Identifiying human MHC supertypes using bioinformatic methods

Classification of MHC molecules into supertypes in terms of peptide-binding specificities is an important issue, with direct implications for the development of epitope-based vaccines with wide population coverage. In view of extremely high MHC polymorphism (948 class I and 633 class II HLA alleles) the experimental solution of this task is presently impossible. In this study, we describe a bioinformatics strategy for classifying MHC molecules into supertypes using information drawn solely from three-dimensional protein structure. Two chemometric techniques–hierarchical clustering and principal component analysis–were used independently on a set of 783 HLA class I molecules to identify supertypes based on structural similarities and molecular interaction fields calculated for the peptide binding site. Eight supertypes were defined: A2, A3, A24, B7, B27, B44, C1, and C4. The two techniques gave 77% consensus, i.e., 605 HLA class I alleles were classified in the same supertype by both methods. The proposed strategy allowed “supertype fingerprints” to be identified. Thus, the A2 supertype fingerprint is Tyr9/Phe9, Arg97, and His114 or Tyr116; the A3-Tyr9/Phe9/Ser9, Ile97/Met97 and Glu114 or Asp116; the A24-Ser9 and Met97; the B7-Asn63 and Leu81; the B27-Glu63 and Leu81; for B44-Ala81; the C1-Ser77; and the C4-Asn77. action fields calculated for the peptide binding site. Eight supertypes were defined: A2, A3, A24, B7, B27, B44, C1, and C4. The two techniques gave 77% consensus, i.e., 605 HLA class I alleles were classified in the same supertype by both methods. The proposed strategy allowed “supertype fingerprints” to be identified. Thus, the A2 supertype fingerprint is Tyr9/Phe9, Arg97, and His114 or Tyr116; the A3-Tyr9/Phe9/Ser9, Ile97/Met97 and Glu114 or Asp116; the A24-Ser9 and Met97; the B7-Asn63 and Leu81; the B27-Glu63 and Leu81; for B44-Ala81; the C1-Ser77; and the C4-Asn77.

