Local Subsets Research Articles

Ethnicity and Clinical Outcomes in Head and Neck Cancer: an Analysis of the SEER Database

We evaluated the differences within various ethnic groups diagnosed with head and neck cancer in the USA with a specific focus on the clinical outcomes of patients of Hispanic ethnicity. We used the Surveillance, Epidemiology, and End Results (SEER) database to examine the clinical outcomes of patients with head and neck cancer by ethnicity, region of origin, place of birth, treatment modality, primary location, age, gender, and SEER tumor stage. For all patients, African American race conferred the worst prognosis for head and neck cancer (1.41; CI 1.361–1.454), while non-US born Hispanics had the best prognosis (0.74; 0.684–0.804). US born Hispanics (0.86; CI 0.823–0.892) had a significantly better prognosis than Whites (ref.). The Hispanic population appears to have a better prognosis compared to their Caucasian peers while both do better than Blacks across all stages and in local and regional disease subsets. Furthermore, non-US born Hispanics had a better prognosis than US born Hispanics. The cause of this disparity is unclear and warrants further investigation.

Local feature selection using Gaussian process regression

Most feature selection methods determine a global subset of features, where all data instances are projected in order to improve classification accuracy. An attractive alternative solution is to adaptively find a local subset of features for each data instance, such that, the classification of each instance is performed according to its own selective subspace. This paper presents a novel application of Gaussian Processes GPs that improves classification performance by learning a set of functions that quantify the discriminative power of each feature. Specifically, GP regressions are used to build for each available feature a function that estimates its discriminative properties over all its input space. Afterwards, by locally joining these regressions it is possible to obtain a discriminative subspace for any position of the input space. New instances are then classified by using a K-NN classifier that operates in the local subspaces. Experimental results show that by using local discriminative subspaces, we are able to reach higher levels of classification accuracy than alternative state-of-the-art feature selection approaches.

The effect of spatially marginal localities in modelling species niches and distributions

AbstractAimWe introduce and evaluate the potential effect of spatially marginal localities (specifically those protruding into unsuitable regions), in overestimating species niches and distributions when using ecological niche models (ENMs).LocationNorth‐western South America.MethodsWe built anENMfor the Caribbean spiny pocket mouse (Heteromys anomalus) usingMaxEntand climatic variables. This species typically inhabits extensive tropical forests but can also range into drier habitats through patches of mesic vegetation. We ranked occurrence records according to the suitability value they received, and retrieved habitat information from collectors' field notes and the literature to determine whether those receiving lower values correspond to spatially marginal localities protruding into unsuitable regions. We then built a model excluding a subset of such localities and compared its geographic and environmental prediction with that of the original model.ResultsModels differed substantially in their estimates of suitability. The original model resulted in an overly extensive prediction, considering as suitable hot and dry regions dominated by xerophytic vegetation. Records receiving the lowest suitability values in this model corresponded mainly to captures in patches of mesic forest surrounded by thorn scrub or savannas. The model calibrated without such records restricted suitability mostly to regions characterized by the typical habitat of the species.Main conclusionsWhen it is not possible to use variables that are more proximal or have finer resolutions, we recommend building complementary models that, together, can provide a more realistic estimate of the species' niche and corresponding geographic distribution. Jointly interpreting these models, researchers may better differentiate between areas harbouring typical habitat and those where the species can be found only if locally favourable conditions exist. Such a distinction is of relevance for a wide range of applications relying on inferences obtained fromENMs.

Electoral Rules or Electoral Leverage? Explaining Muslim Representation in England

Immigration has fundamentally altered the ethnic and religious makeup of most advanced democracies, but substantial variation is observed in the political representation of immigrant-origin minority groups across countries and cities. Though existing research has highlighted the role of electoral institutions in explaining minority representation, it is often difficult to isolate their effects across contexts. Focusing on Muslims in England and employing a new data set containing over 42,000 candidate-level observations, this article explains Muslim candidate election and selection. To do this, the author makes use of a rule change whereby a subset of localities switched from the use of multimember elections to the use of single-member elections. She finds that these electoral rules have no significant effect on the share of Muslims that gets elected but that they do influence the selection process: in a given election, Muslims are half as likely to be selected when only one seat is up for election as compared with when three seats are in play. Yet parties balance the slate across consecutive single-member elections, leading to similar results across systems. Further, the more undesirable the seat, the more likely it is to have a Muslim on the ticket, but this effect holds only in single-member elections, and it reverses as Muslims gain electoral leverage. Overall electoral leverage proves crucial: the effect of institutions and the potential for institution-based discrimination are conditional on the size and concentration of the local Muslim population and the votes it can deliver at both the election and the selection stages.

