"As streaming techniques and wireless access networks become more widely deployed, a streaming multimedia connection with the "last mile" being a wireless network is becoming increasingly common. However, since current streaming techniques are primarily designed for wired networks, streaming multimedia applications can perform poorly in wireless networks. Recent research has shown that the wireless network conditions, such as the wireless link layer rate adaptation, contending traffic, and interference can significantly degrade the performance of streaming media applications. This performance degradation includes increased multimedia frame losses and lower image quality caused by packet loss, and multiple rebuffering events that stop the media playout. This dissertation presents the model, design, implementation and evaluation of an application layer solution for improving streaming multimedia application performance in IEEE 802.11 wireless networks by using enhanced bandwidth estimation techniques. The solution includes two parts: 1) a new Wireless Bandwidth estimation tool (WBest) designed for fast, non-intrusive, accurate estimation of available bandwidth in IEEE 802.11 networks, which can be used by streaming multimedia applications to improve the performance in wireless networks; 2) a Buffer and Rate Optimization for Streaming (BROS) algorithm using WBest to guide the streaming rate selection and initial buffer optimization. WBest and BROS are implemented and incorporated into an emulated streaming client-server system, Emulated Streaming (EmuS), in Linux and evaluated under a variety of wireless conditions. The evaluations show that with WBest and BROS, the performance of streaming multimedia applications in wireless networks can be significantly improved in terms of multimedia frame loss, rebuffer events and buffer delay."

The phenomenal growth in the use of the World Wide Web often places a heavy load on networks and servers, threatening to increase Web server response time and raising scalability issues for both the network and the server. With the advances in the field of optical networking and the increasing use of broadband technologies like cable modems and DSL, the server and not the network, is more likely to be the bottleneck. Many clients are willing to receive a degraded, less resource intensive version of the requested content as an alternative to connection failures. In this thesis, we present an adaptive content delivery system that transparently switches content depending on the load on the server in order to serve more clients. Our system is designed to work for dynamic Web pages and streaming multimedia traffic, which are not currently supported by other adaptive content approaches. We have designed a system which is capable of quantifying the load on the server and then performing the necessary adaptation. We designed a streaming MPEG server and client which can react to the server load by scaling the quality of frames transmitted. The main benefits of our approach include: transparent content switching for content adaptation, alleviating server load by a graceful degradation of server performance and no requirement of modification to existing server software, browsers or the HTTP protocol. We experimentally evaluate our adaptive server system and compare it with an unadaptive server. We find that adaptive content delivery can support as much as 25% more static requests, 15% more dynamic requests and twice as many multimedia requests as a non-adaptive server. Our, client-side experiments performed on the Internet show that the response time savings from our system are quite significant.

General purpose operating systems have been designed to provide fast, loss-free disk service to all applications. However, multimedia applications are capable of tolerating some data loss, but are very sensitive to variation in disk service timing. Present research efforts to handle multimedia applications assume pessimistic disk behaviour when deciding to admit new multimedia connections so as not to violate the real-time application constraints. However, since multimedia applications are ``soft' real-time applications that can tolerate some loss, we propose an optimistic scheme for admission control which uses average case values for disk access. Typically, disk scheduling mechanisms for multimedia applications reduce disk access times by only trying to minimize movement to subsequent blocks after sequencing based on Earliest Deadline First. We propose to implement a disk scheduling algorithm that uses knowledge of the media stored and permissible loss and jitter for each client, in addition to the physical parameters used by the other scheduling algorithms. We will evaluate our approach by implementing our admission control policy and disk scheduling algorithm in Linux and measuring the quality of various multimedia streams. If successful, the contributions of this thesis are the development of new admission control and flexible disk scheduling algorithm for improved multimedia quality of service.

This dissertation examines the practices of leisure and consumption in early twentieth-century Buenos Aires through both intellectual and mass cultural productions. Using works by authors such as Horacio Quiroga and Roberto Arlt together with articles, images, and texts from the Argentine mass media, I examine how national, social and civic identities were intimately tied to, and were constituted through, technologically mediated leisure practices. Sports and film spectatorship, the reception of radio and the reading popular texts were all activities that opened spaces for the rehearsal of forms of citizenship and encouraged the formation of communities and publics both in line with and contrary to the hegemonic and disciplinary mechanisms of the state.

