A Clearer Picture of Total Variation Blind Deconvolution

Blind deconvolution is the problem of recovering a sharp image and a blur kernel from a noisy blurry image. Recently, there has been a significant effort on understanding the basic mechanisms to solve blind deconvolution. While this effort resulted in the deployment of effective algorithms, the theoretical findings generated contrasting views on why these approaches worked. On the one hand, one could observe experimentally that alternating energy minimization algorithms converge to the desired solution. On the other hand, it has been shown that such alternating minimization algorithms should fail to converge and one should instead use a so-called Variational Bayes approach. To clarify this conundrum, recent work showed that a good image and blur prior is instead what makes a blind deconvolution algorithm work. Unfortunately, this analysis did not apply to algorithms based on total variation regularization. In this manuscript, we provide both analysis and experiments to get a clearer picture of blind deconvolution. Our analysis reveals the very reason why an algorithm based on total variation works. We also introduce an implementation of this algorithm and show that, in spite of its extreme simplicity, it is very robust and achieves a performance comparable to the top performing algorithms.

(Table 1) Abundance of tunicate spicules along the Great Barrier Reef-Queensland Plateau transect

The abundance patterns of tunicate spicules are documented for the Pliocene-Pleistocene sediments at seven sites along the Great Barrier Reef-Queensland Plateau transect. The spatial distribution pattern indicates that tunicate spicules were limited to waters shallower than 900 m. The occurrences of tunicate spicules at Sites 822 and 823 that are deeper than 900 m are ascribed to downslope transport, and their distribution patterns can be used to monitor downslope transport processes. The first common occurrence of tunicate spicules at Sites 822 and 823 around 1.6 Ma may suggest the initiation of the central Great Barrier Reef at this time. The morphology of tunicate spicules varies greatly and appears to be gradational among different forms. Older tunicate assemblages are less diverse than those in younger sediments, presumably because of diagenesis. Tunicate spicules do not appear to be a promising biostratigraphic tool for the Pliocene-Pleistocene.

The Inefficiency of Self-Sufficiency: Uzbekistan's Response to CASA-1000

Thesis (Master's)--University of Washington, 2016-06

Preserved obstacle avoidance during reaching in patients with left visual neglect

We asked 12 patients with left visual neglect to bisect the gap between two cylinders or to reach rapidly between them to a more distal target zone. Both tasks demanded a motor response but these responses were quite different in nature. The bisection response was a communicative act whereby the patient indicated the perceived midpoint. The reaching task carried no imperative to bisect the gap, only to maintain a safe distance from either cylinder while steering to the target zone. Optimal performance on either task could only be achieved by reference to the location of both cylinders. Our analysis focused upon the relative influence of the left and right cylinders on the lateral location of the response. In the bisection task, all neglect patients showed qualitatively the same asymmetry, with the left cylinder exerting less influence than the right. In the reaching task, the neglect group behaved like normal subjects, being influenced approximately equally by the two cylinders. This was true for all bar two of the patients, who showed clear neglect in both tasks. We conclude that the visuomotor processing underlying obstacle avoidance during reaching is preserved in most patients with left visual neglect. (C) 2004 Elsevier Ltd. All rights reserved.

Identifying the molecular phenotype of renal progenitor cells

Although many of the molecular interactions in kidney development are now well understood, the molecules involved in the specification of the metanephric mesenchyme from surrounding intermediate mesoderm and, hence, the formation of the renal progenitor population are poorly characterized. In this study, cDNA microarrays were used to identify genes enriched in the murine embryonic day 10.5 (E10.5) uninduced metanephric mesenchyme, the renal progenitor population, in comparison with more rostral derivatives of the intermediate mesoderm. Microarray data were analyzed using R statistical software to determine accurately genes differentially expressed between these populations. Microarray outliers were biologically verified, and the spatial expression pattern of these genes at E10.5 and subsequent stages of early kidney development was determined by RNA in situ hybridization. This approach identified 21 genes preferentially expressed by the E10.5 metanephric mesenchyme, including Ewing sarcoma homolog, 14-3-3 theta, retinoic acid receptor-alpha, stearoyl-CoA desaturase 2, CD24, and cadherin-11, that may be important in formation of renal progenitor cells. Cell surface proteins such as CD24 and cadherin-11 that were strongly and specifically expressed in the uninduced metanephric mesenchyme and mark the renal progenitor population may prove useful in the purification of renal progenitor cells by FACS. These findings may assist in the isolation and characterization of potential renal stem cells for use in cellular therapies for kidney disease.