Bandit-based local feature subset selection

In this work we propose a method for local feature subset selection, where we simultaneously partition the sample space into localities and select features for them. The partitions and the corresponding local features are represented using a novel notion of feature tree. The problem of finding an appropriate feature tree is then formulated as a reinforcement learning problem. A value-based Monte Carlo tree search with the corresponding credit assignment policy is devised to learn near-optimal feature trees. Furthermore, the Monte Carlo tree search is enhanced in a way to be applicable for large numbers of actions (i.e., features). This objective is achieved by taking into account a bandit-based explorative policy while having a soft exploitive estimation policy. The results for synthetic datasets show that when local features are present in data, the proposed method can outperform other feature selection methods. Furthermore, the results for microarray classification show that the method can obtain results comparable to the state of the art, using a simple KNN classifier.

Mining histopathological images via composite hashing and online learning.

With a continuous growing amount of annotated histopathological images, large-scale and data-driven methods potentially provide the promise of bridging the semantic gap between these images and their diagnoses. The purpose of this paper is to increase the scale at which automated systems can entail scalable analysis of histopathological images in massive databases. Specifically, we propose a principled framework to unify hashing-based image retrieval and supervised learning. Concretely, composite hashing is designed to simultaneously fuse and compress multiple high-dimensional image features into tens of binary hash bits, enabling scalable image retrieval with a very low computational cost. Upon a local data subset that retains the retrieved images, supervised learning methods are applied on-the-fly to model image structures for accurate classification. Our framework is validated thoroughly on 1120 lung microscopic tissue images by differentiating adenocarcinoma and squamous carcinoma. The average accuracy as 87.5% with only 17ms running time, which compares favorably with other commonly used methods.

An artificial immune network for multiobjective optimization problems

Multiobjective optimization is an important problem of great complexity and evolutionary algorithms have been established as a dominant approach in the field. This article suggests a method for approximating the Pareto front of a given function based on artificial immune networks. The proposed method uses cloning and mutation on a population of antibodies to create local subsets of the Pareto front. Elements of these local fronts are combined, in a way that maximizes diversity, to form the complete Pareto front of the function. The method is tested on a number of well-known benchmark problems, as well as an engineering problem. Its performance is compared against state-of-the-art algorithms, yielding promising results.

Reducing subject tree browsing complexity

Many large digital collections are currently organized by subject; although useful, these information organization structures are large and complex and thus difficult to browse. Current online tools and visualization prototypes show small, localized subsets and do not provide the ability to explore the predominant patterns of the overall subject structure. This study describes subject tree modifications that facilitate browsing for documents by capitalizing on the highly uneven distribution of real‐world collections. The approach is demonstrated on two large collections organized by the Library of Congress Subject Headings (LCSH) and Medical Subject Headings (MeSH). Results show that the LCSH subject tree can be reduced to 49% of its initial complexity while maintaining access to 83% of the collection, and the MeSH tree can be reduced to 45% of its initial complexity while maintaining access to 97% of the collection. A simple solution to negate the loss of access is discussed. The visual impact is demonstrated by using traditional outline views and a slider control allowing searchers to change the subject structure dynamically according to their needs. This study has implications for the development of information organization theory and human–information interaction techniques for subject trees.

Semi-supervised projected model-based clustering

We present an adaptation of model-based clustering for partially labeled data, that is capable of finding hidden cluster labels. All the originally known and discoverable clusters are represented using localized feature subset selections (subspaces), obtaining clusters unable to be discovered by global feature subset selection. The semi-supervised projected model-based clustering algorithm (SeSProC) also includes a novel model selection approach, using a greedy forward search to estimate the final number of clusters. The quality of SeSProC is assessed using synthetic data, demonstrating its effectiveness, under different data conditions, not only at classifying instances with known labels, but also at discovering completely hidden clusters in different subspaces. Besides, SeSProC also outperforms three related baseline algorithms in most scenarios using synthetic and real data sets.

