Image processing systems have typically exhibited a high degree of application specificity. This makes it unlikely that any of the processing tasks for one application can be applied to another. Research has also found that such systems had not made use of knowledge about human perception although, it can be argued, such knowledge is independent of the application. This thesis takes the view that the structures which people perceive in pictures are of considerable importance to supporting picture-based human computer interaction. However, the design of systems which process pictures based on knowledge of human perception presents two major challenges. The first relates to determining the appropriate psychological knowledge that the computational modelling effort will be based upon. The second relates to the testing and evaluation of systems that aim to generate perceptually-valid structures, which poses significant problems because such systems are concerned with perceptual structures rather than semantic interpretations. An approach to developing picture processing systems is proposed that overcomes these problems. The approach represents a development strategy that exploits psychological theory in constructing image processing algorithms. The aim of these algorithms is to generate structures that match those perceived by humans. The approach enables new algorithms to be developed as additional psychological knowledge becomes available. Given appropriate training, users of such systems are able to see in images the structures that are generated by the system. The thesis presents the results of a study of perceptual and computational approaches to the processing of pictures. Motivations for and the development of a computational model based on perceptual grouping are described. A thorough evaluation of the computational model is performed which highlights both its value and limitations. Specific applications of the computational model in areas such as shape emergence in design, and pictorial databases, are then considered.

No description available.

The use and dissemination of remotely-sensed data is an important resource that can be used for environmental, commercial and educational purposes. Because of this, the use and availability of remotely-sensed data has increased dramatically in recent years. This usefulness, however, is often overshadowed by the difficulty encountered with trying to deal with this type of data. The amount of data available is immense. Storing, searching and retrieving the data of interest is often difficult, time consuming and inefficient. This is particularly true when these types of data need to be rapidly and continually accessed via the Internet, or combined with other types of remotely-sensed data, such as combining Aerial Photography with US Census vector data. This thesis addresses some of these difficulties, a two-fold approach has been taken. First, a database schema which can store various types of remotely-sensed data in one database has been designed for use in a Semantic Object-Oriented Database System (Sem-ODB). This database schema includes in its design a linear addressing scheme for remotely-sensed objects which maps an object’s 2-dimentional (latitude/longitude) location information to a 1-dimensional integrated integer value. The advantages of using this Semantic schema with remotely-sensed data is discussed and the use of this addressing scheme to rapidly search for and retrieve point-based vector data is investigated. In conjunction with this, an algorithm for transforming a remotely-sensed range search into a number of linear segments of objects in the 1-dimensional array is investigated. The main issues and the combination of solutions involved are discussed.

Master of Science

The Naval Message Analyzer was developed to show that a distributed computerized message-analysis system could be built using the CODER (COmposite Document Expert!/extended/effective Retrieval) system, a testbed for artificial intelligence methods in Information Storage and Retrieval, as the foundation of the project. The Naval Message Analyzer reads messages contained in disk files and indexes them according to dates, locations and words. Words are used as indexes to documents according to headword entries in the Collins English Dictionary. Dates, sources and destinations are read from the header information as well as the body of each message. Header information is in one standard format, but other information is detected using clues within the text, and is further analyzed by expert modules. A database of world-knowledge is included as a repository of proper names for ships, cities, countries and abbreviations. A database maintenance facility is an integral part of the Naval Message Analyzer, so that facts can be added or changed as necessary. The system retrieves the closest matches, in descending order, to one or more ship names, locations, dates and keywords provided by the user. / Master of Science

No description available.

This thesis is devoted to designing and analyzing efficient text indexing data structures and associated algorithms for processing text data. The general problem is to preprocess a given text or a collection of texts into a space-efficient index to quickly answer various queries on this data. Basic queries such as counting/reporting a given pattern's occurrences as substrings of the original text are useful in modeling critical bioinformatics applications. This line of research has witnessed many breakthroughs, such as the suffix trees, suffix arrays, FM-index, etc. In this work, we revisit the following problems: 1. The Heaviest Induced Ancestors problem 2. Range Longest Common Prefix problem 3. Range Shortest Unique Substrings problem 4. Non-Overlapping Indexing problem For the first problem, we present two new space-time trade-offs that improve the space, query time, or both of the existing solutions by roughly a logarithmic factor. For the second problem, our solution takes linear space, which improves the previous result by a logarithmic factor. The techniques developed are then extended to obtain an efficient solution for our third problem, which is newly formulated. Finally, we present a new framework that yields efficient solutions for the last problem in both cache-aware and cache-oblivious models.

