Specialized Mappings Architecture with Applications to Vision-based Estimation of Articulated Body Pose

A fundamental task of vision systems is to infer the state of the world given some form of visual observations. From a computational perspective, this often involves facing an ill-posed problem; e.g., information is lost via projection of the 3D world into a 2D image. Solution of an ill-posed problem requires additional information, usually provided as a model of the underlying process. It is important that the model be both computationally feasible as well as theoretically well-founded. In this thesis, a probabilistic, nonlinear supervised computational learning model is proposed: the Specialized Mappings Architecture (SMA). The SMA framework is demonstrated in a computer vision system that can estimate the articulated pose parameters of a human body or human hands, given images obtained via one or more uncalibrated cameras. The SMA consists of several specialized forward mapping functions that are estimated automatically from training data, and a possibly known feedback function. Each specialized function maps certain domains of the input space (e.g., image features) onto the output space (e.g., articulated body parameters). A probabilistic model for the architecture is first formalized. Solutions to key algorithmic problems are then derived: simultaneous learning of the specialized domains along with the mapping functions, as well as performing inference given inputs and a feedback function. The SMA employs a variant of the Expectation-Maximization algorithm and approximate inference. The approach allows the use of alternative conditional independence assumptions for learning and inference, which are derived from a forward model and a feedback model. Experimental validation of the proposed approach is conducted in the task of estimating articulated body pose from image silhouettes. Accuracy and stability of the SMA framework is tested using artificial data sets, as well as synthetic and real video sequences of human bodies and hands.

Scheduling Flows with Unknown Sizes: Approximate Analysis

Previous studies have shown that giving preferential treatment to short jobs helps reduce the average system response time, especially when the job size distribution possesses the heavy-tailed property. Since it has been shown that the TCP flow length distribution also has the same property, it is natural to let short TCP flows enjoy better service inside the network. Analyzing such discriminatory system requires modification to traditional job scheduling models since usually network traffic managers do not have detailed knowledge about individual flows such as their lengths. The Multi-Level (ML) queue, proposed by Kleinrock, can b e used to characterize such system. In an ML queueing system, the priority of a flow is reduced as the flow stays longer. We present an approximate analysis of the ML queueing system to obtain a closed-form solution of the average system response time function for general flow size distributions. We show that the response time of short flows can be significantly reduced without penalizing long flows.

Fast Approximate Reconciliation of Set Differences

We present new, simple, efficient data structures for approximate reconciliation of set differences, a useful standalone primitive for peer-to-peer networks and a natural subroutine in methods for exact reconciliation. In the approximate reconciliation problem, peers A and B respectively have subsets of elements SA and SB of a large universe U. Peer A wishes to send a short message M to peer B with the goal that B should use M to determine as many elements in the set SB–SA as possible. To avoid the expense of round trip communication times, we focus on the situation where a single message M is sent. We motivate the performance tradeoffs between message size, accuracy and computation time for this problem with a straightforward approach using Bloom filters. We then introduce approximation reconciliation trees, a more computationally efficient solution that combines techniques from Patricia tries, Merkle trees, and Bloom filters. We present an analysis of approximation reconciliation trees and provide experimental results comparing the various methods proposed for approximate reconciliation.

Indexing Distances in Large Graphs and Applications in Search Tasks

This thesis elaborates on the problem of preprocessing a large graph so that single-pair shortest-path queries can be answered quickly at runtime. Computing shortest paths is a well studied problem, but exact algorithms do not scale well to real-world huge graphs in applications that require very short response time. The focus is on approximate methods for distance estimation, in particular in landmarks-based distance indexing. This approach involves choosing some nodes as landmarks and computing (offline), for each node in the graph its embedding, i.e., the vector of its distances from all the landmarks. At runtime, when the distance between a pair of nodes is queried, it can be quickly estimated by combining the embeddings of the two nodes. Choosing optimal landmarks is shown to be hard and thus heuristic solutions are employed. Given a budget of memory for the index, which translates directly into a budget of landmarks, different landmark selection strategies can yield dramatically different results in terms of accuracy. A number of simple methods that scale well to large graphs are therefore developed and experimentally compared. The simplest methods choose central nodes of the graph, while the more elaborate ones select central nodes that are also far away from one another. The efficiency of the techniques presented in this thesis is tested experimentally using five different real world graphs with millions of edges; for a given accuracy, they require as much as 250 times less space than the current approach which considers selecting landmarks at random. Finally, they are applied in two important problems arising naturally in large-scale graphs, namely social search and community detection.