H-ras, K-ras, and inner plasma membrane raft proteins operate in nanoclusters with differential dependence on the actin cytoskeleton

Plasma membrane compartmentalization imposes lateral segregation on membrane proteins that is important for regulating signal transduction. We use computational modeling of immunogold spatial point patterns on intact plasma membrane sheets to test different models of inner plasma membrane organization. We find compartmentalization at the nanoscale level but show that a classical raft model of preexisting stable domains into which lipid raft proteins partition is incompatible with the spatial point patterns generated by the immunogold labeling of a palmitoylated raft marker protein. Rather, approximate to 30% of the raft protein exists in cholesterol-dependent nanoclusters, with approximate to 70% distributed as monomers. The cluster/monomer ratio (number of proteins in clusters/number of proteins outside clusters) is independent of expression level. H-rasG12V and K-rasG12V proteins also operate in nanoclusters with fixed cluster/monomer ratios that are independent of expression level. Detailed calibration of the immunogold imaging protocol suggests that radii of raft and RasG12V protein nanoclusters may be as small as 11 and 6 nm, respectively, and shows that the nanoclusters contain small numbers (6.0-7.7) of proteins. Raft nanoclusters do not form if the actin cytoskeleton is disassembled. The formation of K-rasG12V but not H-rasG12V nanoclusters also is actin-dependent. K-rasG12V but not H-rasG12V signaling is abrogated by actin cytoskeleton disassembly, which shows that nanoclustering is critical for Ras function. These findings argue against stable preexisting domains on the inner plasma membrane in favor of dynamic actively regulated nanoclusters similar to those proposed for the outer plasma membrane. RasG12V nanoclusters may facilitate the assembly of essential signal transduction complexes.

Global adaptation of spring bread and durum wheat lines near-isogenic for major reduced height genes

The effect of major dwarfing genes, Rht-B1 and Rht-D1, in bread (Triticum aestivum L.) and durum (Triticum turgidum L. var. durum) wheats varies with environment. Six reduced-height near-isogenic spring wheat lines, included in the International Adaptation Trial (IAT), were grown in 81 trials around the world. Of the 56 IAT trials yielding > 3 Mg ha(-1), the mean yield of semidwarfs was significantly greater than tails in 54% of trials; in the 27 trials yielding < 3 Mg ha-1, semidwarfs were superior in only 24%. Sixteen pairs of semidwarf-tall near-isolines were grown in six managed drought environment trials (DETs) in northwestern Mexico. In these trials, semidwarfs outyielded talls in all but the most droughted environment (2.5 Mg ha(-1)). The effect of the height alleles varied with genetic background and environment. For both yield and height, variance components for allele and environment by allele interaction were larger than those for genetic background and genetic background by environment. Pattern analysis showed that tall and semidwarf lines had similar adaptation to stressed environments (< 2.8 Mg ha(-1), low rainfall), while semidwarfs yielded more in less stressed environments (> 4.3 Mg ha(-1), high rainfall). The best adapted near-isogenic pair had a Kauz background, where the tall was only 16% taller than the dwarf. In the Kauz-derived pair, the semidwarf outyielded the tall in only 13% of trials with no differences in low yielding trials. This supports the idea that '' short talls '' may be useful in marginal environments (yield < 3 Mg ha(-1)).

Using generative models for handwritten digit recognition

We describe a method of recognizing handwritten digits by fitting generative models that are built from deformable B-splines with Gaussian ``ink generators'' spaced along the length of the spline. The splines are adjusted using a novel elastic matching procedure based on the Expectation Maximization (EM) algorithm that maximizes the likelihood of the model generating the data. This approach has many advantages. (1) After identifying the model most likely to have generated the data, the system not only produces a classification of the digit but also a rich description of the instantiation parameters which can yield information such as the writing style. (2) During the process of explaining the image, generative models can perform recognition driven segmentation. (3) The method involves a relatively small number of parameters and hence training is relatively easy and fast. (4) Unlike many other recognition schemes it does not rely on some form of pre-normalization of input images, but can handle arbitrary scalings, translations and a limited degree of image rotation. We have demonstrated our method of fitting models to images does not get trapped in poor local minima. The main disadvantage of the method is it requires much more computation than more standard OCR techniques.