Editorial overview:new protein production tools for structural biology

Full text: The idea of producing proteins from recombinant DNA hatched almost half a century ago. In his PhD thesis, Peter Lobban foresaw the prospect of inserting foreign DNA (from any source, including mammalian cells) into the genome of a λ phage in order to detect and recover protein products from Escherichia coli [ 1 and 2]. Only a few years later, in 1977, Herbert Boyer and his colleagues succeeded in the first ever expression of a peptide-coding gene in E. coli — they produced recombinant somatostatin [ 3] followed shortly after by human insulin. The field has advanced enormously since those early days and today recombinant proteins have become indispensable in advancing research and development in all fields of the life sciences. Structural biology, in particular, has benefitted tremendously from recombinant protein biotechnology, and an overwhelming proportion of the entries in the Protein Data Bank (PDB) are based on heterologously expressed proteins. Nonetheless, synthesizing, purifying and stabilizing recombinant proteins can still be thoroughly challenging. For example, the soluble proteome is organized to a large part into multicomponent complexes (in humans often comprising ten or more subunits), posing critical challenges for recombinant production. A third of all proteins in cells are located in the membrane, and pose special challenges that require a more bespoke approach. Recent advances may now mean that even these most recalcitrant of proteins could become tenable structural biology targets on a more routine basis. In this special issue, we examine progress in key areas that suggests this is indeed the case. Our first contribution examines the importance of understanding quality control in the host cell during recombinant protein production, and pays particular attention to the synthesis of recombinant membrane proteins. A major challenge faced by any host cell factory is the balance it must strike between its own requirements for growth and the fact that its cellular machinery has essentially been hijacked by an expression construct. In this context, Bill and von der Haar examine emerging insights into the role of the dependent pathways of translation and protein folding in defining high-yielding recombinant membrane protein production experiments for the common prokaryotic and eukaryotic expression hosts. Rather than acting as isolated entities, many membrane proteins form complexes to carry out their functions. To understand their biological mechanisms, it is essential to study the molecular structure of the intact membrane protein assemblies. Recombinant production of membrane protein complexes is still a formidable, at times insurmountable, challenge. In these cases, extraction from natural sources is the only option to prepare samples for structural and functional studies. Zorman and co-workers, in our second contribution, provide an overview of recent advances in the production of multi-subunit membrane protein complexes and highlight recent achievements in membrane protein structural research brought about by state-of-the-art near-atomic resolution cryo-electron microscopy techniques. E. coli has been the dominant host cell for recombinant protein production. Nonetheless, eukaryotic expression systems, including yeasts, insect cells and mammalian cells, are increasingly gaining prominence in the field. The yeast species Pichia pastoris, is a well-established recombinant expression system for a number of applications, including the production of a range of different membrane proteins. Byrne reviews high-resolution structures that have been determined using this methylotroph as an expression host. Although it is not yet clear why P. pastoris is suited to producing such a wide range of membrane proteins, its ease of use and the availability of diverse tools that can be readily implemented in standard bioscience laboratories mean that it is likely to become an increasingly popular option in structural biology pipelines. The contribution by Columbus concludes the membrane protein section of this volume. In her overview of post-expression strategies, Columbus surveys the four most common biochemical approaches for the structural investigation of membrane proteins. Limited proteolysis has successfully aided structure determination of membrane proteins in many cases. Deglycosylation of membrane proteins following production and purification analysis has also facilitated membrane protein structure analysis. Moreover, chemical modifications, such as lysine methylation and cysteine alkylation, have proven their worth to facilitate crystallization of membrane proteins, as well as NMR investigations of membrane protein conformational sampling. Together these approaches have greatly facilitated the structure determination of more than 40 membrane proteins to date. It may be an advantage to produce a target protein in mammalian cells, especially if authentic post-translational modifications such as glycosylation are required for proper activity. Chinese Hamster Ovary (CHO) cells and Human Embryonic Kidney (HEK) 293 cell lines have emerged as excellent hosts for heterologous production. The generation of stable cell-lines is often an aspiration for synthesizing proteins expressed in mammalian cells, in particular if high volumetric yields are to be achieved. In his report, Buessow surveys recent structures of proteins produced using stable mammalian cells and summarizes both well-established and novel approaches to facilitate stable cell-line generation for structural biology applications. The ambition of many biologists is to observe a protein's structure in the native environment of the cell itself. Until recently, this seemed to be more of a dream than a reality. Advances in nuclear magnetic resonance (NMR) spectroscopy techniques, however, have now made possible the observation of mechanistic events at the molecular level of protein structure. Smith and colleagues, in an exciting contribution, review emerging ‘in-cell NMR’ techniques that demonstrate the potential to monitor biological activities by NMR in real time in native physiological environments. A current drawback of NMR as a structure determination tool derives from size limitations of the molecule under investigation and the structures of large proteins and their complexes are therefore typically intractable by NMR. A solution to this challenge is the use of selective isotope labeling of the target protein, which results in a marked reduction of the complexity of NMR spectra and allows dynamic processes even in very large proteins and even ribosomes to be investigated. Kerfah and co-workers introduce methyl-specific isotopic labeling as a molecular tool-box, and review its applications to the solution NMR analysis of large proteins. Tyagi and Lemke next examine single-molecule FRET and crosslinking following the co-translational incorporation of non-canonical amino acids (ncAAs); the goal here is to move beyond static snap-shots of proteins and their complexes and to observe them as dynamic entities. The encoding of ncAAs through codon-suppression technology allows biomolecules to be investigated with diverse structural biology methods. In their article, Tyagi and Lemke discuss these approaches and speculate on the design of improved host organisms for ‘integrative structural biology research’. Our volume concludes with two contributions that resolve particular bottlenecks in the protein structure determination pipeline. The contribution by Crepin and co-workers introduces the concept of polyproteins in contemporary structural biology. Polyproteins are widespread in nature. They represent long polypeptide chains in which individual smaller proteins with different biological function are covalently linked together. Highly specific proteases then tailor the polyprotein into its constituent proteins. Many viruses use polyproteins as a means of organizing their proteome. The concept of polyproteins has now been exploited successfully to produce hitherto inaccessible recombinant protein complexes. For instance, by means of a self-processing synthetic polyprotein, the influenza polymerase, a high-value drug target that had remained elusive for decades, has been produced, and its high-resolution structure determined. In the contribution by Desmyter and co-workers, a further, often imposing, bottleneck in high-resolution protein structure determination is addressed: The requirement to form stable three-dimensional crystal lattices that diffract incident X-ray radiation to high resolution. Nanobodies have proven to be uniquely useful as crystallization chaperones, to coax challenging targets into suitable crystal lattices. Desmyter and co-workers review the generation of nanobodies by immunization, and highlight the application of this powerful technology to the crystallography of important protein specimens including G protein-coupled receptors (GPCRs). Recombinant protein production has come a long way since Peter Lobban's hypothesis in the late 1960s, with recombinant proteins now a dominant force in structural biology. The contributions in this volume showcase an impressive array of inventive approaches that are being developed and implemented, ever increasing the scope of recombinant technology to facilitate the determination of elusive protein structures. Powerful new methods from synthetic biology are further accelerating progress. Structure determination is now reaching into the living cell with the ultimate goal of observing functional molecular architectures in action in their native physiological environment. We anticipate that even the most challenging protein assemblies will be tackled by recombinant technology in the near future.