Fast Shortest Path Distance Estimation in Large Networks

We study the problem of preprocessing a large graph so that point-to-point shortest-path queries can be answered very fast. Computing shortest paths is a well studied problem, but exact algorithms do not scale to huge graphs encountered on the web, social networks, and other applications. In this paper we focus on approximate methods for distance estimation, in particular using landmark-based distance indexing. This approach involves selecting a subset of nodes as landmarks and computing (offline) the distances from each node in the graph to those landmarks. At runtime, when the distance between a pair of nodes is needed, we can estimate it quickly by combining the precomputed distances of the two nodes to the landmarks. We prove that selecting the optimal set of landmarks is an NP-hard problem, and thus heuristic solutions need to be employed. Given a budget of memory for the index, which translates directly into a budget of landmarks, different landmark selection strategies can yield dramatically different results in terms of accuracy. A number of simple methods that scale well to large graphs are therefore developed and experimentally compared. The simplest methods choose central nodes of the graph, while the more elaborate ones select central nodes that are also far away from one another. The efficiency of the suggested techniques is tested experimentally using five different real world graphs with millions of edges; for a given accuracy, they require as much as 250 times less space than the current approach in the literature which considers selecting landmarks at random. Finally, we study applications of our method in two problems arising naturally in large-scale networks, namely, social search and community detection.

Efficient Nearest Neighbor Classification Using a Cascade of Approximate Similarity Measures

Nearest neighbor classification using shape context can yield highly accurate results in a number of recognition problems. Unfortunately, the approach can be too slow for practical applications, and thus approximation strategies are needed to make shape context practical. This paper proposes a method for efficient and accurate nearest neighbor classification in non-Euclidean spaces, such as the space induced by the shape context measure. First, a method is introduced for constructing a Euclidean embedding that is optimized for nearest neighbor classification accuracy. Using that embedding, multiple approximations of the underlying non-Euclidean similarity measure are obtained, at different levels of accuracy and efficiency. The approximations are automatically combined to form a cascade classifier, which applies the slower approximations only to the hardest cases. Unlike typical cascade-of-classifiers approaches, that are applied to binary classification problems, our method constructs a cascade for a multiclass problem. Experiments with a standard shape data set indicate that a two-to-three order of magnitude speed up is gained over the standard shape context classifier, with minimal losses in classification accuracy.

Articulated Pose Estimation in a Learned Smooth Space of Feasible Solutions

A learning based framework is proposed for estimating human body pose from a single image. Given a differentiable function that maps from pose space to image feature space, the goal is to invert the process: estimate the pose given only image features. The inversion is an ill-posed problem as the inverse mapping is a one to many process. Hence multiple solutions exist, and it is desirable to restrict the solution space to a smaller subset of feasible solutions. For example, not all human body poses are feasible due to anthropometric constraints. Since the space of feasible solutions may not admit a closed form description, the proposed framework seeks to exploit machine learning techniques to learn an approximation that is smoothly parameterized over such a space. One such technique is Gaussian Process Latent Variable Modelling. Scaled conjugate gradient is then used find the best matching pose in the space of feasible solutions when given an input image. The formulation allows easy incorporation of various constraints, e.g. temporal consistency and anthropometric constraints. The performance of the proposed approach is evaluated in the task of upper-body pose estimation from silhouettes and compared with the Specialized Mapping Architecture. The estimation accuracy of the Specialized Mapping Architecture is at least one standard deviation worse than the proposed approach in the experiments with synthetic data. In experiments with real video of humans performing gestures, the proposed approach produces qualitatively better estimation results.

