The curricular structure and the complexity of the prerequisite dependencies in a curriculum are essential factors that impact student progression, and ultimately graduation rates. However, we are not aware of any closed-form methods for quantifying the relationship between the complexity of a curriculum and the graduation rate of those attempting to complete the curriculum. This paper introduces a new method that quantifies this relationship using Markov Decision Processes (MDP). The non-deterministic nature of student progress along with their evolving states at each semester make MDP a suitable framework for this work. We propose a novel model that is useful due to the fact that it provides a closed-form solution approach that can be utilized to perform “what-if” analyses around student progress through a curriculum. The results confirm the inverse relationship between the complexity of a curriculum and the graduation rate of those students attempting to complete it. This is validated using a Monte Carlo simulation method. The results also provide useful insights that may guide future work in this area.

In this paper, we develop a decentralized probabilistic method for performance optimization of cloud services. We focus on Infrastructure-as-a-Service where the user is provided with the ability of configuring virtual resources on demand in order to satisfy specific computational requirements. This novel approach is strongly supported by a theoretical framework based on tail probabilities and sample complexity analysis. It allows not only the inclusion of performance metrics for the cloud but the incorporation of security metrics based on cryptographic algorithms for data storage. To the best of the authors' knowledge this is the first unified approach to provision performance and security on demand subject to the Service Level Agreement between the client and the cloud service provider. The quality of the service is guaranteed given certain values of accuracy and confidence. We present some experimental results using the Amazon Web Services, Amazon Elastic Compute Cloud service to validate our probabilistic optimization method.

Numerous factors may impact a student's ability to succeed and ultimately graduate, including pre-university preparation, as well as the student support services provided by a university. In this work we study and analyze the impact of such factors on the graduation rates of a university using three predictive models: Support Vector Machines (SVMs), Gaussian Processes (GPs) and Deep Boltzmann Machines (DBMs). We train those models using actual student data. In particular, we used high school GPA, ACT score, gender and ethnicity as the main feature set for training those models. The results show that the DBMs edges out SVMs and GPs in some regards, which has been discussed in detail in the paper, although the difference in performance among the models is negligible with respect to overall accuracies obtained.

Many universities are working to improve their graduation rates. The factors that correlate to student success and hence graduation rates are many, varying from pre-institutional factors, including high school GPA and admissions scores, to institutional factors, including student support services and the quality of faculty. An essential institutional factor that is often overlooked is the structure of the curriculum. In this paper we consider the degree to which the underlying curriculum that a student must traverse in order to earn a degree impacts progress. Using data mining methods, complex network analysis and graph theory, this paper proposes a framework for analyzing university course enrollment networks at the program level. The analyses we provide are based on quantifying the importance of course enrollment sequences on a student's final GPA, a metric that is highly correlated to graduation rates. In particular, we investigate the orderings of courses enrollment sequences that best contribute to student performance and achievement. Experimental results, using data from the University of New Mexico, show that Electrical Engineering students who graduated with "high GPA" values tend to follow a common course enrollment sequence that is quite different than that of students who graduated with relatively "low GPA" values. This work may be useful to both students and decision makers at universities as it presents a robust framework for analyzing the ease of flow of students through curricula, which may lead to improvements that facilitate improved student success.

The field of machine learning strives to develop algorithms that, through learning, lead to generalization; that is, the ability of a machine to perform a task that it was not explicitly trained for. An added challenge arises when the problem domain is dynamic or non-stationary with the data distributions or categorizations changing over time. This phenomenon is known as concept drift. Game-theoretic algorithms are often iterative by nature, consisting of repeated game play rather than a single interaction. Effectively, rather than requiring extensive retraining to update a learning model, a game-theoretic approach can adjust strategies as a novel approach to concept drift. In this paper we present a variant of our Support Vector Machine (SVM) Game classifier which may be used in an adaptive manner with repeated play to address concept drift, and show results of applying this algorithm to synthetic as well as real data.

Security and resource optimization are two of the most critical concerns in cloud computing. A cloud provider must ensure customers with appropriate security, while optimizing the use of cloud resources. In this paper, we present a framework which optimizes both the use of cloud resources and security provided to the customers in an infrastructure as a service (IaaS) cloud. Our framework offers secure usage control of sensitive data within secure virtual machines (VMs), which are dynamically instantiated while optimizing both security and resources allocated to the VMs. These resources are then allocated to the VMs using an optimization model based upon randomized algorithms. We demonstrate that both security and resources can be efficiently optimized within a cloud setting using our formal mathematical model and usage management framework.

