Papers by David Windridge
IEEE Transactions on Quantum Engineering
Building a quantum analog of classical deep neural networks represents a fundamental challenge in... more Building a quantum analog of classical deep neural networks represents a fundamental challenge in quantum computing. A key issue is how to address the inherent non-linearity of classical deep learning, a problem in the quantum domain due to the fact that the composition of an arbitrary number of quantum gates, consisting of a series of sequential unitary transformations, is intrinsically linear. This problem has been variously approached in the literature, principally via the introduction of measurements between layers of unitary transformations. In this paper, we introduce the Quantum Path Kernel, a formulation of quantum machine learning capable of replicating those aspects of deep machine learning typically associated with superior generalization performance in the classical domain, specifically, hierarchical feature learning. Our approach generalizes the notion of Quantum Neural Tangent Kernel, which has been used to study the dynamics of classical and quantum machine learning models. The Quantum Path Kernel exploits the parameter trajectory, i.e. the curve delineated by model parameters as they evolve during training, enabling the representation of differential layer-wise convergence behaviors, or the formation of hierarchical parametric dependencies, in terms of their manifestation in the gradient space of the predictor function. We evaluate our approach with respect to variants of the classification of Gaussian XOR mixtures -an artificial but emblematic problem that intrinsically requires multilevel learning in order to achieve optimal class separation.
IEEE MultiMedia, Apr 1, 2015
Full bibliographic details must be given when referring to, or quoting from full items including ... more Full bibliographic details must be given when referring to, or quoting from full items including the author's name, the title of the work, publication details where relevant (place, publisher, date), pagination, and for theses or dissertations the awarding institution, the degree type awarded, and the date of the award. If you believe that any material held in the repository infringes copyright law, please contact the
Perception-Action Learning as an Epistemologically-Consistent Model for Self-Updating Cognitive Representation
Advances in Experimental Medicine and Biology, Nov 21, 2009
As well as having the ability to formulate models of the world capable of experimental falsificat... more As well as having the ability to formulate models of the world capable of experimental falsification, it is evident that human cognitive capability embraces some degree of representational plasticity, having the scope (at least in infancy) to modify the primitives in terms of which the world is delineated. We hence employ the term 'cognitive bootstrapping' to refer to the autonomous updating of an embodied agent's perceptual framework in response to the perceived requirements of the environment in such a way as to retain the ability to refine the environment model in a consistent fashion across perceptual changes.We will thus argue that the concept of cognitive bootstrapping is epistemically ill-founded unless there exists an a priori percept/motor interrelation capable of maintaining an empirical distinction between the various possibilities of perceptual categorization and the inherent uncertainties of environment modeling.As an instantiation of this idea, we shall specify a very general, logically-inductive model of perception-action learning capable of compact re-parameterization of the percept space. In consequence of the a priori percept/action coupling, the novel perceptual state transitions so generated always exist in bijective correlation with a set of novel action states, giving rise to the required empirical validation criterion for perceptual inferences. Environmental description is correspondingly accomplished in terms of progressively higher-level affordance conjectures which are likewise validated by exploratory action.Application of this mechanism within simulated perception-action environments indicates that, as well as significantly reducing the size and specificity of the a priori perceptual parameter-space, the method can significantly reduce the number of iterations required for accurate convergence of the world-model. It does so by virtue of the active learning characteristics implicit in the notion of cognitive bootstrapping.
The Practical Performance Characteristics of Tomographically Filtered Multiple Classifier Fusion
Springer eBooks, 2003
ABSTRACT In this paper we set out to give an indication both of the classification performance an... more ABSTRACT In this paper we set out to give an indication both of the classification performance and the robustness to estimation error of the authors’ ‘tomographic’ classifier fusion methodology in a comparative field test with the sum and product classier fusion methodologies. In encompassing this, we find evidence to confirm that the tomographic methodology represents a generally superior fusion strategy across the entire range of problem dimensionalities, final results indicating an as much as 25% improvement on the next nearest performing combination scheme at the extremity of the tested range.