Quantitative Measurements of Anaerobic Digestion Process Parameters Using near Infrared Spectroscopy and Local Calibration Models

The performance of local calibration models for quantitative measurements of ammonium and acetate on samples from an anaerobic digestion process was examined. The local calibration methods used were locally weighted regression (LWR) and multi-layer partial least squares (ML-PLS) regression. The results of these two methods were compared to each other and to the results from the global partial least squares (PLS) model regression as well. For ammonium, both the local methods performed excellently in comparison with global PLS models. However, the results from the 150 LWR models regressed for ammonium also showed that the accuracy can be highly dependent on the different combination alternatives for model parameter settings and pre-processing alternatives. For this reason, a number of distance measures were evaluated as local subset selection methods in ML-PLS. The benefits of an optimised layer structure and the iterative approach in ML-PLS were also evaluated for ammonium. This showed that some benefits can be obtained by optimising the layer structure, at least in the sense that the number of layers can be reduced, and that there can be a significant advantage in using an iterative approach in the selection of the local subset of calibration data. The local calibration methods were also evaluated for acetate but, in this case, the benefits compared to global PLS calibration models were fairly insignificant with ML-PLS and none at all with LWR.

Plasma Levels of Soluble HLA-E and HLA-F at Diagnosis May Predict Overall Survival of Neuroblastoma Patients

The purpose of this study was to identify the plasma/serum biomarkers that are able to predict overall survival (OS) of neuroblastoma (NB) patients. Concentration of soluble (s) biomarkers was evaluated in plasma (sHLA-E, sHLA-F, chromogranin, and B7H3) or serum (calprotectin) samples from NB patients or healthy children. The levels of biomarkers that were significantly higher in NB patients were then analyzed considering localized or metastatic subsets. Finally, biomarkers that were significantly different in these two subsets were correlated with patient's outcome. With the exception of B7H3, levels of all molecules were significantly higher in NB patients than those in controls. However, only chromogranin, sHLA-E, and sHLA-F levels were different between patients with metastatic and localized tumors. sHLA-E and -F levels correlated with each other but not chromogranin. Chromogranin levels correlated with different event-free survival (EFS), whereas sHLA-E and -F levels also correlated with different OS. Association with OS was also detected considering only patients with metastatic disease. In conclusion, low levels of sHLA-E and -F significantly associated with worse EFS/OS in the whole cohort of NB patients and in patients with metastatic NB. Thus, these molecules deserve to be tested in prospective studies to evaluate their predictive power for high-risk NB patients.

Exact versus estimated pruning of subject hierarchies

AbstractMany large digital collections are organized by subject; these useful information organization structures are large and complex, thus difficult to browse. Current online tools and visualization prototypes show small localized subsets and do not provide the ability to explore the predominant patterns of the overall subject structure. This study builds on existing work concerning automatic subject hierarchy modification techniques that aim to facilitate browsing for documents by capitalizing on the highly uneven distribution of real‐world collections. Specifically, previous work used an estimation of the number of accessible documents offered by each subject term, while the current study uses the exact number of accessible documents. The impact is demonstrated on a large collection organized using Medical Subject Headings (MeSH). Results show that, although computationally more demanding, pruning the MeSH hierarchy based on the exact access produces a different subject hierarchy under some conditions. The visual impact is demonstrated using traditional outline views. This study has implications for the development of information organization theory and human‐information interaction techniques for subject hierarchies.

Oil windfalls in Brazil and their long-run social impacts

In 1997, Brazil changed the regulatory framework of its oil and gas sector. After this change, a restrict subset of localities started to receive large amounts of oil rents. The goal of this paper is to evaluate whether such rents distributed under this law contributed to improving some social indicators relative to the national average in the eligible municipalities. We used a “difference-in-differences” measurement to compare changes in selected social indicators within affected Municipalities in the last two decades, taking the unaffected districts as control group. The Municipal data on social indicators were collected from three National Censuses, conducted in 1991, 2000 and 2010. Results show that royalties had a positive and statistically significant impact on household’s access to electric wiring, piped water and waste collection, as well as in the decrease of illiteracy rate. This means the eligible districts were able to improve some of their social indicators in the long-run.

