Search

In certain judgmental situations where a 'correct' decision is presumed to exist, optimal decision making requires evaluation of the decision-makers' capabilities and the selection of the appropriate aggregation rule. The major and so far unresolved difficulty is the former necessity. This article presents the optimal aggregation rule that simultaneously satisfies these two interdependent necessary requirements. In our setting, some record of the voters' past decisions is available, but the correct decisions are not known. We observe that any arbitrary evaluation of the decision-makers' capabilities as probabilities yields some optimal aggregation rule that, in turn, yields a maximum-likelihood estimation of decisional skills. Thus, a skill-evaluation equilibrium can be defined as an evaluation of decisional skills that yields itself as a maximum-likelihood estimation of decisional skills. We show that such equilibrium exists and offer a procedure for finding one. The obtained equilibrium is locally optimal and is shown empirically to generally be globally optimal in terms of the correctness of the resulting collective decisions. Interestingly, under minimally competent (almost symmetric) skill distributions that allow unskilled decision makers, the optimal rule considerably outperforms the common simple majority rule (SMR). Furthermore, a sufficient record of past decisions ensures that the collective probability of making a correct decision converges to 1, as opposed to accuracy of about 0.7 under SMR. Our proposed optimal voting procedure relaxes the fundamental (and sometimes unrealistic) assumptions in Condorcet's celebrated theorem and its extensions, such as sufficiently high decision-making quality, skill homogeneity or existence of a sufficiently large group of decision makers. [ABSTRACT FROM AUTHOR]

Given the judgments of multiple voters regarding some issue, it is generally assumed that the best way to arrive at some collective judgment is by following the majority. We consider here the now common case in which each voter expresses some (binary) judgment regarding each of a multiplicity of independent issues and assume that each voter has some fixed (unknown) probability of making a correct judgment for any given issue. We leverage the fact that multiple votes by each voter are known in order to demonstrate, both analytically and empirically, that a method based on maximum likelihood estimation is superior to the simple majority rule for arriving at true collective judgments. [ABSTRACT FROM AUTHOR]

This paper presents a procedure for automatic extraction and segmentation of a class-specific object (or region) by learning class-specific boundaries. We describe and evaluate the method with a specific focus on the detection of lesion regions in uterine cervix images. The watershed segmentation map of the input image is modeled using a Markov random field (MRF) in which watershed regions correspond to binary random variables indicating whether the region is part of the lesion tissue or not. The local pairwise factors on the arcs of the watershed map indicate whether the arc is part of the object boundary. The factors are based on supervised learning of a visual word distribution. The final lesion region segmentation is obtained using a loopy belief propagation applied to the watershed arc-level MRF. Experimental results on real data show state-of-the-art segmentation results on this very challenging task that, if necessary, can be interactively enhanced. [ABSTRACT FROM AUTHOR]

Mixture of Gaussians (MoG) model is a useful tool in statistical learning. In many learning processes that are based on mixture models, computational requirements are very demanding due to the large number of components involved in the model. We propose a novel algorithm for learning a simplified representation of a Gaussian mixture that is based on the Unscented Transform, which was introduced for filtering nonlinear dynamical systems. The superiority of the proposed method is validated on both simulation experiments and categorization of a real image database. The proposed categorization methodology is based on modeling each image using a Gaussian mixture model. A category model is obtained by learning a simplified mixture model from all the images in the category. [ABSTRACT FROM AUTHOR]

Low-density parity-check (LDPC) codes are usually decoded by running an iterative belief-propagation algorithm over the factor graph of the code. In the traditional message-passing schedule, in each iteration all the variable nodes, and subsequently all the factor nodes, pass new messages to their neighbors. Recently several studies show that serial scheduling, in which messages are generated using the latest available information, significantly improves the convergence speed in terms of number of iterations. It was observed experimentally in several studies that the serial schedule converges in exactly half the number of iterations compared to the standard parallel schedule. In this correspondence we provide a theoretical motivation for this observation by proving it for single-path graphs. [ABSTRACT FROM AUTHOR]