A Bimodality Test in High Dimensions

We present a test for identifying clusters in high dimensional data based on the k-means algorithm when the null hypothesis is spherical normal. We show that projection techniques used for evaluating validity of clusters may be misleading for such data. In particular, we demonstrate that increasingly well-separated clusters are identified as the dimensionality increases, when no such clusters exist. Furthermore, in a case of true bimodality, increasing the dimensionality makes identifying the correct clusters more difficult. In addition to the original conservative test, we propose a practical test with the same asymptotic behavior that performs well for a moderate number of points and moderate dimensionality. ACM Computing Classification System (1998): I.5.3.

Generalized KPCA by adaptive rules in feature space

Principal component analysis (PCA) is well recognized in dimensionality reduction, and kernel PCA (KPCA) has also been proposed in statistical data analysis. However, KPCA fails to detect the nonlinear structure of data well when outliers exist. To reduce this problem, this paper presents a novel algorithm, named iterative robust KPCA (IRKPCA). IRKPCA works well in dealing with outliers, and can be carried out in an iterative manner, which makes it suitable to process incremental input data. As in the traditional robust PCA (RPCA), a binary field is employed for characterizing the outlier process, and the optimization problem is formulated as maximizing marginal distribution of a Gibbs distribution. In this paper, this optimization problem is solved by stochastic gradient descent techniques. In IRKPCA, the outlier process is in a high-dimensional feature space, and therefore kernel trick is used. IRKPCA can be regarded as a kernelized version of RPCA and a robust form of kernel Hebbian algorithm. Experimental results on synthetic data demonstrate the effectiveness of IRKPCA. © 2010 Taylor & Francis.

Improved data visualisation through multiple dissimilarity modelling

Popular dimension reduction and visualisation algorithms rely on the assumption that input dissimilarities are typically Euclidean, for instance Metric Multidimensional Scaling, t-distributed Stochastic Neighbour Embedding and the Gaussian Process Latent Variable Model. It is well known that this assumption does not hold for most datasets and often high-dimensional data sits upon a manifold of unknown global geometry. We present a method for improving the manifold charting process, coupled with Elastic MDS, such that we no longer assume that the manifold is Euclidean, or of any particular structure. We draw on the benefits of different dissimilarity measures allowing for the relative responsibilities, under a linear combination, to drive the visualisation process.

DIMUSE: An integrated framework for distributed multimedia system with database management and security support

With the recent explosion in the complexity and amount of digital multimedia data, there has been a huge impact on the operations of various organizations in distinct areas, such as government services, education, medical care, business, entertainment, etc. To satisfy the growing demand of multimedia data management systems, an integrated framework called DIMUSE is proposed and deployed for distributed multimedia applications to offer a full scope of multimedia related tools and provide appealing experiences for the users. This research mainly focuses on video database modeling and retrieval by addressing a set of core challenges. First, a comprehensive multimedia database modeling mechanism called Hierarchical Markov Model Mediator (HMMM) is proposed to model high dimensional media data including video objects, low-level visual/audio features, as well as historical access patterns and frequencies. The associated retrieval and ranking algorithms are designed to support not only the general queries, but also the complicated temporal event pattern queries. Second, system training and learning methodologies are incorporated such that user interests are mined efficiently to improve the retrieval performance. Third, video clustering techniques are proposed to continuously increase the searching speed and accuracy by architecting a more efficient multimedia database structure. A distributed video management and retrieval system is designed and implemented to demonstrate the overall performance. The proposed approach is further customized for a mobile-based video retrieval system to solve the perception subjectivity issue by considering individual user's profile. Moreover, to deal with security and privacy issues and concerns in distributed multimedia applications, DIMUSE also incorporates a practical framework called SMARXO, which supports multilevel multimedia security control. SMARXO efficiently combines role-based access control (RBAC), XML and object-relational database management system (ORDBMS) to achieve the target of proficient security control. A distributed multimedia management system named DMMManager (Distributed MultiMedia Manager) is developed with the proposed framework DEMUR; to support multimedia capturing, analysis, retrieval, authoring and presentation in one single framework.