A New Method Applied to the Multi-Parametric Nonlinear Compensations of Computer Numeric Control Bender

Complex Moving Least Squares and its key attributes as basis functions and support region radius were discussed and studied in this paper. Owing to the weak performance of traditional Moving Least Squares when applied to asymmetrical sampling points set, an improved Moving Least Squares named dynamic radius Moving Least Squares was designed, whose support regions radius at fitting point dynamically adjust as the density of local point subset. At last, it is applied to multi-parametric nonlinear compensation of computer numeric control bender through comparative fitting experiments and simulations on couples of data compared to traditional Moving Least Squares. The results of experiments and simulations indicate that the new algorithm makes a great progress on computational complexity, fitting smoothness and fitting accuracy, especially the accuracy in dense points region compared to traditional Moving Least Squares.

ACOSampling: An ant colony optimization-based undersampling method for classifying imbalanced DNA microarray data

In DNA microarray data, class imbalance problem occurs frequently, causing poor prediction performance for minority classes. Moreover, its other features, such as high-dimension, small sample, high noise etc., intensify this damage. In this study, we propose ACOSampling that is a novel undersampling method based on the idea of ant colony optimization (ACO) to address this problem. The algorithm starts with feature selection technology to eliminate noisy genes in data. Then we randomly and repeatedly divided the original training set into two groups: training set and validation set. In each division, one modified ACO algorithm as a variant of our previous work is conducted to filter less informative majority samples and search the corresponding optimal training sample subset. At last, the statistical results from all local optimal training sample subsets are given in the form of frequence list, where each frequence indicates the importance of the corresponding majority sample. We only extracted those high frequency ones and combined them with all minority samples to construct the final balanced training set. We evaluated the method on four benchmark skewed DNA microarray datasets by support vector machine (SVM) classifier, showing that the proposed method outperforms many other sampling approaches, which indicates its superiority.

NMR shielding tensors for density fitted local second-order Møller-Plesset perturbation theory using gauge including atomic orbitals

An efficient method for the calculation of nuclear magnetic resonance (NMR) shielding tensors is presented, which treats electron correlation at the level of second-order Mo̸ller-Plesset perturbation theory. It uses spatially localized functions to span occupied and virtual molecular orbital spaces, respectively, which are expanded in a basis of gauge including atomic orbitals (GIAOs or London atomic orbitals). Doubly excited determinants are restricted to local subsets of the virtual space and pair energies with an interorbital distance beyond a certain threshold are omitted. Furthermore, density fitting is employed to factorize the electron repulsion integrals. Ordinary Gaussians are employed as fitting functions. It is shown that the errors in the resulting NMR shielding constant, introduced (i) by the local approximation and (ii) by density fitting, are very small or even negligible. The capabilities of the new program are demonstrated by calculations on some extended molecular systems, such as the cyclobutane pyrimidine dimer photolesion with adjacent nucleobases in the native intrahelical DNA double strand (ATTA sequence). Systems of that size were not accessible to correlated ab initio calculations of NMR spectra before. The presented method thus opens the door to new and interesting applications in this area.

Relationship of CUL4A gene underexpression and prognosis in localized high-risk soft tissue sarcoma (STS) patients of limbs or trunk wall.

