. This paper analyzes the intrinsic rules and behavioral exhibitions of inter-domain routing system from the view of self-organization, and evaluates the mending methods to BGP protocol for improving the scalability, convergence, stability and s