Exploring the integration of data mining and data visualization

Due to the rapid advances in computing and sensing technologies, enormous amounts of data are being generated everyday in various applications. The integration of data mining and data visualization has been widely used to analyze these massive and complex data sets to discover hidden patterns. For both data mining and visualization to be effective, it is important to include the visualization techniques in the mining process and to generate the discovered patterns for a more comprehensive visual view. In this dissertation, four related problems: dimensionality reduction for visualizing high dimensional datasets, visualization-based clustering evaluation, interactive document mining, and multiple clusterings exploration are studied to explore the integration of data mining and data visualization. In particular, we 1) propose an efficient feature selection method (reliefF + mRMR) for preprocessing high dimensional datasets; 2) present DClusterE to integrate cluster validation with user interaction and provide rich visualization tools for users to examine document clustering results from multiple perspectives; 3) design two interactive document summarization systems to involve users efforts and generate customized summaries from 2D sentence layouts; and 4) propose a new framework which organizes the different input clusterings into a hierarchical tree structure and allows for interactive exploration of multiple clustering solutions.

An Indexing Structure and Application Model for Vehicles Moving on Road Networks

Moving objects database systems are the most challenging sub-category among Spatio-Temporal database systems. A database system that updates in real-time the location information of GPS-equipped moving vehicles has to meet even stricter requirements. Currently existing data storage models and indexing mechanisms work well only when the number of moving objects in the system is relatively small. This dissertation research aimed at the real-time tracking and history retrieval of massive numbers of vehicles moving on road networks. A total solution has been provided for the real-time update of the vehicles’ location and motion information, range queries on current and history data, and prediction of vehicles’ movement in the near future. To achieve these goals, a new approach called Segmented Time Associated to Partitioned Space (STAPS) was first proposed in this dissertation for building and manipulating the indexing structures for moving objects databases. Applying the STAPS approach, an indexing structure of associating a time interval tree to each road segment was developed for real-time database systems of vehicles moving on road networks. The indexing structure uses affordable storage to support real-time data updates and efficient query processing. The data update and query processing performance it provides is consistent without restrictions such as a time window or assuming linear moving trajectories. An application system design based on distributed system architecture with centralized organization was developed to maximally support the proposed data and indexing structures. The suggested system architecture is highly scalable and flexible. Finally, based on a real-world application model of vehicles moving in region-wide, main issues on the implementation of such a system were addressed.

Detection and Classification of Whale Acoustic Signals

This dissertation focuses on two vital challenges in relation to whale acoustic signals: detection and classification.

In detection, we evaluated the influence of the uncertain ocean environment on the spectrogram-based detector, and derived the likelihood ratio of the proposed Short Time Fourier Transform detector. Experimental results showed that the proposed detector outperforms detectors based on the spectrogram. The proposed detector is more sensitive to environmental changes because it includes phase information.

In classification, our focus is on finding a robust and sparse representation of whale vocalizations. Because whale vocalizations can be modeled as polynomial phase signals, we can represent the whale calls by their polynomial phase coefficients. In this dissertation, we used the Weyl transform to capture chirp rate information, and used a two dimensional feature set to represent whale vocalizations globally. Experimental results showed that our Weyl feature set outperforms chirplet coefficients and MFCC (Mel Frequency Cepstral Coefficients) when applied to our collected data.

Since whale vocalizations can be represented by polynomial phase coefficients, it is plausible that the signals lie on a manifold parameterized by these coefficients. We also studied the intrinsic structure of high dimensional whale data by exploiting its geometry. Experimental results showed that nonlinear mappings such as Laplacian Eigenmap and ISOMAP outperform linear mappings such as PCA and MDS, suggesting that the whale acoustic data is nonlinear.

We also explored deep learning algorithms on whale acoustic data. We built each layer as convolutions with either a PCA filter bank (PCANet) or a DCT filter bank (DCTNet). With the DCT filter bank, each layer has different a time-frequency scale representation, and from this, one can extract different physical information. Experimental results showed that our PCANet and DCTNet achieve high classification rate on the whale vocalization data set. The word error rate of the DCTNet feature is similar to the MFSC in speech recognition tasks, suggesting that the convolutional network is able to reveal acoustic content of speech signals.

Argument structure and specific language impairment: retrospect and prospect

This article reflects a collaboration between the Universities of Groningen and Reading of which Frans Zwarts was the promoter. One of the outcomes was a close attention to the learning of various aspects of argument structure by children with specific language impairment (SLI) in Dutch and English. At that time and since, the focus on deficits in grammatical morphology in these children has left verb complementation as something of a syntactic Cinderella. Here we review the findings from our studies in the 1990s. We confirm that children with SLI in both languages have problems with verb specificity, with argument structure alternations and with resultative verb predicates. The very limited number of subsequent studies on verb syntax appear to support our findings. We conclude that this is an area which will repay further scrutiny – it is high time argument structure received an invitation to the ball.