10079 Background: Genes involved in cell cycle checkpoints and chromosome stability seem to be relevant in distant metastases appearance in STS. CUL4A is an E3 ubiquitin ligase that has been related to both p16 activation and maintenance of genomic stability. Expression of CUL4A has been scarcely described in some tumors with divergent outcome and there are no publications regarding its expression in STS patients. We previously showed that protein underexpression of CUL4A in metastatic STS was related to poor survival. Methods: The expression level of CUL4A was determined by quantitative qPCR and was retrospectively performed in a subset of 39 high-risk (grade 3, deep tumours and > than 5 cm) localized STS of limbs or trunk wall patients for which histological material and clinical data were available. Patients of this series were enrolled prospectively into a randomized clinical trial conducted by the Italian and Spanish sarcoma groups (ISG-GEIS 0101) comparing 3 vs 5 cycles of epirrubicin and ifosfamide. Differences in PFS, RFS, and OS were calculated using log-rank test Results: The median age was 51 years with a median follow-up of 52 months. The pathologic subtypes were undifferentiated pleomorphic sarcoma (35.3%), synovial sarcoma (23.5%) and mixed subtypes (41.2%). Location was distributed as inferior limbs (66.6%), upper limbs (25.5%) and trunk wall (7.9%). No relationship was seen between CUL4A expression and clinical variables (histologic types, location, size or response to neoadjuvant treatment). The expression level of CUL4A was significantly associated with events of progression, 65% vs 32%, p =0.037, for expression values under and over the median respectively. Regarding 4 y actuarial survival, there was a significant worse PFS and RFS for patients underexpressing CUL4A (p=0.041 and 0.009 respectively) and a trend towards worse OS (p=0.056). Conclusions: Expression of CUL4A gene shows a prognostic role in sarcomas within this high risk localized subset of STS patients. The outcome of this exploratory analysis is aligned with our hypothesis that genes implicated in cell cycle regulation and chromosome stability could play a relevant prognostic role in STS.

ProClusEnsem: Predicting membrane protein types by fusing different modes of pseudo amino acid composition

Knowing the type of an uncharacterized membrane protein often provides a useful clue in both basic research and drug discovery. With the explosion of protein sequences generated in the post genomic era, determination of membrane protein types by experimental methods is expensive and time consuming. It therefore becomes important to develop an automated method to find the possible types of membrane proteins. In view of this, various computational membrane protein prediction methods have been proposed. They extract protein feature vectors, such as PseAAC (pseudo amino acid composition) and PsePSSM (pseudo position-specific scoring matrix) for representation of protein sequence, and then learn a distance metric for the KNN (K nearest neighbor) or NN (nearest neighbor) classifier to predicate the final type. Most of the metrics are learned using linear dimensionality reduction algorithms like Principle Components Analysis (PCA) and Linear Discriminant Analysis (LDA). Such metrics are common to all the proteins in the dataset. In fact, they assume that the proteins lie on a uniform distribution, which can be captured by the linear dimensionality reduction algorithm. We doubt this assumption, and learn local metrics which are optimized for local subset of the whole proteins. The learning procedure is iterated with the protein clustering. Then a novel ensemble distance metric is given by combining the local metrics through Tikhonov regularization. The experimental results on a benchmark dataset demonstrate the feasibility and effectiveness of the proposed algorithm named ProClusEnsem.

Efficient user selection for multi-cell multi-user MIMO systems with limited backhaul

Multi-cell multi-user multiple-input multiple-output (MC-MU-MIMO) is a promising technique to eliminate inter-user interference and inter-cell cochannel interference in wireless telecommunication systems. As the large number of users in the system and the limited number of simultaneously supportable users with MC-MU-MIMO, it is necessary to select a subset of users to maximize the total throughput. However, the fully centralized user selection algorithms used in single cell system, which will incur high complexity and backhaul load in multi-cell cooperative processing (MCP) systems, are not suitable to MC-MU-MIMO systems. This article presents a two cascaded user selection method for MCP systems with multi-cell block diagonalization. In this paper, a local optimal subset of users, which can maximize the local sum capacity, is first chosen by the greedy method in every cooperative base station in parallel. Then, all the cooperative base stations report their local optimal users to the central unit (CU). Finally, the global optimal users, which can maximize the global sum capacity of MCP systems, are selected from the aggregated local optimal users at the CU. The simulation results show that the proposed method performs closely to the optimal and centralized algorithm. Meanwhile, the complexity and backhaul load are reduced dramatically.

Determinants of the Drosophila Odorant Receptor Pattern

In most olfactory systems studied to date, neurons that express the same odorant receptor (Or) gene are scattered across sensory epithelia, intermingled with neurons that express different Or genes. In Drosophila, olfactory sensilla that express the same Or gene are dispersed on the antenna and the maxillary palp. Here we show that Or identity is specified in a spatially stereotyped pattern by the cell-autonomous activity of the transcriptional regulators Engrailed and Dachshund. Olfactory sensilla then become highly motile and disperse beneath the epidermis. Thus, positional information and cell motility underlie the dispersed patterns of Drosophila Or gene expression.