Conventionally, in each low-density parity-check (LDPC) decoding iteration all the variable nodes and subsequently all the check nodes send messages to their neighbors (flooding schedule). An alternative, more efficient, approach is to update the nodes' messages serially (serial schedule). A theoretical analysis of serial message passing decoding schedules is presented. In particular, the evolution of the computation tree under serial scheduling is analyzed. It shows that the tree grows twice as fast in comparison to the flooding schedule's one, indicating that the serial schedule propagates information twice as fast in the code's underlying graph. Furthermore, an asymptotic analysis of the serial schedule's convergence rate is done using the density evolution (DE) algorithm. Applied to various ensembles of LDPC codes, it shows that for long codes the serial schedule is expected to converge in half the number of iterations compared to the standard flooding schedule, when working near the ensemble's threshold. This observation is generally proved for the binary erasure channel (BEC) under some natural assumptions. Finally, an accompanying concentration theorem is proved. [ABSTRACT FROM AUTHOR]

An automated algorithm for tissue segmentation of noisy, low-contrast magnetic resonance (MR) images of the brain is presented. A mixture model composed of a large number of Gaussians is used to represent the brain image. Each tissue is represented by a large number of Gaussian components to capture the complex tissue spatial layout. The intensity of a tissue is considered a global feature and is incorporated into the model through tying of all the related Gaussian parameters. The expectation-maximization (EM) algorithm is utilized to learn the parameter-tied, constrained Gaussian mixture model. An elaborate initialization scheme is suggested to link the set of Gaussians per tissue type, such that each Gaussian in the set has similar intensity characteristics with minimal overlapping spatial supports. Segmentation of the brain image is achieved by the affiliation of each voxel to the component of the model that maximized the a posteriori probability. The presented algorithm is used to segment three-dimensional, T1-weighted, simulated and real MR images of the brain into three different tissues, under varying noise conditions. Results are compared with state-of-the-art algorithms in the literature. The algorithm does not use an atlas for initialization or parameter learning. Registration processes are therefore not required and the applicability of the framework can be extended to diseased brains and neonatal brains. [ABSTRACT FROM AUTHOR]

We describe an algorithm for context-based segmentation of visual data. New frames in an image sequence (video) are segmented based on the prior segmentation of earlier frames in the sequence. The segmentation is performed by adapting a probabilistic model learned on previous frames, according to the content of the new frame. We utilize the maximum a posteriori version of the EM algorithm to segment the new image. The Gaussian mixture distribution that is used to model the current frame is transformed into a conjugate-prior distribution for the parametric model describing the segmentation of the new frame. This semisupervised method improves the segmentation quality and consistency and enables a propagation of segments along the segmented images. The performance of the proposed approach is illustrated on both simulated and real image data. [ABSTRACT FROM AUTHOR]

In this paper, we combine discrete and continuous image models with information-theoretic-based criteria for unsupervised hierarchical image-set clustering. The continuous image modeling is based on mixture of Gaussian densities. The unsupervised image-set clustering is based on a generalized version of a recently introduced information-theoretic principle, the information bottleneck principle. Images are clustered such that the mutual information between the clusters and the image content is maximally preserved. Experimental results demonstrate the performance of the proposed framework for image clustering on a large image set. Information theoretic tools are used to evaluate cluster quality. Particular emphasis is placed on the application of the clustering for efficient image search and retrieval. [ABSTRACT FROM AUTHOR]

Abstract: In this work we address the problem of projective reconstruction from multiple views with missing data. Factorization based algorithms require point correspondences across all the views. In many applications this is an unrealistic assumption. Current methods that solve the problem of projective reconstruction with missing data require correspondence information across triplets of images. We propose a projective reconstruction method that yields a consistent camera set given the fundamental matrices between pairs of views without directly using the image correspondences. The algorithm is based on breaking the reconstruction problem into small steps. In each step, we eliminate as much uncertainty as possible. [Copyright &y& Elsevier]

In this paper, we describe a statistical video representation and modeling scheme. Video representation schemes are needed to segment a video stream into meaningful video-objects, useful for later indexing and retrieval applications. In the proposed methodology, unsupervised clustering via Gaussian mixture modeling extracts coherent space-time regions in feature space, and corresponding coherent segments (video-regions) in the video content. A key feature of the system is the analysis of video input as a single entity as opposed to a sequence of separate frames. Space and time are treated uniformly. The probabilistic space-time video representation scheme is extended to a piecewise GMM framework in which a succession of GMMs are extracted for the video sequence, instead of a single global model for the entire sequence. The piecewise GMM framework allows for the analysis of extended video sequences and the description of nonlinear, nonconvex motion patterns. The extracted space-time regions allow for the detection and recognition of video events. Results of segmenting video content into static versus dynamic video regions and video content editing are presented. Index Terms—Video representation; video segmentation; detection of events in video; Gaussian mixture model. [ABSTRACT FROM AUTHOR]