An Entropy-Based Approach to the Hierarchical Acquisition of Perception-Action Capabilities
Lecture Notes in Computer Science, 2008
We detail an approach to the autonomous acquisition of hierarchical perception-action competences... more We detail an approach to the autonomous acquisition of hierarchical perception-action competences in which capabilities are bootstrapped using an information-based saliency measure. Our principle aim is hence to accelerate learning in embodied autonomous agents by aggregating novel motor capabilities and their corresponding perceptual representations using a subsumption-based strategy. The method seeks to allocate affordance parameterizations according to the current (possibly
arXiv (Cornell University), Aug 29, 2019
Contrast Enhanced Magnetic Resonance imaging (DCE-MRI) requires careful segmentation of the renal... more Contrast Enhanced Magnetic Resonance imaging (DCE-MRI) requires careful segmentation of the renal region of interest (ROI). Traditionally, human experts are required to manually delineate the kidney ROI across multiple images in the dynamic sequence. This approach is costly, time-consuming and labour intensive, and therefore acts to limit patient throughout and acts as one of the factors limiting the wider adoption of DCR-MRI in clinical practice. Therefore, to address this issue, we present the first use of Dynamic Mode Decomposition (DMD) as a the basis for automatic segmentation of a dynamic sequence, in this case, kidney ROIs in DCE-MRI. Using DMD coupled combined with thresholding and connected component analysis is first validated on synthetically generated data with known ground-truth, and then applied to ten healthy volunteers DCE-MRI datasets.
Economic Tomographic Classifier Fusion: Eliminating Redundant Hogborn Deconvolution Cycles in the Sum-Rule Domain
We consider a multiple classifier system which combines the hard decisions of experts by voting. ... more We consider a multiple classifier system which combines the hard decisions of experts by voting. We argue that the individual experts should not set their own decision thresholds. The respective thresholds should be selected jointly as this will allow for compensation of the weaknesses of some experts by the relative strengths of the others. We perform the joint optimization of decision thresholds for a multiple expert system by a systematic sampling of the multidimensional decision threshold space. We show the effectiveness of this approach on the important practical application of video shot cut detection.
Hidden Markov chain estimation and parameterisation via ICA-based
IEEE TRANS. ON INFORMATION FORENSICS AND SECURITY 1 Detection of Face Spoofing Using Visual Dynamics
Abstract—Rendering a face recognition system robust is vital in order to safeguard it against spo... more Abstract—Rendering a face recognition system robust is vital in order to safeguard it against spoof attacks carried out by using printed pictures of a victim (also known as print attack) or a replayed video of the person (replay attack). A key property in distinguishing a live, valid access from printed media or replayed videos is by exploiting the information dynamics of the video content, such as blinking eyes, moving lips, and facial dynamics. We advance the state of the art in facial anti-spoofing by applying a recently developed algorithm called Dynamic Mode Decomposition (DMD) as a general-purpose, entirely data-driven approach to capture the above liveness cues. We propose a classification pipeline consisting of DMD, Local Binary Patterns (LBP), and Support Vector Machines (SVM) with a histogram intersection kernel. A unique property of DMD is its ability to conveniently represent the temporal information of the entire video as a single image with the same dimensions as those...
We describe our approach to addressing Mini Challenge 2 of the 2016 IEEE VAST Challenge. We descr... more We describe our approach to addressing Mini Challenge 2 of the 2016 IEEE VAST Challenge. We describe four tools: POLAR Kermode, Classifier, Excel with conditional formatting on a power wall and Data Timelines.
Sparse Multimodal Classification of EEG Signals from Rapid Serial Visual Presentation of Diagnostic Images
Neural Computing and Applications, 2020
A Linguistic Feature Vector for the Visual
This paper presents a novel approach to sign language recognition that provides extremely high cl... more This paper presents a novel approach to sign language recognition that provides extremely high classification rates on minimal training data. Key to this approach is a 2 stage classification procedure where an initial classification stage extracts a high level description of hand shape and motion. This high level description is based upon sign linguistics and describes actions at a conceptual level easily understood by humans. Moreover, such a description broadly generalises temporal activities naturally overcoming variability of people and environments. A second stage of classification is then used to model the temporal transitions of individual signs using a classifier bank of Markov chains combined with Independent Component Analysis. We demonstrate classification rates as high as 97.67% for a lexicon of 43 words using only single instance training outperforming previous approaches where thousands of training examples are required.