Bayesian Inference in Large-scale Problems

Many modern applications fall into the category of "large-scale" statistical problems, in which both the number of observations n and the number of features or parameters p may be large. Many existing methods focus on point estimation, despite the continued relevance of uncertainty quantification in the sciences, where the number of parameters to estimate often exceeds the sample size, despite huge increases in the value of n typically seen in many fields. Thus, the tendency in some areas of industry to dispense with traditional statistical analysis on the basis that "n=all" is of little relevance outside of certain narrow applications. The main result of the Big Data revolution in most fields has instead been to make computation much harder without reducing the importance of uncertainty quantification. Bayesian methods excel at uncertainty quantification, but often scale poorly relative to alternatives. This conflict between the statistical advantages of Bayesian procedures and their substantial computational disadvantages is perhaps the greatest challenge facing modern Bayesian statistics, and is the primary motivation for the work presented here.

Two general strategies for scaling Bayesian inference are considered. The first is the development of methods that lend themselves to faster computation, and the second is design and characterization of computational algorithms that scale better in n or p. In the first instance, the focus is on joint inference outside of the standard problem of multivariate continuous data that has been a major focus of previous theoretical work in this area. In the second area, we pursue strategies for improving the speed of Markov chain Monte Carlo algorithms, and characterizing their performance in large-scale settings. Throughout, the focus is on rigorous theoretical evaluation combined with empirical demonstrations of performance and concordance with the theory.

One topic we consider is modeling the joint distribution of multivariate categorical data, often summarized in a contingency table. Contingency table analysis routinely relies on log-linear models, with latent structure analysis providing a common alternative. Latent structure models lead to a reduced rank tensor factorization of the probability mass function for multivariate categorical data, while log-linear models achieve dimensionality reduction through sparsity. Little is known about the relationship between these notions of dimensionality reduction in the two paradigms. In Chapter 2, we derive several results relating the support of a log-linear model to nonnegative ranks of the associated probability tensor. Motivated by these findings, we propose a new collapsed Tucker class of tensor decompositions, which bridge existing PARAFAC and Tucker decompositions, providing a more flexible framework for parsimoniously characterizing multivariate categorical data. Taking a Bayesian approach to inference, we illustrate empirical advantages of the new decompositions.

Latent class models for the joint distribution of multivariate categorical, such as the PARAFAC decomposition, data play an important role in the analysis of population structure. In this context, the number of latent classes is interpreted as the number of genetically distinct subpopulations of an organism, an important factor in the analysis of evolutionary processes and conservation status. Existing methods focus on point estimates of the number of subpopulations, and lack robust uncertainty quantification. Moreover, whether the number of latent classes in these models is even an identified parameter is an open question. In Chapter 3, we show that when the model is properly specified, the correct number of subpopulations can be recovered almost surely. We then propose an alternative method for estimating the number of latent subpopulations that provides good quantification of uncertainty, and provide a simple procedure for verifying that the proposed method is consistent for the number of subpopulations. The performance of the model in estimating the number of subpopulations and other common population structure inference problems is assessed in simulations and a real data application.

In contingency table analysis, sparse data is frequently encountered for even modest numbers of variables, resulting in non-existence of maximum likelihood estimates. A common solution is to obtain regularized estimates of the parameters of a log-linear model. Bayesian methods provide a coherent approach to regularization, but are often computationally intensive. Conjugate priors ease computational demands, but the conjugate Diaconis--Ylvisaker priors for the parameters of log-linear models do not give rise to closed form credible regions, complicating posterior inference. In Chapter 4 we derive the optimal Gaussian approximation to the posterior for log-linear models with Diaconis--Ylvisaker priors, and provide convergence rate and finite-sample bounds for the Kullback-Leibler divergence between the exact posterior and the optimal Gaussian approximation. We demonstrate empirically in simulations and a real data application that the approximation is highly accurate, even in relatively small samples. The proposed approximation provides a computationally scalable and principled approach to regularized estimation and approximate Bayesian inference for log-linear models.