Colleges and universities are increasingly interested in tracking student progress as they monitor and work to improve their retention and graduation rates. Ideally, early indicators of student progress, or lack thereof, can be used to provide appropriate interventions that increase the likelihood of student success. In this paper we present a framework that uses machine learning, and in particular, a Bayesian Belief Network (BBN), to predict the performance of students early in their academic careers. The results obtained show that the proposed framework can predict student progress, specifically student grade point average (GPA) within the intended major, with minimal error after observing a single semester of performance. Furthermore, as additional performance is observed, the predicted GPA in subsequent semesters becomes increasingly accurate, providing the ability to advise students regarding likely success outcomes early in their academic careers.

Many universities have recently focused significant efforts on enhancing their graduation rates. Numerous factors may impact a student's ability to succeed and ultimately graduate, including pre-university preparation, as well as the student support services provided by a university. However, even the best efforts to improve in these areas may fail if other institutional factors overwhelm their ability to facilitate student progress. Specifically, in this paper we consider degree to which the underlying curriculum that a student must traverse in order to earn a degree impacts progress. Using complex network analysis and graph theory, this paper proposes a framework for analyzing university course networks at the university, college and departmental levels. The analyses we provide are based on quantifying the importance of a course based on its delay and blocking factors, as well as the number of curricula that incorporate the course, leading to a metric we refer to as the course cruciality. Experimental results, using data from the University of New Mexico, show that the distribution of course cruciality follows a power law distribution. Applications of the proposed framework are extended to study the complexity of curricula within colleges as well as the tendency of a university's disciplines to associate with others that are unlike them. This work may be useful to both students and decision makers at universities as it presents a robust framework for analyzing the ease of flow of students through curricula, which may lead to improvements that facilitate improved student success.

Colleges and universities are increasingly interested in tracking student progress as they monitor and work to improve their retention and graduation rates. Ideally, early indicators of student progress, or lack thereof, can be used to provide appropriate interventions that increase the likelihood of student success. In this paper we present a framework that uses data mining and machine learning techniques, and in particular, linear regression and a Markov network (MN), to predict the performance of students early in their academic careers. The results obtained show that the proposed framework can predict student progress, specifically student grade point average (GPA) within the intended major, with minimal error after observing a single semester of performance. Furthermore, as additional performance is observed, the predicted GPA in subsequent semesters becomes increasingly accurate, providing the ability to advise students regarding likely success outcomes early in their academic careers.

Herein, we consider how overlay architectures can be deployed in order to support usage management in cloud computing. We contrast filter-centric perimeter and policy-centric overlay approaches toward usage management of sensitive content from the perspective of cross-domain solutions and then demonstrate a path from current systems to distributed policy-centric overlay methods. We first cover the need for overlay networks supporting usage management by highlighting the pending migration to utility computing models and the current shortcomings of those kinds of systems from a variety of perspectives. We also briefly cover the state of usage management within these kinds of systems, also examining the known state of the art in proposed cross domain solutions. We then compare and contrast the advantages of policy-based overlay systems to solving usage management issues over sensitive materials within multi-level security environments. We then demonstrate a migration path from current to future solutions. We close with our current and future directions of research.

User concerns regarding data handling within the cloud will gain increasing importance as cloud computing becomes more pervasive. Existing service level agreement (SLA) frameworks are not designed for flexibly handling even relatively straightforward usage policies. This paper introduces the notion and importance of usage management in cloud computing. It provides an analysis of features and challenges involved in deploying a usage management framework over a distributed cloud environment to enable automated and actionable interpretation, reasoning and enforcement of usage policies. Finally, a preliminary architecture for such a framework is proposed.

In this paper we describe the development of a domain specific language (DSL) for expressing usage management policies and associating those policies with managed artifacts. We begin by framing a model for the language, including generalized use cases, a domain model, a general supported life-cycle, and specific extension requirements. We then develop the language from that model, demonstrating key syntactic elements and highlighting the technology behind the language while tracing features back to the initial model. We then demonstrate how the DSL supports common usage management and DRM-centric environments, including creative commons, the extensible rights markup language (XrML), and the open digital rights language (ODRL).