Perception-Action Learning as an Epistemologically-Consistent Model for Self-Updating Cognitive Representation
Advances in Experimental Medicine and Biology, 2009
As well as having the ability to formulate models of the world capable of experimental falsificat... more As well as having the ability to formulate models of the world capable of experimental falsification, it is evident that human cognitive capability embraces some degree of representational plasticity, having the scope (at least in infancy) to modify the primitives in terms of which the world is delineated. We hence employ the term 'cognitive bootstrapping' to refer to the autonomous updating of an embodied agent's perceptual framework in response to the perceived requirements of the environment in such a way as to retain the ability to refine the environment model in a consistent fashion across perceptual changes.We will thus argue that the concept of cognitive bootstrapping is epistemically ill-founded unless there exists an a priori percept/motor interrelation capable of maintaining an empirical distinction between the various possibilities of perceptual categorization and the inherent uncertainties of environment modeling.As an instantiation of this idea, we shall specify a very general, logically-inductive model of perception-action learning capable of compact re-parameterization of the percept space. In consequence of the a priori percept/action coupling, the novel perceptual state transitions so generated always exist in bijective correlation with a set of novel action states, giving rise to the required empirical validation criterion for perceptual inferences. Environmental description is correspondingly accomplished in terms of progressively higher-level affordance conjectures which are likewise validated by exploratory action.Application of this mechanism within simulated perception-action environments indicates that, as well as significantly reducing the size and specificity of the a priori perceptual parameter-space, the method can significantly reduce the number of iterations required for accurate convergence of the world-model. It does so by virtue of the active learning characteristics implicit in the notion of cognitive bootstrapping.
Induced Decision Fusion in Automated Sign Language Interpretation: Using ICA to Isolate the Underlying Components of Sign
Lecture Notes in Computer Science, 2004
ABSTRACT We utilise the techniques of independent component analysis and principle component anal... more ABSTRACT We utilise the techniques of independent component analysis and principle component analysis to derive an independent set of gestural primitives for visual sign-language, employing existing sign linguistics as a reference point in the feature reduction. In this way it is possible both to reduce (by several orders of magnitude) the requisite quantity of HMM computation involved in word classification, as well as to significantly improve performance through having transformed the initial classification problem into one of decision fusion. Moreover, the independent and optimally-compact representation of the gestural primitives ensures a maximum of classifier diversity prior to combination.
Lecture Notes in Computer Science, 2005
In conventional computer vision systems symbol grounding is invariably established via supervised... more In conventional computer vision systems symbol grounding is invariably established via supervised learning. We investigate unsupervised symbol grounding mechanisms that rely on perception action coupling 1 . The mechanisms involve unsupervised clustering of observed actions and percepts. Their association gives rise to behaviours that emulate human action. The capability of the system is demonstrated on the problem of mimicking shape puzzle solving. It is argued that the same mechanisms support unsupervised cognitive bootstrapping in cognitive vision.
A Relational Cognitive Bootstrapping Mechanism For Active Perception-Action Learning
Convex Support and Relevance Vector Machines for selective multimodal pattern recognition
ABSTRACT We address the problem of featureless pattern recognition under the assumption that pair... more ABSTRACT We address the problem of featureless pattern recognition under the assumption that pair-wise comparison of objects is arbitrarily scored by real numbers. Such a linear embedding is much more general than the traditional kernel-based approach, which demands positive semi-definiteness of the matrix of object comparisons. This demand is frequently prohibitive and is further complicated if there exist a large number of comparison functions, i.e., multiple modalities of object representation. In these cases, the experimenter typically also has the problem of eliminating redundant modalities and objects. In the context of the general pair-wise comparison space this problem becomes mathematically analogous to that of wrapper-based feature selection. The resulting convex SVM-like training criterion is analogous to Tipping's Relevance Vector Machine, but essentially generalizes it via the presence of a structural parameter controlling the selectivity level.
A generalised solution to the problem of multiple expert fusion
Uploads
Papers by David Windridge