A generative model for online depth fusion

We present a probabilistic, online, depth map fusion framework, whose generative model for the sensor measurement process accurately incorporates both long-range visibility constraints and a spatially varying, probabilistic outlier model. In addition, we propose an inference algorithm that updates the state variables of this model in linear time each frame. Our detailed evaluation compares our approach against several others, demonstrating and explaining the improvements that this model offers, as well as highlighting a problem with all current methods: systemic bias. © 2012 Springer-Verlag.

Bayesian Learning with Dependency Structures via Latent Factors, Mixtures, and Copulas

Bayesian methods offer a flexible and convenient probabilistic learning framework to extract interpretable knowledge from complex and structured data. Such methods can characterize dependencies among multiple levels of hidden variables and share statistical strength across heterogeneous sources. In the first part of this dissertation, we develop two dependent variational inference methods for full posterior approximation in non-conjugate Bayesian models through hierarchical mixture- and copula-based variational proposals, respectively. The proposed methods move beyond the widely used factorized approximation to the posterior and provide generic applicability to a broad class of probabilistic models with minimal model-specific derivations. In the second part of this dissertation, we design probabilistic graphical models to accommodate multimodal data, describe dynamical behaviors and account for task heterogeneity. In particular, the sparse latent factor model is able to reveal common low-dimensional structures from high-dimensional data. We demonstrate the effectiveness of the proposed statistical learning methods on both synthetic and real-world data.

Security and Trust in Distributed Computation

We propose three research problems to explore the relations between trust and security in the setting of distributed computation. In the first problem, we study trust-based adversary detection in distributed consensus computation. The adversaries we consider behave arbitrarily disobeying the consensus protocol. We propose a trust-based consensus algorithm with local and global trust evaluations. The algorithm can be abstracted using a two-layer structure with the top layer running a trust-based consensus algorithm and the bottom layer as a subroutine executing a global trust update scheme. We utilize a set of pre-trusted nodes, headers, to propagate local trust opinions throughout the network. This two-layer framework is flexible in that it can be easily extensible to contain more complicated decision rules, and global trust schemes. The first problem assumes that normal nodes are homogeneous, i.e. it is guaranteed that a normal node always behaves as it is programmed. In the second and third problems however, we assume that nodes are heterogeneous, i.e, given a task, the probability that a node generates a correct answer varies from node to node. The adversaries considered in these two problems are workers from the open crowd who are either investing little efforts in the tasks assigned to them or intentionally give wrong answers to questions. In the second part of the thesis, we consider a typical crowdsourcing task that aggregates input from multiple workers as a problem in information fusion. To cope with the issue of noisy and sometimes malicious input from workers, trust is used to model workers' expertise. In a multi-domain knowledge learning task, however, using scalar-valued trust to model a worker's performance is not sufficient to reflect the worker's trustworthiness in each of the domains. To address this issue, we propose a probabilistic model to jointly infer multi-dimensional trust of workers, multi-domain properties of questions, and true labels of questions. Our model is very flexible and extensible to incorporate metadata associated with questions. To show that, we further propose two extended models, one of which handles input tasks with real-valued features and the other handles tasks with text features by incorporating topic models. Our models can effectively recover trust vectors of workers, which can be very useful in task assignment adaptive to workers' trust in the future. These results can be applied for fusion of information from multiple data sources like sensors, human input, machine learning results, or a hybrid of them. In the second subproblem, we address crowdsourcing with adversaries under logical constraints. We observe that questions are often not independent in real life applications. Instead, there are logical relations between them. Similarly, workers that provide answers are not independent of each other either. Answers given by workers with similar attributes tend to be correlated. Therefore, we propose a novel unified graphical model consisting of two layers. The top layer encodes domain knowledge which allows users to express logical relations using first-order logic rules and the bottom layer encodes a traditional crowdsourcing graphical model. Our model can be seen as a generalized probabilistic soft logic framework that encodes both logical relations and probabilistic dependencies. To solve the collective inference problem efficiently, we have devised a scalable joint inference algorithm based on the alternating direction method of multipliers. The third part of the thesis considers the problem of optimal assignment under budget constraints when workers are unreliable and sometimes malicious. In a real crowdsourcing market, each answer obtained from a worker incurs cost. The cost is associated with both the level of trustworthiness of workers and the difficulty of tasks. Typically, access to expert-level (more trustworthy) workers is more expensive than to average crowd and completion of a challenging task is more costly than a click-away question. In this problem, we address the problem of optimal assignment of heterogeneous tasks to workers of varying trust levels with budget constraints. Specifically, we design a trust-aware task allocation algorithm that takes as inputs the estimated trust of workers and pre-set budget, and outputs the optimal assignment of tasks to workers. We derive the bound of total error probability that relates to budget, trustworthiness of crowds, and costs of obtaining labels from crowds naturally. Higher budget, more trustworthy crowds, and less costly jobs result in a lower theoretical bound. Our allocation scheme does not depend on the specific design of the trust evaluation component. Therefore, it can be combined with generic trust evaluation algorithms.