Another issue studied in the framework is Distance Metric Learning (DML). Learning distance metrics is critical to many machine learning tasks, especially when contextual information is available. To learn effective metrics from pairwise contextual constraints, two novel methods, Discriminative Component Analysis (DCA) and Kernel DCA, are proposed to learn both linear and nonlinear distance metrics. Empirical results on data clustering validate the advantages of the algorithms. / Based on this unified learning framework, a novel scheme is suggested for learning Unified Kernel Machines (UKM). The UKM scheme combines supervised kernel machine learning, unsupervised kernel de sign, semi-supervised kernel learning, and active learning in an effective fashion. A key component in the UKM scheme is to learn kernels from both labeled and unlabeled data. To this purpose; a new Spectral Kernel Learning (SKL) algorithm is proposed, which is related to a quadratic program. Empirical results show that the UKM technique is promising for classification tasks. / In addition to the above methodologies, this thesis also addresses some practical issues in applying machine learning techniques to real-world applications. For example, in a time-dependent data mining application, in order to design a domain-specific kernel, marginalized kernel techniques are suggested to formulate an effective kernel aimed at web data mining tasks. / Last, the thesis investigates statistical machine learning techniques with applications to multimedia retrieval and addresses some practical issues, such as robustness to noise and scalability. To bridge semantic gap issues of multimedia retrieval, a Collaborative Multimedia Retrieval (CMR) scheme is proposed to exploit historical log data of users' relevance feedback for improving retrieval tasks. Two types of learning tasks in the CMR scheme are identified and two innovative algorithms are proposed to effectively solve the problems respectively. / Statistical machine learning techniques have been widely applied in data mining and multimedia information retrieval. While traditional methods; such as supervised learning, unsupervised learning, and active learning, have been extensively studied separately, there are few comprehensive schemes to investigate these techniques in a unified approach. This thesis proposes a unified learning paradigm (ULP) framework that integrates several machine learning techniques including supervised learning; unsupervised learning, semi-supervised learning, active learning and metric learning in a synergistic way to maximize the effectiveness of a learning task. / Within the unified learning framework, this thesis further explores two important challenging tasks. One is Batch Mode Active Learning (BMAL). In contrast to traditional approaches, the BMAL method searches a batch of informative examples for labeling. To develop an effective algorithm, the BMAL task is formulated into a convex optimization problem and a novel bound optimization algorithm is proposed to efficiently solve it with global optima. Extensive evaluations on text categorization tasks show that the BMAL algorithm is superior to traditional methods. / Hoi Chu Hong. / "September 2006." / Adviser: Michael R. Lyu. / Source: Dissertation Abstracts International, Volume: 68-03, Section: B, page: 1723. / Thesis (Ph.D.)--Chinese University of Hong Kong, 2006. / Includes bibliographical references (p. 203-223). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. [Ann Arbor, MI] : ProQuest Information and Learning, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstracts in English and Chinese. / School code: 1307.

Yue Ho-Yin. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2001. / Includes bibliographical references (leaves 129-137). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.iv / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Problem Definition --- p.7 / Chapter 1.2 --- Contributions --- p.8 / Chapter 1.3 --- Thesis Organization --- p.10 / Chapter 2 --- Literature Review --- p.11 / Chapter 2.1 --- "Content-based Retrieval, Background and Indexing Problem" --- p.11 / Chapter 2.1.1 --- Feature Extraction --- p.12 / Chapter 2.1.2 --- Nearest-neighbor Search --- p.13 / Chapter 2.1.3 --- Content-based Indexing Methods --- p.15 / Chapter 2.2 --- Indexing Problems --- p.25 / Chapter 2.3 --- Data Clustering Methods for Indexing --- p.26 / Chapter 2.3.1 --- Probabilistic Clustering --- p.27 / Chapter 2.3.2 --- Possibilistic Clustering --- p.34 / Chapter 3 --- Fuzzy Clustering Algorithms --- p.37 / Chapter 3.1 --- Fuzzy Competitive Clustering --- p.38 / Chapter 3.2 --- Sequential Fuzzy Competitive Clustering --- p.40 / Chapter 3.3 --- Experiments --- p.43 / Chapter 3.3.1 --- Experiment 1: Data set with different number of samples --- p.44 / Chapter 3.3.2 --- Experiment 2: Data set on different dimensionality --- p.46 / Chapter 3.3.3 --- Experiment 3: Data set with different number of natural clusters inside --- p.55 / Chapter 3.3.4 --- Experiment 4: Data set with different noise level --- p.56 / Chapter 3.3.5 --- Experiment 5: Clusters with different geometry size --- p.60 / Chapter 3.3.6 --- Experiment 6: Clusters with different number of data instances --- p.67 / Chapter 3.3.7 --- Experiment 7: Performance on real data set --- p.71 / Chapter 3.4 --- Discussion --- p.72 / Chapter 3.4.1 --- "Differences Between FCC, SFCC, and Others Clustering Algorithms" --- p.72 / Chapter 3.4.2 --- Variations on SFCC --- p.75 / Chapter 3.4.3 --- Why SFCC? --- p.75 / Chapter 4 --- Hierarchical Indexing based on Natural Clusters Information --- p.77 / Chapter 4.1 --- The Hierarchical Approach --- p.77 / Chapter 4.2 --- The Sequential Fuzzy Competitive Clustering Binary Tree (SFCC- b-tree) --- p.79 / Chapter 4.2.1 --- Data Structure of SFCC-b-tree --- p.80 / Chapter 4.2.2 --- Tree Building of SFCC-b-Tree --- p.82 / Chapter 4.2.3 --- Insertion of SFCC-b-tree --- p.83 / Chapter 4.2.4 --- Deletion of SFCC-b-Tree --- p.84 / Chapter 4.2.5 --- Searching in SFCC-b-Tree --- p.84 / Chapter 4.3 --- Experiments --- p.88 / Chapter 4.3.1 --- Experimental Setting --- p.88 / Chapter 4.3.2 --- Experiment 8: Test for different leaf node sizes --- p.90 / Chapter 4.3.3 --- Experiment 9: Test for different dimensionality --- p.97 / Chapter 4.3.4 --- Experiment 10: Test for different sizes of data sets --- p.104 / Chapter 4.3.5 --- Experiment 11: Test for different data distributions --- p.109 / Chapter 4.4 --- Summary --- p.113 / Chapter 5 --- A Case Study on SFCC-b-tree --- p.114 / Chapter 5.1 --- Introduction --- p.114 / Chapter 5.2 --- Data Collection --- p.115 / Chapter 5.3 --- Data Pre-processing --- p.116 / Chapter 5.4 --- Experimental Results --- p.119 / Chapter 5.5 --- Summary --- p.121 / Chapter 6 --- Conclusion --- p.122 / Chapter 6.1 --- An Efficiency Formula --- p.122 / Chapter 6.1.1 --- Motivation --- p.122 / Chapter 6.1.2 --- Regression Model --- p.123 / Chapter 6.1.3 --- Discussion --- p.124 / Chapter 6.2 --- Future Directions --- p.127 / Chapter 6.3 --- Conclusion --- p.128 / Bibliography --- p.129

