Simultaneous Learning of Nonlinear Manifold and Dynamical Models for High-dimensional Time Series

The goal of this work is to learn a parsimonious and informative representation for high-dimensional time series. Conceptually, this comprises two distinct yet tightly coupled tasks: learning a low-dimensional manifold and modeling the dynamical process. These two tasks have a complementary relationship as the temporal constraints provide valuable neighborhood information for dimensionality reduction and conversely, the low-dimensional space allows dynamics to be learnt efficiently. Solving these two tasks simultaneously allows important information to be exchanged mutually. If nonlinear models are required to capture the rich complexity of time series, then the learning problem becomes harder as the nonlinearities in both tasks are coupled. The proposed solution approximates the nonlinear manifold and dynamics using piecewise linear models. The interactions among the linear models are captured in a graphical model. By exploiting the model structure, efficient inference and learning algorithms are obtained without oversimplifying the model of the underlying dynamical process. Evaluation of the proposed framework with competing approaches is conducted in three sets of experiments: dimensionality reduction and reconstruction using synthetic time series, video synthesis using a dynamic texture database, and human motion synthesis, classification and tracking on a benchmark data set. In all experiments, the proposed approach provides superior performance.

Simultaneous Learning of Non-Linear Manifold and Dynamical Models for High-Dimensional Time Series

The goal of this work is to learn a parsimonious and informative representation for high-dimensional time series. Conceptually, this comprises two distinct yet tightly coupled tasks: learning a low-dimensional manifold and modeling the dynamical process. These two tasks have a complementary relationship as the temporal constraints provide valuable neighborhood information for dimensionality reduction and conversely, the low-dimensional space allows dynamics to be learnt efficiently. Solving these two tasks simultaneously allows important information to be exchanged mutually. If nonlinear models are required to capture the rich complexity of time series, then the learning problem becomes harder as the nonlinearities in both tasks are coupled. The proposed solution approximates the nonlinear manifold and dynamics using piecewise linear models. The interactions among the linear models are captured in a graphical model. The model structure setup and parameter learning are done using a variational Bayesian approach, which enables automatic Bayesian model structure selection, hence solving the problem of over-fitting. By exploiting the model structure, efficient inference and learning algorithms are obtained without oversimplifying the model of the underlying dynamical process. Evaluation of the proposed framework with competing approaches is conducted in three sets of experiments: dimensionality reduction and reconstruction using synthetic time series, video synthesis using a dynamic texture database, and human motion synthesis, classification and tracking on a benchmark data set. In all experiments, the proposed approach provides superior performance.

Learning Embeddings for Indexing, Retrieval, and Classification, with Applications to Object and Shape Recognition in Image Databases

Nearest neighbor retrieval is the task of identifying, given a database of objects and a query object, the objects in the database that are the most similar to the query. Retrieving nearest neighbors is a necessary component of many practical applications, in fields as diverse as computer vision, pattern recognition, multimedia databases, bioinformatics, and computer networks. At the same time, finding nearest neighbors accurately and efficiently can be challenging, especially when the database contains a large number of objects, and when the underlying distance measure is computationally expensive. This thesis proposes new methods for improving the efficiency and accuracy of nearest neighbor retrieval and classification in spaces with computationally expensive distance measures. The proposed methods are domain-independent, and can be applied in arbitrary spaces, including non-Euclidean and non-metric spaces. In this thesis particular emphasis is given to computer vision applications related to object and shape recognition, where expensive non-Euclidean distance measures are often needed to achieve high accuracy. The first contribution of this thesis is the BoostMap algorithm for embedding arbitrary spaces into a vector space with a computationally efficient distance measure. Using this approach, an approximate set of nearest neighbors can be retrieved efficiently - often orders of magnitude faster than retrieval using the exact distance measure in the original space. The BoostMap algorithm has two key distinguishing features with respect to existing embedding methods. First, embedding construction explicitly maximizes the amount of nearest neighbor information preserved by the embedding. Second, embedding construction is treated as a machine learning problem, in contrast to existing methods that are based on geometric considerations. The second contribution is a method for constructing query-sensitive distance measures for the purposes of nearest neighbor retrieval and classification. In high-dimensional spaces, query-sensitive distance measures allow for automatic selection of the dimensions that are the most informative for each specific query object. It is shown theoretically and experimentally that query-sensitivity increases the modeling power of embeddings, allowing embeddings to capture a larger amount of the nearest neighbor structure of the original space. The third contribution is a method for speeding up nearest neighbor classification by combining multiple embedding-based nearest neighbor classifiers in a cascade. In a cascade, computationally efficient classifiers are used to quickly classify easy cases, and classifiers that are more computationally expensive and also more accurate are only applied to objects that are harder to classify. An interesting property of the proposed cascade method is that, under certain conditions, classification time actually decreases as the size of the database increases, a behavior that is in stark contrast to the behavior of typical nearest neighbor classification systems. The proposed methods are evaluated experimentally in several different applications: hand shape recognition, off-line character recognition, online character recognition, and efficient retrieval of time series. In all datasets, the proposed methods lead to significant improvements in accuracy and efficiency compared to existing state-of-the-art methods. In some datasets, the general-purpose methods introduced in this thesis even outperform domain-specific methods that have been custom-designed for such datasets.