Improved kinematic sensing for motion control applications

New compensation methods are presented that can greatly reduce the slit errors (i.e. transition location errors) and interval errors induced due to non-idealities in optical incremental encoders (square-wave). An M/T-type, constant sample-time digital tachometer (CSDT) is selected for measuring the velocity of the sensor drives. Using this data, three encoder compensation techniques (two pseudoinverse based methods and an iterative method) are presented that improve velocity measurement accuracy. The methods do not require precise knowledge of shaft velocity. During the initial learning stage of the compensation algorithm (possibly performed in-situ), slit errors/interval errors are calculated through pseudoinversebased solutions of simple approximate linear equations, which can provide fast solutions, or an iterative method that requires very little memory storage. Subsequent operation of the motion system utilizes adjusted slit positions for more accurate velocity calculation. In the theoretical analysis of the compensation of encoder errors, encoder error sources such as random electrical noise and error in estimated reference velocity are considered. Initially, the proposed learning compensation techniques are validated by implementing the algorithms in MATLAB software, showing a 95% to 99% improvement in velocity measurement. However, it is also observed that the efficiency of the algorithm decreases with the higher presence of non-repetitive random noise and/or with the errors in reference velocity calculations. The performance improvement in velocity measurement is also demonstrated experimentally using motor-drive systems, each of which includes a field-programmable gate array (FPGA) for CSDT counting/timing purposes, and a digital-signal-processor (DSP). Results from open-loop velocity measurement and closed-loop servocontrol applications, on three optical incremental square-wave encoders and two motor drives, are compiled. While implementing these algorithms experimentally on different drives (with and without a flywheel) and on encoders of different resolutions, slit error reductions of 60% to 86% are obtained (typically approximately 80%).

A new proposition on the martingale representation theorem and on the approximate hedging of contingent claim in mean-variance criterion

In this work we revisit the problem of the hedging of contingent claim using mean-square criterion. We prove that in incomplete market, some probability measure can be identified so that becomes -martingale under .This is in fact a new proposition on the martingale representation theorem. The new results also identify a weight function that serves to be an approximation to the Radon-Nikodým derivative of the unique neutral martingale measure.

Multiple Dynamic Processes Contribute to the Complex Steady Shear Behavior of Cross-Linked Supramolecular Networks of Semidilute Entangled Polymer Solutions.

Molecular theories of shear thickening and shear thinning in associative polymer networks are typically united in that they involve a single kinetic parameter that describes the network -- a relaxation time that is related to the lifetime of the associative bonds. Here we report the steady-shear behavior of two structurally identical metallo-supramolecular polymer networks, for which single-relaxation parameter models break down in dramatic fashion. The networks are formed by the addition of reversible cross-linkers to semidilute entangled solutions of PVP in DMSO, and they differ only in the lifetime of the reversible cross-links. Shear thickening is observed for cross-linkers that have a slower dissociation rate (17 s(-1)), while shear thinning is observed for samples that have a faster dissociation rate (ca. 1400 s(-1)). The difference in the steady shear behavior of the unentangled vs. entangled regime reveals an unexpected, additional competing relaxation, ascribed to topological disentanglement in the semidilute entangled regime that contributes to the rheological properties.

Magnetic resonance water proton relaxation in protein solutions and tissue: T(1rho) dispersion characterization.

BACKGROUND: Image contrast in clinical MRI is often determined by differences in tissue water proton relaxation behavior. However, many aspects of water proton relaxation in complex biological media, such as protein solutions and tissue are not well understood, perhaps due to the limited empirical data. PRINCIPAL FINDINGS: Water proton T(1), T(2), and T(1rho) of protein solutions and tissue were measured systematically under multiple conditions. Crosslinking or aggregation of protein decreased T(2) and T(1rho), but did not change high-field T(1). T(1rho) dispersion profiles were similar for crosslinked protein solutions, myocardial tissue, and cartilage, and exhibited power law behavior with T(1rho)(0) values that closely approximated T(2). The T(1rho) dispersion of mobile protein solutions was flat above 5 kHz, but showed a steep curve below 5 kHz that was sensitive to changes in pH. The T(1rho) dispersion of crosslinked BSA and cartilage in DMSO solvent closely resembled that of water solvent above 5 kHz but showed decreased dispersion below 5 kHz. CONCLUSIONS: Proton exchange is a minor pathway for tissue T(1) and T(1rho) relaxation above 5 kHz. Potential models for relaxation are discussed, however the same molecular mechanism appears to be responsible across 5 decades of frequencies from T(1rho) to T(1).