Luo Bo. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 55-60). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgments --- p.vi / Table of Contents --- p.vii / List of Figures --- p.ix / List of Tables --- p.x / List of Abbreviations --- p.xi / Chapter Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Background --- p.1 / Chapter 1.2 --- Text in Videos --- p.1 / Chapter 1.3 --- Related Work --- p.4 / Chapter 1.3.1 --- Connected Component Based Methods --- p.4 / Chapter 1.3.2 --- Texture Classification Based Methods --- p.5 / Chapter 1.3.3 --- Edge Detection Based Methods --- p.5 / Chapter 1.3.4 --- Multi-frame Enhancement --- p.7 / Chapter 1.4 --- Our Contribution --- p.9 / Chapter Chapter 2 --- Caption Segmentation --- p.10 / Chapter 2.1 --- Temporal Feature Vectors --- p.10 / Chapter 2.2 --- Principal Component Analysis --- p.14 / Chapter 2.3 --- PCA of Temporal Feature Vectors --- p.16 / Chapter Chapter 3 --- Caption (Dis)Appearance Detection --- p.20 / Chapter 3.1 --- Abstract Image Sequence --- p.20 / Chapter 3.2 --- Abstract Image Refinement --- p.23 / Chapter 3.2.1 --- Refinement One --- p.23 / Chapter 3.2.2 --- Refinement Two --- p.24 / Chapter 3.2.3 --- Discussions --- p.24 / Chapter 3.3 --- Detection of Caption (Dis)Appearance --- p.26 / Chapter Chapter 4 --- System Overview --- p.31 / Chapter 4.1 --- System Implementation --- p.31 / Chapter 4.2 --- Computation of the System --- p.35 / Chapter Chapter 5 --- Experiment Results and Performance Analysis --- p.36 / Chapter 5.1 --- The Gaussian Classifier --- p.36 / Chapter 5.2 --- Training Samples --- p.37 / Chapter 5.3 --- Testing Data --- p.38 / Chapter 5.4 --- Caption (Dis)appearance Detection --- p.38 / Chapter 5.5 --- Caption Segmentation --- p.43 / Chapter 5.6 --- Text Line Extraction --- p.45 / Chapter 5.7 --- Caption Recognition --- p.50 / Chapter Chapter 6 --- Summary --- p.53 / Bibliography --- p.55