Another challenging and somewhat non-standard joint modeling problem is inference on tail dependence in stochastic processes. In applications where extreme dependence is of interest, data are almost always time-indexed. Existing methods for inference and modeling in this setting often cluster extreme events or choose window sizes with the goal of preserving temporal information. In Chapter 5, we propose an alternative paradigm for inference on tail dependence in stochastic processes with arbitrary temporal dependence structure in the extremes, based on the idea that the information on strength of tail dependence and the temporal structure in this dependence are both encoded in waiting times between exceedances of high thresholds. We construct a class of time-indexed stochastic processes with tail dependence obtained by endowing the support points in de Haan's spectral representation of max-stable processes with velocities and lifetimes. We extend Smith's model to these max-stable velocity processes and obtain the distribution of waiting times between extreme events at multiple locations. Motivated by this result, a new definition of tail dependence is proposed that is a function of the distribution of waiting times between threshold exceedances, and an inferential framework is constructed for estimating the strength of extremal dependence and quantifying uncertainty in this paradigm. The method is applied to climatological, financial, and electrophysiology data.

The remainder of this thesis focuses on posterior computation by Markov chain Monte Carlo. The Markov Chain Monte Carlo method is the dominant paradigm for posterior computation in Bayesian analysis. It has long been common to control computation time by making approximations to the Markov transition kernel. Comparatively little attention has been paid to convergence and estimation error in these approximating Markov Chains. In Chapter 6, we propose a framework for assessing when to use approximations in MCMC algorithms, and how much error in the transition kernel should be tolerated to obtain optimal estimation performance with respect to a specified loss function and computational budget. The results require only ergodicity of the exact kernel and control of the kernel approximation accuracy. The theoretical framework is applied to approximations based on random subsets of data, low-rank approximations of Gaussian processes, and a novel approximating Markov chain for discrete mixture models.

Data augmentation Gibbs samplers are arguably the most popular class of algorithm for approximately sampling from the posterior distribution for the parameters of generalized linear models. The truncated Normal and Polya-Gamma data augmentation samplers are standard examples for probit and logit links, respectively. Motivated by an important problem in quantitative advertising, in Chapter 7 we consider the application of these algorithms to modeling rare events. We show that when the sample size is large but the observed number of successes is small, these data augmentation samplers mix very slowly, with a spectral gap that converges to zero at a rate at least proportional to the reciprocal of the square root of the sample size up to a log factor. In simulation studies, moderate sample sizes result in high autocorrelations and small effective sample sizes. Similar empirical results are observed for related data augmentation samplers for multinomial logit and probit models. When applied to a real quantitative advertising dataset, the data augmentation samplers mix very poorly. Conversely, Hamiltonian Monte Carlo and a type of independence chain Metropolis algorithm show good mixing on the same dataset.

Learning from Geometry

Subspaces and manifolds are two powerful models for high dimensional signals. Subspaces model linear correlation and are a good fit to signals generated by physical systems, such as frontal images of human faces and multiple sources impinging at an antenna array. Manifolds model sources that are not linearly correlated, but where signals are determined by a small number of parameters. Examples are images of human faces under different poses or expressions, and handwritten digits with varying styles. However, there will always be some degree of model mismatch between the subspace or manifold model and the true statistics of the source. This dissertation exploits subspace and manifold models as prior information in various signal processing and machine learning tasks.

A near-low-rank Gaussian mixture model measures proximity to a union of linear or affine subspaces. This simple model can effectively capture the signal distribution when each class is near a subspace. This dissertation studies how the pairwise geometry between these subspaces affects classification performance. When model mismatch is vanishingly small, the probability of misclassification is determined by the product of the sines of the principal angles between subspaces. When the model mismatch is more significant, the probability of misclassification is determined by the sum of the squares of the sines of the principal angles. Reliability of classification is derived in terms of the distribution of signal energy across principal vectors. Larger principal angles lead to smaller classification error, motivating a linear transform that optimizes principal angles. This linear transformation, termed TRAIT, also preserves some specific features in each class, being complementary to a recently developed Low Rank Transform (LRT). Moreover, when the model mismatch is more significant, TRAIT shows superior performance compared to LRT.

The manifold model enforces a constraint on the freedom of data variation. Learning features that are robust to data variation is very important, especially when the size of the training set is small. A learning machine with large numbers of parameters, e.g., deep neural network, can well describe a very complicated data distribution. However, it is also more likely to be sensitive to small perturbations of the data, and to suffer from suffer from degraded performance when generalizing to unseen (test) data.