Processing a vast volume of data generated by web, mobile and Internet-enabled devices, necessitates a scalable and flexible data management system. Database-as-a-Service (DBaaS) is a new cloud computing paradigm, promising a cost-effective and scalable, fully-managed database functionality meeting the requirements of online data processing. Although DBaaS offers many benefits it also introduces new threats and vulnerabilities. While many traditional data processing threats remain, DBaaS introduces new challenges such as confidentiality violation and information leakage in the presence of privileged malicious insiders and adds new dimension to the data security. We address the problem of building a secure DBaaS for a public cloud infrastructure where, the Cloud Service Provider (CSP) is not completely trusted by the data owner. We present a high level description of several architectures combining modern cryptographic primitives for achieving this goal. A novel searchable security scheme is proposed to leverage secure query processing in presence of a malicious cloud insider without disclosing sensitive information. A holistic database security scheme comprised of data confidentiality and information leakage prevention is proposed in this dissertation. The main contributions of our work are: (i) A searchable security scheme for non-relational databases of the cloud DBaaS; (ii) Leakage minimization in the untrusted cloud. The analysis of experiments that employ a set of established cryptographic techniques to protect databases and minimize information leakage, proves that the performance of the proposed solution is bounded by communication cost rather than by the cryptographic computational effort.

Modern information era gives rise to the persistent generation of large amounts of data with rapid speed and broad geographical distribution. Obtaining knowledge and understanding via analysis and learning from such data have invaluable worth. Features of such data analytical tasks commonly include: data can be large scale and geographically distributed; computing capability demand can be enormous; tasks can be time-critical; some data can be private; participants can have heterogeneous capabilities and non-IID data; and multiple simultaneously submitted data analytical tasks can be possible. These bring challenges to contemporary computing infrastructure and learning models. In view of this, we develop techniques with the purpose of tackling above challenges together towards more efficient collaborative distributed data analysis and learning. We propose a hierarchical framework that supports data analytics on multiple Apache Spark clusters. We propose reinforcement learning based resource management approaches to improve overall efficiency and reduce deadline violations for scheduling general and time-critical data analytical workflows among computing resources. We establish a new hybrid framework for efficient privacy-preserving federated learning and further propose an algorithm upon it for improving asynchronous federated learning of heterogeneous participants having non-IID data. We also propose an asynchronous stochastic gradient descent algorithm for general distributed learning of heterogeneous participants having non-IID data with convergence analysis. Experiments have shown the efficacy of our proposed approaches.

Thesis (M. A. (Information Studies)) -- University of Limpopo, 2020 / This study examined whether postgraduate students in the Faculty of Humanities at the University of Limpopo are aware of, and are using electronic databases optimally to locate information for their academic research. The study employed a quantitative research design through the use of a questionnaire as a data collection method to determine students’ accessibility and levels of awareness of electronic databases; to measure the extent to which they use electronic databases for academic research; to assess the form of training which they attended on the use of electronic databases; to identify factors that determine their usage and non-usage of electronic databases; and to establish challenges that they encounter in using electronic databases. The accidental sampling method was used to select the participants. The study found that most participants are aware of the existence of electronic databases. Even if they indicated to have used some of these databases at UL library, it appears that they are referring to Google and Google Scholar. This is despite the fact that the majority of them showed to have attended some sort of training on the use of electronic databases. Factors that influence their choice and use of specific electronic databases include familiarity, unlimited access, multidisciplinary as well as their capabilities. Problems and challenges encountered in the use of electronic databases are related to remote access. It becomes difficult for them to access these databases when they are not on campus. Lack of knowledge and skills to search these electronic databases effectively hindered their optimal usage. Other problems identified by postgraduate students include: slow internet connectivity; inadequately networked computers; lack of access to low-cost printing facilities in the library; inability to use advanced search strategies on most databases; and a lack of awareness of most e-resources. It is recommended that studies of this nature should be conducted in other faculties so that the library should identify electronic databases that are not used and to consider cancelling subscriptions to unused electronic databases in order to save costs.