Antibiotic resistance-the need for global solutions.

The causes of antibiotic resistance are complex and include human behaviour at many levels of society; the consequences affect everybody in the world. Similarities with climate change are evident. Many efforts have been made to describe the many different facets of antibiotic resistance and the interventions needed to meet the challenge. However, coordinated action is largely absent, especially at the political level, both nationally and internationally. Antibiotics paved the way for unprecedented medical and societal developments, and are today indispensible in all health systems. Achievements in modern medicine, such as major surgery, organ transplantation, treatment of preterm babies, and cancer chemotherapy, which we today take for granted, would not be possible without access to effective treatment for bacterial infections. Within just a few years, we might be faced with dire setbacks, medically, socially, and economically, unless real and unprecedented global coordinated actions are immediately taken. Here, we describe the global situation of antibiotic resistance, its major causes and consequences, and identify key areas in which action is urgently needed.

Clinical Research Environment in India: Challenges and Proposed Solutions.

India has compelling need and keen aspirations for indigenous clinical research. Notwithstanding this need and previously reported growth the expected expansion of Indian clinical research has not materialized. We reviewed the scientific literature, lay press reports, and ClinicalTrials.gov data for information and commentary on projections, progress, and impediments associated with clinical trials in India. We also propose targeted solutions to identified challenges. The Indian clinical trial sector grew by (+) 20.3% CAGR (compound annual growth rate) between 2005 and 2010 and contracted by (-) 14.6% CAGR between 2010 and 2013. Phase-1 trials grew by (+) 43.5% CAGR from 2005-2013, phase-2 trials grew by (+) 19.8% CAGR from 2005-2009 and contracted by (-) 12.6% CAGR from 2009-2013, and phase-3 trials grew by (+) 13.0% CAGR from 2005-2010 and contracted by (-) 28.8% CAGR from 2010-2013. This was associated with a slowing of the regulatory approval process, increased media coverage and activist engagement, and accelerated development of regulatory guidelines and recuperative initiatives. We propose the following as potential targets for restorative interventions: Regulatory overhaul (leadership and enforcement of regulations, resolution of ambiguity in regulations, staffing, training, guidelines, and ethical principles [e.g., compensation]).Education and training of research professionals, clinicians, and regulators.Public awareness and empowerment. After a peak in 2009-2010, the clinical research sector in India appears to be experiencing a contraction. There are indications of challenges in regulatory enforcement of guidelines; training of clinical research professionals; and awareness, participation, partnership, and the general image amongst the non-professional media and public. Preventative and corrective principles and interventions are outlined with the goal of realizing the clinical research potential in India.

Identifying Barriers and Practical Solutions to Conducting Site-Based Research in North America: Exploring Acute Heart Failure Trials As a Case Study.

Although the prognosis of ambulatory heart failure (HF) has improved dramatically there have been few advances in the management of acute HF (AHF). Despite regional differences in patient characteristics, background therapy, and event rates, AHF clinical trial enrollment has transitioned from North America and Western Europe to Eastern Europe, South America, and Asia-Pacific where regulatory burden and cost of conducting research may be less prohibitive. It is unclear if the results of clinical trials conducted outside of North America are generalizable to US patient populations. This article uses AHF as a paradigm and identifies barriers and practical solutions to successfully conducting site-based research in North America.

On approximate calculations

The ability to undertake approximate calcutations and to get a “rough feel” for data is an important skill which must not be overlooked.