From the perspective of complexity of function classes, such a learning machine has a huge capacity (complexity), which tends to overfit. The manifold model provides us with a way of regularizing the learning machine, so as to reduce the generalization error, therefore mitigate overfiting. Two different overfiting-preventing approaches are proposed, one from the perspective of data variation, the other from capacity/complexity control. In the first approach, the learning machine is encouraged to make decisions that vary smoothly for data points in local neighborhoods on the manifold. In the second approach, a graph adjacency matrix is derived for the manifold, and the learned features are encouraged to be aligned with the principal components of this adjacency matrix. Experimental results on benchmark datasets are demonstrated, showing an obvious advantage of the proposed approaches when the training set is small.

Stochastic optimization makes it possible to track a slowly varying subspace underlying streaming data. By approximating local neighborhoods using affine subspaces, a slowly varying manifold can be efficiently tracked as well, even with corrupted and noisy data. The more the local neighborhoods, the better the approximation, but the higher the computational complexity. A multiscale approximation scheme is proposed, where the local approximating subspaces are organized in a tree structure. Splitting and merging of the tree nodes then allows efficient control of the number of neighbourhoods. Deviation (of each datum) from the learned model is estimated, yielding a series of statistics for anomaly detection. This framework extends the classical {\em changepoint detection} technique, which only works for one dimensional signals. Simulations and experiments highlight the robustness and efficacy of the proposed approach in detecting an abrupt change in an otherwise slowly varying low-dimensional manifold.

Seismic Response Analysis of a Full-Scale Base-Isolated Structure via Measurements and Modeling

The full-scale base-isolated structure studied in this dissertation is the only base-isolated building in South Island of New Zealand. It sustained hundreds of earthquake ground motions from September 2010 and well into 2012. Several large earthquake responses were recorded in December 2011 by NEES@UCLA and by GeoNet recording station nearby Christchurch Women's Hospital. The primary focus of this dissertation is to advance the state-of-the art of the methods to evaluate performance of seismic-isolated structures and the effects of soil-structure interaction by developing new data processing methodologies to overcome current limitations and by implementing advanced numerical modeling in OpenSees for direct analysis of soil-structure interaction.

This dissertation presents a novel method for recovering force-displacement relations within the isolators of building structures with unknown nonlinearities from sparse seismic-response measurements of floor accelerations. The method requires only direct matrix calculations (factorizations and multiplications); no iterative trial-and-error methods are required. The method requires a mass matrix, or at least an estimate of the floor masses. A stiffness matrix may be used, but is not necessary. Essentially, the method operates on a matrix of incomplete measurements of floor accelerations. In the special case of complete floor measurements of systems with linear dynamics, real modes, and equal floor masses, the principal components of this matrix are the modal responses. In the more general case of partial measurements and nonlinear dynamics, the method extracts a number of linearly-dependent components from Hankel matrices of measured horizontal response accelerations, assembles these components row-wise and extracts principal components from the singular value decomposition of this large matrix of linearly-dependent components. These principal components are then interpolated between floors in a way that minimizes the curvature energy of the interpolation. This interpolation step can make use of a reduced-order stiffness matrix, a backward difference matrix or a central difference matrix. The measured and interpolated floor acceleration components at all floors are then assembled and multiplied by a mass matrix. The recovered in-service force-displacement relations are then incorporated into the OpenSees soil structure interaction model.

Numerical simulations of soil-structure interaction involving non-uniform soil behavior are conducted following the development of the complete soil-structure interaction model of Christchurch Women's Hospital in OpenSees. In these 2D OpenSees models, the superstructure is modeled as two-dimensional frames in short span and long span respectively. The lead rubber bearings are modeled as elastomeric bearing (Bouc Wen) elements. The soil underlying the concrete raft foundation is modeled with linear elastic plane strain quadrilateral element. The non-uniformity of the soil profile is incorporated by extraction and interpolation of shear wave velocity profile from the Canterbury Geotechnical Database. The validity of the complete two-dimensional soil-structure interaction OpenSees model for the hospital is checked by comparing the results of peak floor responses and force-displacement relations within the isolation system achieved from OpenSees simulations to the recorded measurements. General explanations and implications, supported by displacement drifts, floor acceleration and displacement responses, force-displacement relations are described to address the effects of soil-structure interaction.