In this paper, we describe a formal framework for usage management that provides a scaffolding upon which interoperable usage management systems can be built. We apply the principles of system design to standardize certain features of the framework, such as the operational semantics, and leave free of standards areas that necessitate choice and innovation. We demonstrate that such an approach enables us to achieve a balance of flexibility and usability for the purpose of interoperability in usage management systems. We provide a formal model that allows us to define formal semantics for interoperability.

Information value has been implicitly utilized and mostly non-subjectively computed in information retrieval (IR) systems. We explicitly define and compute the value of an information piece as a function of two parameters, the first is the potential semantic impact the target information can subjectively have on its recipient's world-knowledge, and the second parameter is trust in the information source. We model these two parameters as properties of RDF graphs. Two graphs are constructed, a target graph representing the semantics of the target body of information and a context graph representing the context of the consumer of that information. We compute information value subjectively as a function of both potential change to the context graph (impact) and the overlap between the two graphs (trust). Graph change is computed as a graph edit distance measuring the dissimilarity between the context graph before and after the learning of the target graph. A particular application of this subjective information valuation is in the construction of a personalized ranking component in Web search engines. Based on our method, we construct a Web re-ranking system that personalizes the information experience for the information-consumer.

Information valuation has typically been carried out implicitly in question-answering and document retrieval systems. We argue that explicit information valuation is needed to move away from the system and process-centric nature of implicit valuation which has also hindered the theoretical study of information value under a unified and explicit framework. In this paper we present a graphical-based model for explicit information valuation. Our model caters to the subjective nature of information quality by measuring the impact a candidate piece of information may have on a knowledge base representing the recipient's world view. Our model is capable of evaluating information semantically at the statement level and is in effect basing information-valuation on information-understanding. However, information value can be computed and predicted using our causal graph model without requiring full logical inference typically needed for information-understanding.

Emergence of different digital rights management (DRM) systems and various rights expression languages (RELs) has led to problems with DRM interoperability and smooth flow of content across different content management systems. The complexity, varied scope, undefined boundaries, and lack of formalism in current RELs pose some of the biggest challenges in addressing DRM interoperability. In this paper, we define a formal language neutral conceptual model for rights expression statements that provides a platform upon which rights statements from different RELs can be mapped, reasoned, and manipulated.

In this paper we cast DRM in a setting that allows us to model a number of current approaches as games. The DRM game is partitioned into two subgames, one that considers the game associated with content acquisition, and a second that considers how a consumer uses the content, along with a vendor's response to this usage. Examples are provided in order to demonstrate how these subgames correspond to real situations associated with content industries, and the conditions under which Nash equilibria will exist. These subgames form the primary stage of a repeated game that models a number of important long-term interactions between consumers and vendors. We analyze current strategies that attempt to influence the outcome of the repeated game, and we also consider a new type of architectural infrastructure that makes novel use of a trust authority in order to create a suitable environment for constructing DRM games that may prove useful in the future.

We present a generalization of Benford’s law for the first significant digit. This generalization is based on keeping two terms of the Fourier expansion of the probability density function of the data in the modular logarithmic domain. We prove that images in the Discrete Cosine Transform domain closely follow this generalization. We use this property to propose an application in image steganalysis, namely, detecting that a given image carries a hidden message.

PageRank is an algorithm used by several search engines to rank web documents according to their assumed relevance and popularity deduced from the Web's link structure. PageRank determines a global ordering of candidate search results according to each page's popularity as determined by the number and importance of pages linking to these results. Personalized and topic-sensitive PageRank are variants of the algorithm that return a local ranking based on each user's preferences as biased by a set of pages they trust or topics they prefer. In this paper we compare personalized and topic-sensitive local PageRanks to the global PageRank showing experimentally how similar or dissimilar results of personalization can be to the original global rank results and to other personalizations. Our approach is to examine a snapshot of the Web and determine how advantageous personalization can be in the best and worst cases and how it performs at various values of the damping factor in the PageRank formula.

In this paper we consider the functionality that a rights expression language (REL) should provide within a digital rights management (DRM) environment. We begin by noting the dearth of applications that make use of RELs, despite the fact that they have now been available since the late 1990's. We posit that one of the main impediments to the use of RELs is the complexity associated with understanding and using them. This results from the fact that the functionality needed to handle a wide variety of possible DRM scenarios is typically built into a REL, and it is often difficult to cleanly partition out only those pieces needed by a particular DRM application. Basing DRM system design on a layered architecture provides one way of achieving a partitioning and points to the need for a simple REL that is exclusively responsible for the expression of rights, while pushing much of the functionality found in current RELs into higher system layers. In order to demonstrate the usefulness of this approach, we provide an example implementation dealing with DRM-based negotiation.