Yeung Siu Fung. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 52-53). / Abstracts in English and Chinese. / Abstract --- p.ii / Acknowledgement --- p.v / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Multi-Key Encryption Theory --- p.7 / Chapter 2.1 --- Reversible Parametric Sequence --- p.7 / Chapter 2.2 --- Implementation of ARPSf --- p.11 / Chapter 3 --- Multimedia Proxy: Architectures and Protocols --- p.16 / Chapter 3.1 --- Operations to Request and Cache Data from the Server --- p.16 / Chapter 3.2 --- Operations to Request Cached Data from the Multimedia Proxy --- p.18 / Chapter 3.3 --- Encryption Configuration Parameters (ECP) --- p.19 / Chapter 4 --- Extension to multi-level proxy --- p.24 / Chapter 5 --- Secure Multimedia Library (SML) --- p.27 / Chapter 5.1 --- Proxy Pre-fetches and Caches Data --- p.27 / Chapter 5.2 --- Client Requests Cached Data From the Proxy --- p.29 / Chapter 6 --- Implementation Results --- p.31 / Chapter 7 --- Related Work --- p.40 / Chapter 8 --- Conclusion --- p.42 / Chapter A --- Function Prototypes of Secure Multimedia Library (SML) --- p.44 / Chapter A.1 --- CONNECTION AND AUTHENTICATION --- p.44 / Chapter A.1.1 --- Create SML Session --- p.44 / Chapter A.1.2 --- Public Key Manipulation --- p.44 / Chapter A.1.3 --- Authentication --- p.45 / Chapter A.1.4 --- Connect and Accept --- p.46 / Chapter A.1.5 --- Close Connection --- p.47 / Chapter A.2 --- SECURE DATA TRANSMISSION --- p.47 / Chapter A.2.1 --- Asymmetric Reversible Parametric Sequence and En- cryption Configuration Parameters --- p.47 / Chapter A.2.2 --- Bulk Data Encryption and Decryption --- p.48 / Chapter A.2.3 --- Entire Data Encryption and Decryption --- p.49 / Chapter A.3 --- Secure Proxy Architecture --- p.49 / Chapter A.3.1 --- Proxy-Server Connection --- p.49 / Chapter A.3.2 --- ARPS and ECP --- p.49 / Chapter A.3.3 --- Initial Sever Encryption --- p.50 / Chapter A.3.4 --- Proxy Re-Encryption --- p.51 / Chapter A.3.5 --- Client Decryption --- p.51 / Bibliography --- p.52

Abstract not available

Although the Internet is the dominant broadband network, it still has a fundamental shortcoming in traffic management that does not properly support efficient use of resources together with differentiated quality of service for mixed traffic types. Even though a number of mechanisms have been proposed under key approaches of resource adaptation, resource reservation, and resource pricing, this problem remains unsolved to date because of its complexity and the way it relates to so many considerations of engineering and economics, and the diverse range of services desired by users. Hence it is considered difficult to devise a perfect mechanism that can universally solve this problem. In this respect, this dissertation contributes to insights into potential combinations and trade-offs of key approaches above in some efficient manner for managing traffic and scarce resources in the Internet. The first part of this work considers the combination of relaxed resource reservation and resource pricing schemes for handling the unfair bandwidth distribution problem in soft-bandwidth guarantee services of the Differentiated Services (DiffServ) Internet. We show that this unfairness problem can be handled using a network-user cooperative approach that addresses both individual user satisfaction and global optimization. We propose to add a mechanism based on price-based feedback signaling to DiffServ-capable routers providing Assured Forwarding (AF) Service so that they can offer per-flow signaling. This still allows them to work at an aggregate traffic level, hence the scalability feature of the DiffServ network can be retained. Our proposed mechanism allows the network to offer reliable service assurances via a distributed optimization algorithm, without introducing special protocols or requiring routers to have access of individual user requirements. Moreover, it can provide incentives for users to cooperate so that optimum performance can be accomplished. This approach has broad applicability and is relevant to all types of assured service classes. The second part of this work considers the combination of a specific resource pricing scheme based on a distributed optimization algorithm and multimedia adaptation schemes. Such a capability has not been available because most utility functions of multimedia applications do not meet the concavity constraints required by optimization algorithm. We proposed to overcome this limitation by redefining user utility functions into equivalent discrete forms and using combinatorial (discrete) optimization so that interfacing the resource-pricing scheme can be done in a normal way. However, to obtain feasible solutions in a scalable manner, an agent is included into the scheme for solving the combinatorial (discrete) optimization on behalf of a small group of users. In return, all users belonging to this group can benefit from social welfare maximization of their utilities over a network resource constraint. The last part of this work considers the combination of relaxed resource reservation and resource pricing schemes to enable a proper charging scheme for adaptive applications in the DiffServ Internet. We provide an optimization formulation of the problems of revenue and social welfare maximization, applied at a service provider (SP) who gives access to the DiffServ Internet. In this scheme, resources are provisioned per QoS-oriented class in a long time scale (service level agreement (SLA) duration), then priced based on user demand in the short time scale. We also show that the proposed charging scheme can provide feedback and incentives for users to use the network resource optimally via the standard packet marking, hence eliminating the need for specific pricing protocol. All of the proposals in this work can be used together, solving these key problems as a coordinated whole, and all use readily available network mechanisms.