Math 412 - Topology with Applications

Highlights of Data Expedition: • Students explored daily observations of local climate data spanning the past 35 years. • Topological Data Analysis, or TDA for short, provides cutting-edge tools for studying the geometry of data in arbitrarily high dimensions. • Using TDA tools, students discovered intrinsic dynamical features of the data and learned how to quantify periodic phenomenon in a time-series. • Since nature invariably produces noisy data which rarely has exact periodicity, students also considered the theoretical basis of almost-periodicity and even invented and tested new mathematical definitions of almost-periodic functions. Summary The dataset we used for this data expedition comes from the Global Historical Climatology Network. “GHCN (Global Historical Climatology Network)-Daily is an integrated database of daily climate summaries from land surface stations across the globe.” Source: https://www.ncdc.noaa.gov/oa/climate/ghcn-daily/ We focused on the daily maximum and minimum temperatures from January 1, 1980 to April 1, 2015 collected from RDU International Airport. Through a guided series of exercises designed to be performed in Matlab, students explore these time-series, initially by direct visualization and basic statistical techniques. Then students are guided through a special sliding-window construction which transforms a time-series into a high-dimensional geometric curve. These high-dimensional curves can be visualized by projecting down to lower dimensions as in the figure below (Figure 1), however, our focus here was to use persistent homology to directly study the high-dimensional embedding. The shape of these curves has meaningful information but how one describes the “shape” of data depends on which scale the data is being considered. However, choosing the appropriate scale is rarely an obvious choice. Persistent homology overcomes this obstacle by allowing us to quantitatively study geometric features of the data across multiple-scales. Through this data expedition, students are introduced to numerically computing persistent homology using the rips collapse algorithm and interpreting the results. In the specific context of sliding-window constructions, 1-dimensional persistent homology can reveal the nature of periodic structure in the original data. I created a special technique to study how these high-dimensional sliding-window curves form loops in order to quantify the periodicity. Students are guided through this construction and learn how to visualize and interpret this information. Climate data is extremely complex (as anyone who has suffered from a bad weather prediction can attest) and numerous variables play a role in determining our daily weather and temperatures. This complexity coupled with imperfections of measuring devices results in very noisy data. This causes the annual seasonal periodicity to be far from exact. To this end, I have students explore existing theoretical notions of almost-periodicity and test it on the data. They find that some existing definitions are also inadequate in this context. Hence I challenged them to invent new mathematics by proposing and testing their own definition. These students rose to the challenge and suggested a number of creative definitions. While autocorrelation and spectral methods based on Fourier analysis are often used to explore periodicity, the construction here provides an alternative paradigm to quantify periodic structure in almost-periodic signals using tools from topological data analysis.

Distributed Feature Selection in Large n and Large p Regression Problems

Fitting statistical models is computationally challenging when the sample size or the dimension of the dataset is huge. An attractive approach for down-scaling the problem size is to first partition the dataset into subsets and then fit using distributed algorithms. The dataset can be partitioned either horizontally (in the sample space) or vertically (in the feature space), and the challenge arise in defining an algorithm with low communication, theoretical guarantees and excellent practical performance in general settings. For sample space partitioning, I propose a MEdian Selection Subset AGgregation Estimator ({\em message}) algorithm for solving these issues. The algorithm applies feature selection in parallel for each subset using regularized regression or Bayesian variable selection method, calculates the `median' feature inclusion index, estimates coefficients for the selected features in parallel for each subset, and then averages these estimates. The algorithm is simple, involves very minimal communication, scales efficiently in sample size, and has theoretical guarantees. I provide extensive experiments to show excellent performance in feature selection, estimation, prediction, and computation time relative to usual competitors.

While sample space partitioning is useful in handling datasets with large sample size, feature space partitioning is more effective when the data dimension is high. Existing methods for partitioning features, however, are either vulnerable to high correlations or inefficient in reducing the model dimension. In the thesis, I propose a new embarrassingly parallel framework named {\em DECO} for distributed variable selection and parameter estimation. In {\em DECO}, variables are first partitioned and allocated to m distributed workers. The decorrelated subset data within each worker are then fitted via any algorithm designed for high-dimensional problems. We show that by incorporating the decorrelation step, DECO can achieve consistent variable selection and parameter estimation on each subset with (almost) no assumptions. In addition, the convergence rate is nearly minimax optimal for both sparse and weakly sparse models and does NOT depend on the partition number m. Extensive numerical experiments are provided to illustrate the performance of the new framework.

For datasets with both large sample sizes and high dimensionality, I propose a new "divided-and-conquer" framework {\em DEME} (DECO-message) by leveraging both the {\em DECO} and the {\em message} algorithm. The new framework first partitions the dataset in the sample space into row cubes using {\em message} and then partition the feature space of the cubes using {\em DECO}. This procedure is equivalent to partitioning the original data matrix into multiple small blocks, each with a feasible size that can be stored and fitted in a computer in parallel. The results are then synthezied via the {\em DECO} and {\em message} algorithm in a reverse order to produce the final output. The whole framework is extremely scalable.