Interoperability is currently seen as one of the most significant problems facing the digital rights management (DRM) industry. In this paper we consider the problem of interoperability among DRM systems from the perspective of a layered architectural framework. The advantage of looking at the problem from this point of view is that the layered framework provides a certain amount of structure that is very helpful in guiding those working on DRM interoperability issues. Specifically, the layered framework we describe provides a useful design abstraction along architectural lines. One of the advantages of this perspective is that it allows us to consider the level within computing/communication architectures at which certain functionality should be provided, and then to address how the functionality between layers should interact in order to provide specific DRM capabilities. The communications that occur between layers, both within a single system and between two communicating systems, are the places where protocols can be defined and possibly standardized. Thus, they provide focal points for studying and addressing interoperability in DRM systems.

The current landscape for digital rights management(DRM) consists of various ad hoc technologies and platforms that largely focus on copy protection. The fragmented nature of the DRM industry in 2004 is somewhat reminiscent of the telecommunications industry in the late 1980's. At that time various networking technologies were available, and what was needed was a technology that could integrate existing networks and provide various services to users. The OSI layered framework and the TCP/IP communications protocol suite provided a solution to this situation. The OSI model divides the process of digital data communications into layers. Likewise, in this paper we divide the process of DRM into layers in which various services are offered to the users of digital content at each layer. Three blocks of layers have been identified. The upper layers deal with the end-to-end functions of the application, the middle layers deal with rights expression and interpretation, and the lower layers ensure rights enforcement. This paper describes how responsibilities might be distributed among the various layers, and considers where in these layers it would be appropriate to define protocols and standards.

Remotely sensed satellite images are an important source of geographical data commonly used as input for various types of classification algorithms. For example, these algorithms are commonly used to classify earth land cover, analyze crop conditions, assess mineral and petroleum deposits, and quantify urban growth. Many vendors of digital images are using or are considering the use of invisible watermarking as a means of protecting their images from theft or unauthorized usage. Indeed, the use of invisible watermarking is routinely considered for use in emerging digital rights management~(DRM) systems that may be deployed to manage and protect the rights associated with satellite imagery, or types of "scientific" imagery~(e.g., in the medical field) that routinely have mathematical analyses applied to them. The concern then is how this watermarking impacts subsequent analyses. Specifically, the invisible watermarking process involves making imperceptible modifications to the pixel values of an image. However, even though these changes may be imperceptible to the human observer, they must be of sufficient magnitude to allow for watermark detection. Because of this, the use of invisible watermarking can also impact the performance of image classification algorithms. This paper is concerned with quantifying the impact that invisible watermarks have on satellite image classification. In particular, Landsat satellite images were watermarked using a number of well-known techniques, and the misclassification that resulted from this watermarking was measured. Experimental results show that even weak watermarking can lead to significant misclassification when common image classification algorithms are applied. Thus, the use of watermarking within DRM systems needs to be carefully considered, with particular attention given to the type of content that the watermarking will be applied to.

The Greek Orthodox Archdiocese of America~(GOA) has amassed a rich and varied collection of artifacts associated with two thousand years of religious and historical tradition, as well as more than a century of chronicles in America. The items in this archive include iconography, art, photographs, letters, and other memorabilia. The GOA has endeavored to digitize these assets in order to preserve them, while at the same time make them more accessible for appropriate and beneficial uses. Specifically, the Department of Internet Ministries at the GOA was tasked with overseeing this digitization effort, as well as with the creation of appropriate tools and technology for accessing the resulting digital archive. The challenges associated with this work included the wide diversity of media types, the need to attach specific restrictions to the use of different items, and the desire to provide a user experience that was transparent and not daunting or discouraging. At a fairly early stage it was determined that emerging digital rights management (DRM) capabilities would be necessary in order to realize the goals of the project. These capabilities, however, had to be implemented while accounting for a previously deployed digital asset management (DAM) and web publishing system, not to mention the existing IT infrastructure. This paper presents a case study that describes the efforts associated with the specification, creation, and deployment of an effective DRM system that incorporates rights enforcement technology, and works in synergy with the previously deployed DAM system at the GOA.