This paper presents a novel deep learning architecture for classifying structured objects in ultrafine-grained datasets, where classes may not be clearly distinguishable by their appearance but rather by their context. We model sequences of images as linear-chain CRFs, and jointly learn the parameters from both local-visual features and neighboring class information. The visual features are learned by convolutional layers, whereas class-structure information is reparametrized by factorizing the CRF pairwise potential matrix. This forms a context-based semantic similarity space, learned alongside the visual similarities, and dramatically increases the learning capacity of contextual information. This new parametrization, however, forms a highly nonlinear objective function which is challenging to optimize. To overcome this, we develop a novel surrogate likelihood which allows for a local likelihood approximation of the original CRF with integrated batch-normalization. This model overcomes the difficulties of existing CRF methods to learn the contextual relationships thoroughly when there is a large number of classes and the data is sparse. The performance of the proposed method is illustrated on a huge dataset that contains images of retail-store product displays, and shows significantly improved results compared to linear CRF parametrization, unnormalized likelihood optimization, and RNN modeling. We also show improved results on a standard OCR dataset. • A method for training a factorized CRF that enables combining batch-normalization. • Combining deep class embedding into a CRF formulation. • Handling datasets with a huge number of similar classes. [ABSTRACT FROM AUTHOR]

The significant heterogeneity of Wilms' tumors between different patients is thought to arise from genetic and epigenetic distortions that occur during various stages of fetal kidney development in a way that is poorly understood. To address this, we characterized the heterogeneity of alternative mRNA splicing in Wilms' tumors using a publicly available RNAseq dataset of high-risk Wilms' tumors and normal kidney samples. Through Pareto task inference and cell deconvolution, we found that the tumors and normal kidney samples are organized according to progressive stages of kidney development within a triangle-shaped region in latent space, whose vertices, or "archetypes", resemble the cap mesenchyme, the nephrogenic stroma, and epithelial tubular structures of the fetal kidney. We identified a set of genes that are alternatively spliced between tumors located in different regions of latent space and found that many of these genes are associated with the epithelial-to-mesenchymal transition (EMT) and muscle development. Using motif enrichment analysis, we identified putative splicing regulators, some of which are associated with kidney development. Our findings provide new insights into the etiology of Wilms' tumors and suggest that specific splicing mechanisms in early stages of development may contribute to tumor development in different patients. [ABSTRACT FROM AUTHOR]

This paper examines data fusion methods for multi-view data classification. We present a decision concept that explicitly takes into account the input multi-view structure, where for each case there is a different subset of relevant views. This data fusion concept, which we dub Mixture of Views, is implemented by a special purpose neural network architecture. The single view decisions are combined by a data-driven decision, into a global decision according to the relevance of each view in a given case. The method was applied to two challenging computer-aided diagnosis (CADx) tasks: the task of classifying breast microcalcifications as benign or malignant based on craniocaudal (CC) and mediolateral oblique (MLO) mammography views and segmenting Multiple Sclerosis (MS) white matter lesions. The experimental results show that our method outperforms previously suggested fusion methods. [ABSTRACT FROM AUTHOR]

A well-known obstacle in the successful implementation of deep learning-based systems to real-world problems is the performance degradation that occurs when applying a network that was trained on data collected in one domain, to data from a different domain. In this study, we focus on the Supervised Domain Adaptation (SDA) setup where we assume the availability of a small amount of labeled data from the target domain. Our approach is based on transferring the gradient history of the pre-training phase to the fine-tuning phase in addition to the parameter set to improve the generalization achieved during pre-training while fine-tuning the model. We present two schemes to transfer the gradient's information. Mixed Minibatch Transfer Learning (MMTL) is based on using examples from both the source and target domains and Optimizer-Continuation Transfer Learning (OCTL) preserves the gradient history when shifting from training to fine-tuning. This approach is also applicable to the more general setup of transfer learning across different tasks. We show that our methods outperform the state-of-the-art at different levels of data scarcity from the target domain, on multiple datasets and tasks involving both scenery and medical images. • We present supervised domain adaptation methods that transfer both the parameter set and its derivative. • In MMTL the adaptation procedure is based on using examples from both the source and target domains. • DMMTL is a variant of MMTL where we only take source examples which are close to the target domain. • The OCTL algorithm preserves the gradient history when shifting from training to fine-tuning. [ABSTRACT FROM AUTHOR]