Cortical Learning of Recognition Categories: A Resolution of the Exemplar Vs. Prototype Debate

Do humans and animals learn exemplars or prototypes when they categorize objects and events in the world? How are different degrees of abstraction realized through learning by neurons in inferotemporal and prefrontal cortex? How do top-down expectations influence the course of learning? Thirty related human cognitive experiments (the 5-4 category structure) have been used to test competing views in the prototype-exemplar debate. In these experiments, during the test phase, subjects unlearn in a characteristic way items that they had learned to categorize perfectly in the training phase. Many cognitive models do not describe how an individual learns or forgets such categories through time. Adaptive Resonance Theory (ART) neural models provide such a description, and also clarify both psychological and neurobiological data. Matching of bottom-up signals with learned top-down expectations plays a key role in ART model learning. Here, an ART model is used to learn incrementally in response to 5-4 category structure stimuli. Simulation results agree with experimental data, achieving perfect categorization in training and a good match to the pattern of errors exhibited by human subjects in the testing phase. These results show how the model learns both prototypes and certain exemplars in the training phase. ART prototypes are, however, unlike the ones posited in the traditional prototype-exemplar debate. Rather, they are critical patterns of features to which a subject learns to pay attention based on past predictive success and the order in which exemplars are experienced. Perturbations of old memories by newly arriving test items generate a performance curve that closely matches the performance pattern of human subjects. The model also clarifies exemplar-based accounts of data concerning amnesia.

Grid Cell Hexagonal Patterns Formed by Fast Self-Organized Learning within Entorhinal Cortex

Grid cells in the dorsal segment of the medial entorhinal cortex (dMEC) show remarkable hexagonal activity patterns, at multiple spatial scales, during spatial navigation. How these hexagonal patterns arise has excited intense interest. It has previously been shown how a selforganizing map can convert firing patterns across entorhinal grid cells into hippocampal place cells that are capable of representing much larger spatial scales. Can grid cell firing fields also arise during navigation through learning within a self-organizing map? A neural model is proposed that converts path integration signals into hexagonal grid cell patterns of multiple scales. This GRID model creates only grid cell patterns with the observed hexagonal structure, predicts how these hexagonal patterns can be learned from experience, and can process biologically plausible neural input and output signals during navigation. These results support a unified computational framework for explaining how entorhinal-hippocampal interactions support spatial navigation.

A Neural Network Architecture for Autonomous Learning, Recognition, and Prediction in a Nonstationary World

In a constantly changing world, humans are adapted to alternate routinely between attending to familiar objects and testing hypotheses about novel ones. We can rapidly learn to recognize and narne novel objects without unselectively disrupting our memories of familiar ones. We can notice fine details that differentiate nearly identical objects and generalize across broad classes of dissimilar objects. This chapter describes a class of self-organizing neural network architectures--called ARTMAP-- that are capable of fast, yet stable, on-line recognition learning, hypothesis testing, and naming in response to an arbitrary stream of input patterns (Carpenter, Grossberg, Markuzon, Reynolds, and Rosen, 1992; Carpenter, Grossberg, and Reynolds, 1991). The intrinsic stability of ARTMAP allows the system to learn incrementally for an unlimited period of time. System stability properties can be traced to the structure of its learned memories, which encode clusters of attended features into its recognition categories, rather than slow averages of category inputs. The level of detail in the learned attentional focus is determined moment-by-moment, depending on predictive success: an error due to over-generalization automatically focuses attention on additional input details enough of which are learned in a new recognition category so that the predictive error will not be repeated. An ARTMAP system creates an evolving map between a variable number of learned categories that compress one feature space (e.g., visual features) to learned categories of another feature space (e.g., auditory features). Input vectors can be either binary or analog. Computational properties of the networks enable them to perform significantly better in benchmark studies than alternative machine learning, genetic algorithm, or neural network models. Some of the critical problems that challenge and constrain any such autonomous learning system will next be illustrated. Design principles that work together to solve these problems are then outlined. These principles are realized in the ARTMAP architecture, which is specified as an algorithm. Finally, ARTMAP dynamics are illustrated by means of a series of benchmark simulations.