The focus of this paper is semi-supervised learning in the context of pattern recognition. Semi-supervised learning (SSL) refers to the semi-supervised construction of clusters during the training phase of exemplar-based classifiers. Using artificially generated data sets we present experimental results of classifiers that follow the SSL paradigm and we show that, especially for difficult pattern recognition problems featuring high class overlap, for exemplar-based classifiers implementing SSL i) the generalization performance improves, while ii) the number of necessary exemplars decreases significantly, when compared to the original versions of the classifiers.

A measure of success for any learning algorithm is how useful it is in a variety of learning situations. Those learning algorithms that support universal function approximation can theoretically be applied to a very large and interesting class of learning problems. Many kinds of neural network architectures have already been shown to support universal approximation. In this paper, we will provide a proof to show that Fuzzy ART augmented with a single layer of perceptrons is a universal approximator. Moreover, the Fuzzy ARTMAP neural network architecture, by itself, will be shown to be a universal approximator.

We present a modification to the Fuzzy ARTMAP neural network architecture for conducting classification in a probabilistic setting. We call this new architecture Hierarchical ARTMAP (HARTMAP). Performance comparisons with Fuzzy ARTMAP, Gaussian ARTMAP and Boosted ARTMAP on some simple two-class problems are discussed. Experimental results indicate that HARTMAP yields better generalization results on problems involving overlap of the underlying pattern distributions.

In this paper we introduce a procedure, based on the max-min clustering method, that identifies a fixed order of training pattern presentation for fuzzy adaptive resonance theory mapping (ARTMAP). This procedure is referred to as the ordering algorithm, and the combination of this procedure with fuzzy ARTMAP is referred to as ordered fuzzy ARTMAP. Experimental results demonstrate that ordered fuzzy ARTMAP exhibits a generalization performance that is better than the average generalization performance of fuzzy ARTMAP, and in certain cases as good as, or better than the best fuzzy ARTMAP generalization performance. We also calculate the number of operations required by the ordering algorithm and compare it to the number of operations required by the training phase of fuzzy ARTMAP. We show that, under mild assumptions, the number of operations required by the ordering algorithm is a fraction of the number of operations required by fuzzy ARTMAP.

The authors have developed an interactive environment for the creation and maintenance of dynamic, active multimedia-based teaching mechanisms. The environment is designed to be user-friendly and to facilitate the creation of educational material. This tool has already been used to create courses in multidimensional signal processing (MDSP) by researchers working together while geographically separated.

A set of nonlinear differential equations that describe the dynamics of the ART1 model are presented, along with the motivation for their use. These equations are extensions of those developed by Carpenter and Grossberg (1987). It is shown how these differential equations allow the ART1 model to be realized as a collective nonlinear dynamical system. Specifically, we present an ART1-based neural network model whose description requires no external control features. That is, the dynamics of the model are completely determined by the set of coupled differential equations that comprise the model. It is shown analytically how the parameters of this model can be selected so as to guarantee a behavior equivalent to that of ART1 in both fast and slow learning scenarios. Simulations are performed in which the trajectories of node and weight activities are determined using numerical approximation techniques.

The analysis of complex neural network models via analytical techniques is often quite difficult due to the large numbers of components involved and the nonlinearities associated with these components. The authors present a framework for simulating neural networks as discrete event nonlinear dynamical systems. This includes neural network models whose components are described by continuous-time differential equations or by discrete-time difference equations. Specifically, the authors consider the design and construction of a concurrent object-oriented discrete event simulation environment for neural networks. The use of an object-oriented language provides the data abstraction facilities necessary to support modification and extension of the simulation system at a high level of abstraction. Furthermore, the ability to specify concurrent processing supports execution on parallel architectures. The use of this system is demonstrated by simulating a specific neural network model on a general-purpose parallel computer.>

We consider the ART1 neural network architecture. It is shown that in the fast learning case, an ART1 network that is repeatedly presented with an arbitrary list of binary input patterns, self-stabilizes the recognition code of every size-l pattern in at most l list presentations.

A parallel hardware implementation of the associative memory neural network introduced by Hopfield is described. The design utilizes the Geometric Arithmetic Parallel Processor (GAPP), a commercially available single-chip VLSI general-purpose array processor consisting of 72 processing elements. The ability to cascade these chips allows large arrays of processors to be easily constructed and used to implement the Hopfield network. The memory requirements and processing times of such arrays are analyzed based on the number of nodes in the network and the number of exemplar patterns. Compared with other digital implementations, this design yields significant improvements in runtime performance and offers the capability of using large neural network associative memories in real-time applications.