Interrogating statistical models of music perception

We have developed a new experimental method for interrogating statistical theories of music perception by implementing these theories as generative music algorithms. We call this method Generation in Context. This method differs from most experimental techniques in music perception in that it incorporates aesthetic judgments. Generation In Context is designed to measure percepts for which the musical context is suspected to play an important role. In particular the method is suitable for the study of perceptual parameters which are temporally dynamic. We outline a use of this approach to investigate David Temperley’s (2007) probabilistic melody model, and provide some provisional insights as to what is revealed about the model. We suggest that Temperley’s model could be improved by dynamically modulating the probability distributions according to the changing musical context.

Ontology-based image annotation

With regard to the long-standing problem of the semantic gap between low-level image features and high-level human knowledge, the image retrieval community has recently shifted its emphasis from low-level features analysis to high-level image semantics extrac- tion. User studies reveal that users tend to seek information using high-level semantics. Therefore, image semantics extraction is of great importance to content-based image retrieval because it allows the users to freely express what images they want. Semantic content annotation is the basis for semantic content retrieval. The aim of image anno- tation is to automatically obtain keywords that can be used to represent the content of images. The major research challenges in image semantic annotation are: what is the basic unit of semantic representation? how can the semantic unit be linked to high-level image knowledge? how can the contextual information be stored and utilized for image annotation? In this thesis, the Semantic Web technology (i.e. ontology) is introduced to the image semantic annotation problem. Semantic Web, the next generation web, aims at mak- ing the content of whatever type of media not only understandable to humans but also to machines. Due to the large amounts of multimedia data prevalent on the Web, re- searchers and industries are beginning to pay more attention to the Multimedia Semantic Web. The Semantic Web technology provides a new opportunity for multimedia-based applications, but the research in this area is still in its infancy. Whether ontology can be used to improve image annotation and how to best use ontology in semantic repre- sentation and extraction is still a worth-while investigation. This thesis deals with the problem of image semantic annotation using ontology and machine learning techniques in four phases as below. 1) Salient object extraction. A salient object servers as the basic unit in image semantic extraction as it captures the common visual property of the objects. Image segmen- tation is often used as the �rst step for detecting salient objects, but most segmenta- tion algorithms often fail to generate meaningful regions due to over-segmentation and under-segmentation. We develop a new salient object detection algorithm by combining multiple homogeneity criteria in a region merging framework. 2) Ontology construction. Since real-world objects tend to exist in a context within their environment, contextual information has been increasingly used for improving object recognition. In the ontology construction phase, visual-contextual ontologies are built from a large set of fully segmented and annotated images. The ontologies are composed of several types of concepts (i.e. mid-level and high-level concepts), and domain contextual knowledge. The visual-contextual ontologies stand as a user-friendly interface between low-level features and high-level concepts. 3) Image objects annotation. In this phase, each object is labelled with a mid-level concept in ontologies. First, a set of candidate labels are obtained by training Support Vectors Machines with features extracted from salient objects. After that, contextual knowledge contained in ontologies is used to obtain the �nal labels by removing the ambiguity concepts. 4) Scene semantic annotation. The scene semantic extraction phase is to get the scene type by using both mid-level concepts and domain contextual knowledge in ontologies. Domain contextual knowledge is used to create scene con�guration that describes which objects co-exist with which scene type more frequently. The scene con�guration is represented in a probabilistic graph model, and probabilistic inference is employed to calculate the scene type given an annotated image. To evaluate the proposed methods, a series of experiments have been conducted in a large set of fully annotated outdoor scene images. These include a subset of the Corel database, a subset of the LabelMe dataset, the evaluation dataset of localized semantics in images, the spatial context evaluation dataset, and the segmented and annotated IAPR TC-12 benchmark.