Adaptive Resonance Associative Map: A Hierarchical ART System for Fast Stable Associative Learning

This paper introduces a new class of predictive ART architectures, called Adaptive Resonance Associative Map (ARAM) which performs rapid, yet stable heteroassociative learning in real time environment. ARAM can be visualized as two ART modules sharing a single recognition code layer. The unit for recruiting a recognition code is a pattern pair. Code stabilization is ensured by restricting coding to states where resonances are reached in both modules. Simulation results have shown that ARAM is capable of self-stabilizing association of arbitrary pattern pairs of arbitrary complexity appearing in arbitrary sequence by fast learning in real time environment. Due to the symmetrical network structure, associative recall can be performed in both directions.

Normal and Amnesis Learning, Recognition, and Memory by a Neural Model of Cortico-Hippocampal Interactions

The processes by which humans and other primates learn to recognize objects have been the subject of many models. Processes such as learning, categorization, attention, memory search, expectation, and novelty detection work together at different stages to realize object recognition. In this article, Gail Carpenter and Stephen Grossberg describe one such model class (Adaptive Resonance Theory, ART) and discuss how its structure and function might relate to known neurological learning and memory processes, such as how inferotemporal cortex can recognize both specialized and abstract information, and how medial temporal amnesia may be caused by lesions in the hippocampal formation. The model also suggests how hippocampal and inferotemporal processing may be linked during recognition learning.

A critical realist examination of the inhibitors of double loop learning in an agile adoption

The original solution to the high failure rate of software development projects was the imposition of an engineering approach to software development, with processes aimed at providing a repeatable structure to maintain a consistency in the ‘production process’. Despite these attempts at addressing the crisis in software development, others have argued that the rigid processes of an engineering approach did not provide the solution. The Agile approach to software development strives to change how software is developed. It does this primarily by relying on empowered teams of developers who are trusted to manage the necessary tasks, and who accept that change is a necessary part of a development project. The use of, and interest in, Agile methods in software development projects has expanded greatly, yet this has been predominantly practitioner driven. There is a paucity of scientific research on Agile methods and how they are adopted and managed. This study aims at addressing this paucity by examining the adoption of Agile through a theoretical lens. The lens used in this research is that of double loop learning theory. The behaviours required in an Agile team are the same behaviours required in double loop learning; therefore, a transition to double loop learning is required for a successful Agile adoption. The theory of triple loop learning highlights that power factors (or power mechanisms in this research) can inhibit the attainment of double loop learning. This study identifies the negative behaviours - potential power mechanisms - that can inhibit the double loop learning inherent in an Agile adoption, to determine how the Agile processes and behaviours can create these power mechanisms, and how these power mechanisms impact on double loop learning and the Agile adoption. This is a critical realist study, which acknowledges that the real world is a complex one, hierarchically structured into layers. An a priori framework is created to represent these layers, which are categorised as: the Agile context, the power mechanisms, and double loop learning. The aim of the framework is to explain how the Agile processes and behaviours, through the teams of developers and project managers, can ultimately impact on the double loop learning behaviours required in an Agile adoption. Four case studies provide further refinement to the framework, with changes required due to observations which were often different to what existing literature would have predicted. The study concludes by explaining how the teams of developers, the individual developers, and the project managers, working with the Agile processes and required behaviours, can inhibit the double loop learning required in an Agile adoption. A solution is then proposed to mitigate these negative impacts. Additionally, two new research processes are introduced to add to the Information Systems research toolkit.

Nested dictionary learning for hierarchical organization of imagery and text

A tree-based dictionary learning model is developed for joint analysis of imagery and associated text. The dictionary learning may be applied directly to the imagery from patches, or to general feature vectors extracted from patches or superpixels (using any existing method for image feature extraction). Each image is associated with a path through the tree (from root to a leaf), and each of the multiple patches in a given image is associated with one node in that path. Nodes near the tree root are shared between multiple paths, representing image characteristics that are common among different types of images. Moving toward the leaves, nodes become specialized, representing details in image classes. If available, words (text) are also jointly modeled, with a path-dependent probability over words. The tree structure is inferred via a nested Dirichlet process, and a retrospective stick-breaking sampler is used to infer the tree depth and width.