A hierarchical latent variable model for data visualization

Visualization has proven to be a powerful and widely-applicable tool the analysis and interpretation of data. Most visualization algorithms aim to find a projection from the data space down to a two-dimensional visualization space. However, for complex data sets living in a high-dimensional space it is unlikely that a single two-dimensional projection can reveal all of the interesting structure. We therefore introduce a hierarchical visualization algorithm which allows the complete data set to be visualized at the top level, with clusters and sub-clusters of data points visualized at deeper levels. The algorithm is based on a hierarchical mixture of latent variable models, whose parameters are estimated using the expectation-maximization algorithm. We demonstrate the principle of the approach first on a toy data set, and then apply the algorithm to the visualization of a synthetic data set in 12 dimensions obtained from a simulation of multi-phase flows in oil pipelines and to data in 36 dimensions derived from satellite images.

Hierarchical GTM: constructing localized non-linear projection manifolds in a principled way

It has been argued that a single two-dimensional visualization plot may not be sufficient to capture all of the interesting aspects of complex data sets, and therefore a hierarchical visualization system is desirable. In this paper we extend an existing locally linear hierarchical visualization system PhiVis ¸iteBishop98a in several directions: bf(1) We allow for em non-linear projection manifolds. The basic building block is the Generative Topographic Mapping (GTM). bf(2) We introduce a general formulation of hierarchical probabilistic models consisting of local probabilistic models organized in a hierarchical tree. General training equations are derived, regardless of the position of the model in the tree. bf(3) Using tools from differential geometry we derive expressions for local directional curvatures of the projection manifold. Like PhiVis, our system is statistically principled and is built interactively in a top-down fashion using the EM algorithm. It enables the user to interactively highlight those data in the ancestor visualization plots which are captured by a child model. We also incorporate into our system a hierarchical, locally selective representation of magnification factors and directional curvatures of the projection manifolds. Such information is important for further refinement of the hierarchical visualization plot, as well as for controlling the amount of regularization imposed on the local models. We demonstrate the principle of the approach on a toy data set and apply our system to two more complex 12- and 18-dimensional data sets.

Multiple system atrophy (MSA): Topographic distribution of the alpha-synuclein-associated pathological changes

To study the topographic distribution of the pathology in multiple system atrophy (MSA). Pattern analysis was carried out using a-synuclein immunohistochemistry in 10 MSA cases. The glial cytoplasmic inclusions (GCI) were distributed randomly or in large clusters. The neuronal inclusions (NI) and abnormal neurons were distributed in regular clusters. Clusters of the NI and abnormal neurons were spatially correlated whereas the GCI were not spatially correlated with either the NI or the abnormal neurons. The data suggest that the GCI represent the primary change in MSA and the neuronal pathology develops secondary to the glial pathology.

Bayesian classification with Gaussian processes

We consider the problem of assigning an input vector to one of m classes by predicting P(c|x) for c=1,...,m. For a two-class problem, the probability of class one given x is estimated by s(y(x)), where s(y)=1/(1+e-y). A Gaussian process prior is placed on y(x), and is combined with the training data to obtain predictions for new x points. We provide a Bayesian treatment, integrating over uncertainty in y and in the parameters that control the Gaussian process prior the necessary integration over y is carried out using Laplace's approximation. The method is generalized to multiclass problems (m>2) using the softmax function. We demonstrate the effectiveness of the method on a number of datasets.

A statistical method for the identification and aggregation of regional linguistic variation

This paper introduces a method for the analysis of regional linguistic variation. The method identifies individual and common patterns of spatial clustering in a set of linguistic variables measured over a set of locations based on a combination of three statistical techniques: spatial autocorrelation, factor analysis, and cluster analysis. To demonstrate how to apply this method, it is used to analyze regional variation in the values of 40 continuously measured, high-frequency lexical alternation variables in a 26-million-word corpus of letters to the editor representing 206 cities from across the United States.