Application of machine learning in fault diagnostics of mechanical systems

A diagnostic method based on Bayesian Networks (probabilistic graphical models) is presented. Unlike conventional diagnostic approaches, in this method instead of focusing on system residuals at one or a few operating points, diagnosis is done by analyzing system behavior patterns over a window of operation. It is shown how this approach can loosen the dependency of diagnostic methods on precise system modeling while maintaining the desired characteristics of fault detection and diagnosis (FDD) tools (fault isolation, robustness, adaptability, and scalability) at a satisfactory level. As an example, the method is applied to fault diagnosis in HVAC systems, an area with considerable modeling and sensor network constraints.

Augmenting traversability maps with ultra-wideband radar to enhance obstacle detection in vegetated environments

Operating in vegetated environments is a major challenge for autonomous robots. Obstacle detection based only on geometric features causes the robot to consider foliage, for example, small grass tussocks that could be easily driven through, as obstacles. Classifying vegetation does not solve this problem since there might be an obstacle hidden behind the vegetation. In addition, dense vegetation typically needs to be considered as an obstacle. This paper addresses this problem by augmenting probabilistic traversability map constructed from laser data with ultra-wideband radar measurements. An adaptive detection threshold and a probabilistic sensor model are developed to convert the radar data to occupancy probabilities. The resulting map captures the fine resolution of the laser map but clears areas from the traversability map that are induced by obstacle-free foliage. Experimental results validate that this method is able to improve the accuracy of traversability maps in vegetated environments.

Systematic integration of human factors in the specification of requirements for new railway technologies

Engineering design processes are necessary to attain the requisite standards of integrity for high-assurance safety-related systems. Additionally, human factors design initiatives can provide critical insights that parameterise their development. Unfortunately, the popular perception of human factors as a “forced marriage” between engineering and psychology often provokes views where the ‘human factor’ is perceived as a threat to systems design. Some popular performance-based standards for developing safety-related systems advocate identifying and managing human factors throughout the system lifecycle. However, they also have a tendency to fall short in their guidance on the application of human factors methods and tools, let alone how the outputs generated can be integrated in to various stages of the design process. This case study describes a project that converged engineering with human factors to develop a safety argument for new low-cost railway level crossing technology for system-wide implementation in Australia. The paper enjoins the perspectives of a software engineer and cognitive psychologist and their involvement in the project over two years of collaborative work to develop a safety argument for low-cost level crossing technology. Safety and reliability requirements were informed by applying human factors analytical tools that supported the evaluation and quantification of human reliability where users interfaced with the technology. The project team was confronted with significant challenges in cross-disciplinary engagement, particularly with the complexities of dealing with incongruences in disciplinary language. They were also encouraged to think ‘outside the box’ as to how users of a system interpreted system states and ehaviour. Importantly, some of these states, while considered safe within the boundary of the constituent systems that implemented safety-related functions, could actually lead the users to engage in deviant behaviour. Psychology explained how user compliance could be eroded to levels that effectively undermined levels of risk reduction afforded by systems. Linking the engineering and psychology disciplines intuitively, overall safety performance was improved by introducing technical requirements and making design decisions that minimized the system states and behaviours that led to user deviancy. As a commentary on the utility of transdisciplinary collaboration for technical specification, the processes used to bridge the two disciplines are conceptualised in a graphical model.

Network Error Correction for Unit-Delay, Memory-Free Networks Using Convolutional Codes

A single source network is said to be memory-free if all of the internal nodes (those except the source and the sinks) do not employ memory but merely send linear combinations of the symbols received at their incoming edges on their outgoing edges. In this work, we introduce network-error correction for single source, acyclic, unit-delay, memory-free networks with coherent network coding for multicast. A convolutional code is designed at the source based on the network code in order to correct network- errors that correspond to any of a given set of error patterns, as long as consecutive errors are separated by a certain interval which depends on the convolutional code selected. Bounds on this interval and the field size required for constructing the convolutional code with the required free distance are also obtained. We illustrate the performance of convolutional network error correcting codes (CNECCs) designed for the unit-delay networks using simulations of CNECCs on an example network under a probabilistic error model.

Damped Belief Propagation Based Near-Optimal Equalization of Severely Delay-Spread UWB MIMO-ISI Channels

In this paper, we present a belief propagation (BP) based equalizer for ultrawideband (UWB) multiple-input multiple-output (MIMO) inter-symbol interference (ISI) channels characterized by severe delay spreads. We employ a Markov random field (MRF) graphical model of the system on which we carry out message passing. The proposed BP equalizer is shown to perform increasingly closer to optimal performance for increasing number of multipath components (MPC) at a much lesser complexity than that of the optimum equalizer. The proposed equalizer performs close to within 0.25 dB of SISO AWGN performance at 10-3 bit error rate on a severely delay-spread MIMO-ISI channel with 20 equal-energy MPCs. We point out that, although MIMO/UWB systems are characterized by fully/densely connected graphical models, the following two proposed features are instrumental in achieving near-optimal performance for large number of MPCs at low complexities: i) use of pairwise compatibility functions in densely connected MRFs, and ii) use of damping of messages.