Use of Web Logs (Blogs) in Learning Environments: Blogs for Student Support

Innovation in technology and communications and particularly the advent of the Web is changing the structure of teaching and learning today. While there is much debate about the use of technology in learning and how e-learning is creating new approaches to delivery of learning there is been very little if any work on the use of of the emerging technologies in providing student support through their learning process. This paper reports on research and development undertaken by the eCentre based at the University of Greenwich School Of Computing in designing and developing a "Project Blog System" in order to address some long standing issues related to supervision of final year degree student projects. The paper will report on the methodology used to design the system and will discuss some of the results from students and staff evaluation of the system developed.

"Practise what we preach": a social constructivist approach to blended learning

This presentation reports on the formal evaluation, through questionnaires, of a new Level 1 undergraduate course, for 130 student teachers, that uses blended learning. The course design seeks to radicalise the department’s approach to teaching, learning and assessment and use students as change agents. Its structure and content, model social constructivist approaches to learning. Building on the student’s experiences of and, reflections on, previous learning, promotes further learning through the support of “able others” (Vygotsky 1978), facilitating and nurturing a secure community of practice for students new to higher education. The course’s design incorporates individual, paired, small and large group activities and exploits online video, audio and text materials. Course units begin and end with face-to-face tutor-led activities. Online elements, including discussions and formative submissions, are tutor-mediated. Students work together face-to-face and online to read articles, write reflections, develop presentations, research and share experiences and resources. Summative joint assignments and peer assessments emphasise the value of collaboration and teamwork for academic, personal and professional development. Initial informal findings are positive, indicating that students have engaged readily with course content and structure, with few reporting difficulties accessing or using technology. Students have welcomed the opportunity to work together to tackle readings in a new genre, pilot presentation skills and receive and give constructive feedback to peers. Course tutors have indicated that depth and quality of study are evident, with regular online formative submissions enabling tutors to identify and engage directly with student’s needs, provide feedback and develop appropriately designed distance and face-to-face teaching materials. Pastoral tutors have indicated that students have reported non-engagement of peers, leading to the rapid application of academic or personal support. Outcomes of the formal evaluation will inform the development of Level 2 and 3 courses and influence the department’s use of blended learning.

RuhrCampusOnline: Hochschulübergreifendes E-Learning in der Universitätsallianz Metropole Ruhr

Das Projekt RuhrCampusOnline zielt darauf ab, hochschulübergreifende Lehre für die Partnerhochschulen der Universitätsallianz Metropole Ruhr (UAMR) auf der Basis von Blended-Learning-Arrangements zu implementieren. Der Beitrag stellt die strategische Bedeutung dieses Vorhabens für die Universitäten im Ruhrgebiet dar und beschreibt die Organisation und Vorgehensweise im Projekt. Ziel ist es, einen Pool von Kursen zu implementieren, die einen hohen Online-Anteil haben, und von Studierenden hochschulübergreifend genutzt werden. Über die Internet-Plattform RCO werden diese Kurse sichtbar gemacht, das Belegen der Veranstaltung realisiert und der Austausch von Leistungspunkten unterstützt. Der Beitrag stellt erste Erfahrungen nach einem Jahr Projektlaufzeit vor. (DIPF/ Orig.)

A Competence-Based Course Authoring Concept for Learning Platforms with Legacy Assignment Tools

This paper is concerned with several of the most important aspects of Competence-Based Learning (CBL): course authoring, assignments, and categorization of learning content. The latter is part of the so-called Bologna Process (BP) and can effectively be supported by integrating knowledge resources like, e.g., standardized skill and competence taxonomies into the target implementation approach, aiming at making effective use of an open integration architecture while fostering the interoperability of hybrid knowledge-based e-learning solutions. Modern scenarios ask for interoperable software solutions to seamlessly integrate existing e-learning infrastructures and legacy tools with innovative technologies while being cognitively efficient to handle. In this way, prospective users are enabled to use them without learning overheads. At the same time, methods of Learning Design (LD) in combination with CBL are getting more and more important for production and maintenance of easy to facilitate solutions. We present our approach of developing a competence-based course-authoring and assignment support software. It is bridging the gaps between contemporary Learning Management Systems (LMS) and established legacy learning infrastructures by embedding existing resources via Learning Tools Interoperability (LTI). Furthermore, the underlying conceptual architecture for this integration approach will be explained. In addition, a competence management structure based on knowledge technologies supporting standardized skill and competence taxonomies will be introduced. The overall goal is to develop a software solution which will not only flawlessly merge into a legacy platform and several other learning environments, but also remain intuitively usable. As a proof of concept, the so-called platform independent conceptual architecture model will be validated by a concrete use case scenario.