Wilms' tumors are pediatric malignancies that are thought to arise from faulty kidney development. They contain a wide range of poorly differentiated cell states resembling various distorted developmental stages of the fetal kidney, and as a result, differ between patients in a continuous manner that is not well understood. Here, we used three computational approaches to characterize this continuous heterogeneity in high-risk blastemal-type Wilms' tumors. Using Pareto task inference, we show that the tumors form a triangle-shaped continuum in latent space that is bounded by three tumor archetypes with "stromal", "blastemal", and "epithelial" characteristics, which resemble the un-induced mesenchyme, the cap mesenchyme, and early epithelial structures of the fetal kidney. By fitting a generative probabilistic "grade of membership" model, we show that each tumor can be represented as a unique mixture of three hidden "topics" with blastemal, stromal, and epithelial characteristics. Likewise, cellular deconvolution allows us to represent each tumor in the continuum as a unique combination of fetal kidney-like cell states. These results highlight the relationship between Wilms' tumors and kidney development, and we anticipate that they will pave the way for more quantitative strategies for tumor stratification and classification. [ABSTRACT FROM AUTHOR]

We present the Atlas of Classifiers (AoC)-a conceptually novel framework for brain MRI segmentation. The AoC is a spatial map of voxel-wise multinomial logistic regression (LR) functions learned from the labeled data. Upon convergence, the resulting fixed LR weights, a few for each voxel, represent the training dataset. It can, therefore, be considered as a light-weight learning machine, which despite its low capacity does not underfit the problem. The AoC construction is independent of the actual intensities of the test images, providing the flexibility to train it on the available labeled data and use it for the segmentation of images from different datasets and modalities. In this sense, it does not overfit the training data, as well. The proposed method has been applied to numerous publicly available datasets for the segmentation of brain MRI tissues and is shown to be robust to noise and outreach commonly used methods. Promising results were also obtained for multi-modal, cross-modality MRI segmentation. Finally, we show how AoC trained on brain MRIs of healthy subjects can be exploited for lesion segmentation of multiple sclerosis patients. [ABSTRACT FROM AUTHOR]

A classification model is calibrated if its predicted probabilities of outcomes reflect their accuracy. Calibrating neural networks is critical in medical analysis applications where clinical decisions rely upon the predicted probabilities. Most calibration procedures, such as temperature scaling, operate as a post processing step by using holdout validation data. In practice, it is difficult to collect medical image data with correct labels due to the complexity of the medical data and the considerable variability across experts. This study presents a network calibration procedure that is robust to label noise. We draw on the fact that the confusion matrix of the noisy labels can be expressed as the matrix product between the confusion matrix of the clean labels and the label noises. The method is based on estimating the noise level as part of a noise-robust training method. The noise level is then used to estimate the network accuracy required by the calibration procedure. We show that despite the unreliable labels, we can still achieve calibration results that are on a par with the results of a calibration procedure using data with reliable labels.

In this study we present an efficient image categorization system for medical image databases utilizing a local patch representation based on both content and location. The system discriminates between healthy and pathological cases and indicates the subregion in the image that is automatically found to be most relevant for the decision. We show an application to pathology-level categorization of chest x-ray data, the most popular examination in radiology. Experimental results are provided on chest radiographs taken from routine hospital examinations.

In this study we treat scribbling motion as a compositional system in which a limited set of elementary strokes are capable of concatenating amongst themselves in an endless number of combinations, thus producing an unlimited repertoire of complex constructs. We broke the continuous scribblings into small units and then calculated the Markovian transition matrix between the trajectory clusters. The Markov states are grouped in a way that minimizes the loss of mutual information between adjacent strokes. The grouping algorithm is based on a novel markov-state bi-clustering algorithm derived from the Information-Bottleneck principle. This approach hierarchically decomposes scribblings into increasingly finer elements. We illustrate the usefulness of this approach by applying it to human scribbling.

We present an automated algorithm for tissue segmentation of noisy, low contrast magnetic resonance (MR) images of the brain. We use a mixture model composed of a large number of Gaussians, with each brain tissue represented by a large number of the Gaussian components in order to capture the complex tissue spatial layout. The intensity of a tissue is considered a global feature and is incorporated into the model through parameter tying of all the related Gaussians. The EM algorithm is utilized to learn the parameter-tied Gaussian mixture model. A new initialization method is applied to guarantee the convergence of the EM algorithm to the global maximum likelihood. Segmentation of the brain image is achieved by the affiliation of each voxel to a selected tissue class. The presented algorithm is used to segment 3D, T1-weighted, simulated and real MR images of the brain into three different tissues, under varying noise conditions. Quantitative results are presented and compared with state-of-the-art results reported in the literature.