Functional analysis of ultra high information rates conveyed by rat vibrissal primary afferents

Sensory receptors determine the type and the quantity of information available for perception. Here, we quantified and characterized the information transferred by primary afferents in the rat whisker system using neural system identification. Quantification of ``how much'' information is conveyed by primary afferents, using the direct method (DM), a classical information theoretic tool, revealed that primary afferents transfer huge amounts of information (up to 529 bits/s). Information theoretic analysis of instantaneous spike-triggered kinematic stimulus features was used to gain functional insight on ``what'' is coded by primary afferents. Amongst the kinematic variables tested-position, velocity, and acceleration-primary afferent spikes encoded velocity best. The other two variables contributed to information transfer, but only if combined with velocity. We further revealed three additional characteristics that play a role in information transfer by primary afferents. Firstly, primary afferent spikes show preference for well separated multiple stimuli (i.e., well separated sets of combinations of the three instantaneous kinematic variables). Secondly, neurons are sensitive to short strips of the stimulus trajectory (up to 10 ms pre-spike time), and thirdly, they show spike patterns (precise doublet and triplet spiking). In order to deal with these complexities, we used a flexible probabilistic neuron model fitting mixtures of Gaussians to the spike triggered stimulus distributions, which quantitatively captured the contribution of the mentioned features and allowed us to achieve a full functional analysis of the total information rate indicated by the DM. We found that instantaneous position, velocity, and acceleration explained about 50% of the total information rate. Adding a 10 ms pre-spike interval of stimulus trajectory achieved 80-90%. The final 10-20% were found to be due to non-linear coding by spike bursts.

Streaming covariance selection with applications to adaptive querying in sensor networks

Sensor networks can be naturally represented as graphical models, where the edge set encodes the presence of sparsity in the correlation structure between sensors. Such graphical representations can be valuable for information mining purposes as well as for optimizing bandwidth and battery usage with minimal loss of estimation accuracy. We use a computationally efficient technique for estimating sparse graphical models which fits a sparse linear regression locally at each node of the graph via the Lasso estimator. Using a recently suggested online, temporally adaptive implementation of the Lasso, we propose an algorithm for streaming graphical model selection over sensor networks. With battery consumption minimization applications in mind, we use this algorithm as the basis of an adaptive querying scheme. We discuss implementation issues in the context of environmental monitoring using sensor networks, where the objective is short-term forecasting of local wind direction. The algorithm is tested against real UK weather data and conclusions are drawn about certain tradeoffs inherent in decentralized sensor networks data analysis. © 2010 The Author. Published by Oxford University Press on behalf of The British Computer Society. All rights reserved.

A reason for unreason: returns-based beliefs in game theory

Players cooperate in experiments more than game theory would predict. We introduce the ‘returns-based beliefs’ approach: the expected returns of a particular strategy in proportion to total expected returns of all strategies. Using a decision analytic solution concept, Luce’s (1959) probabilistic choice model, and ‘hyperpriors’ for ambiguity in players’ cooperability, our approach explains empirical observations in various classes of games including the Prisoner’s and Traveler’s Dilemmas. Testing the closeness of fit of our model on Selten and Chmura (2008) data for completely mixed 2 × 2 games shows that with loss aversion, returns-based beliefs explain the data better than other equilibrium concepts.

Occlusive Components Analysis

We study unsupervised learning in a probabilistic generative model for occlusion. The model uses two types of latent variables: one indicates which objects are present in the image, and the other how they are ordered in depth. This depth order then determines how the positions and appearances of the objects present, specified in the model parameters, combine to form the image. We show that the object parameters can be learnt from an unlabelled set of images in which objects occlude one another. Exact maximum-likelihood learning is intractable. However, we show that tractable approximations to Expectation Maximization (EM) can be found if the training images each contain only a small number of objects on average. In numerical experiments it is shown that these approximations recover the correct set of object parameters. Experiments on a novel version of the bars test using colored bars, and experiments on more realistic data, show that the algorithm performs well in extracting the generating causes. Experiments based on the standard bars benchmark test for object learning show that the algorithm performs well in comparison to other recent component extraction approaches. The model and the learning algorithm thus connect research on occlusion with the research field of multiple-causes component extraction methods.