EMBL-EBI: moving, sharing and controlling large datasets

Supporting presentation slides as part of the Janet network end to end performance initiative

Scaling up data mining techniques to large datasets using parallel and distributed processing

Advances in hardware and software technology enable us to collect, store and distribute large quantities of data on a very large scale. Automatically discovering and extracting hidden knowledge in the form of patterns from these large data volumes is known as data mining. Data mining technology is not only a part of business intelligence, but is also used in many other application areas such as research, marketing and financial analytics. For example medical scientists can use patterns extracted from historic patient data in order to determine if a new patient is likely to respond positively to a particular treatment or not; marketing analysts can use extracted patterns from customer data for future advertisement campaigns; finance experts have an interest in patterns that forecast the development of certain stock market shares for investment recommendations. However, extracting knowledge in the form of patterns from massive data volumes imposes a number of computational challenges in terms of processing time, memory, bandwidth and power consumption. These challenges have led to the development of parallel and distributed data analysis approaches and the utilisation of Grid and Cloud computing. This chapter gives an overview of parallel and distributed computing approaches and how they can be used to scale up data mining to large datasets.

Efficient supervised optimum-path forest classification for large datasets

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Improving the accuracy of the optimum-path forest supervised classifier for large datasets

In this work, a new approach for supervised pattern recognition is presented which improves the learning algorithm of the Optimum-Path Forest classifier (OPF), centered on detection and elimination of outliers in the training set. Identification of outliers is based on a penalty computed for each sample in the training set from the corresponding number of imputable false positive and false negative classification of samples. This approach enhances the accuracy of OPF while still gaining in classification time, at the expense of a slight increase in training time. © 2010 Springer-Verlag.

A performance based study on gender recognition in large datasets

[EN]Gender recognition has achieved impressive results based on the face appearance in controlled datasets. Its application in the wild and large datasets is still a challenging task for researchers. In this paper, we make use of classical techniques to analyze their performance in controlled and uncontrolled condition respectively with the LFW and MORPH datasets. For both sets the benchmarking protocol follows the 5-fold cross-validation proposed by the BEFIT challenge.

Computational Techniques for Spatial Logistic Regression with Large Datasets

In epidemiological work, outcomes are frequently non-normal, sample sizes may be large, and effects are often small. To relate health outcomes to geographic risk factors, fast and powerful methods for fitting spatial models, particularly for non-normal data, are required. We focus on binary outcomes, with the risk surface a smooth function of space. We compare penalized likelihood models, including the penalized quasi-likelihood (PQL) approach, and Bayesian models based on fit, speed, and ease of implementation. A Bayesian model using a spectral basis representation of the spatial surface provides the best tradeoff of sensitivity and specificity in simulations, detecting real spatial features while limiting overfitting and being more efficient computationally than other Bayesian approaches. One of the contributions of this work is further development of this underused representation. The spectral basis model outperforms the penalized likelihood methods, which are prone to overfitting, but is slower to fit and not as easily implemented. Conclusions based on a real dataset of cancer cases in Taiwan are similar albeit less conclusive with respect to comparing the approaches. The success of the spectral basis with binary data and similar results with count data suggest that it may be generally useful in spatial models and more complicated hierarchical models.

Pragmatic algorithms for implementing geostatistics with large datasets

With the ability to collect and store increasingly large datasets on modern computers comes the need to be able to process the data in a way that can be useful to a Geostatistician or application scientist. Although the storage requirements only scale linearly with the number of observations in the dataset, the computational complexity in terms of memory and speed, scale quadratically and cubically respectively for likelihood-based Geostatistics. Various methods have been proposed and are extensively used in an attempt to overcome these complexity issues. This thesis introduces a number of principled techniques for treating large datasets with an emphasis on three main areas: reduced complexity covariance matrices, sparsity in the covariance matrix and parallel algorithms for distributed computation. These techniques are presented individually, but it is also shown how they can be combined to produce techniques for further improving computational efficiency.

Projected sequential Gaussian processes:a C++ tool for interpolation of large datasets with heterogeneous noise

Heterogeneous datasets arise naturally in most applications due to the use of a variety of sensors and measuring platforms. Such datasets can be heterogeneous in terms of the error characteristics and sensor models. Treating such data is most naturally accomplished using a Bayesian or model-based geostatistical approach; however, such methods generally scale rather badly with the size of dataset, and require computationally expensive Monte Carlo based inference. Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential Bayesian framework for inference in such projected processes is presented. The observations are considered one at a time which avoids the need for high dimensional integrals typically required in a Bayesian approach. A C++ library, gptk, which is part of the INTAMAP web service, is introduced which implements projected, sequential estimation and adds several novel features. In particular the library includes the ability to use a generic observation operator, or sensor model, to permit data fusion. It is also possible to cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the covariance parameters is explored, including the impact of the projected process approximation on likelihood profiles. We illustrate the projected sequential method in application to synthetic and real datasets. Limitations and extensions are discussed. © 2010 Elsevier Ltd.

Fast regridding of large, complex geospatial datasets

In the earth sciences, data are commonly cast on complex grids in order to model irregular domains such as coastlines, or to evenly distribute grid points over the globe. It is common for a scientist to wish to re-cast such data onto a grid that is more amenable to manipulation, visualization, or comparison with other data sources. The complexity of the grids presents a significant technical difficulty to the regridding process. In particular, the regridding of complex grids may suffer from severe performance issues, in the worst case scaling with the product of the sizes of the source and destination grids. We present a mechanism for the fast regridding of such datasets, based upon the construction of a spatial index that allows fast searching of the source grid. We discover that the most efficient spatial index under test (in terms of memory usage and query time) is a simple look-up table. A kd-tree implementation was found to be faster to build and to give similar query performance at the expense of a larger memory footprint. Using our approach, we demonstrate that regridding of complex data may proceed at speeds sufficient to permit regridding on-the-fly in an interactive visualization application, or in a Web Map Service implementation. For large datasets with complex grids the new mechanism is shown to significantly outperform algorithms used in many scientific visualization packages.

A combined neural network and DEA for measuring efficiency of large scale datasets

Data Envelopment Analysis (DEA) is one of the most widely used methods in the measurement of the efficiency and productivity of Decision Making Units (DMUs). DEA for a large dataset with many inputs/outputs would require huge computer resources in terms of memory and CPU time. This paper proposes a neural network back-propagation Data Envelopment Analysis to address this problem for the very large scale datasets now emerging in practice. Neural network requirements for computer memory and CPU time are far less than that needed by conventional DEA methods and can therefore be a useful tool in measuring the efficiency of large datasets. Finally, the back-propagation DEA algorithm is applied to five large datasets and compared with the results obtained by conventional DEA.

Would you like data with that? Library provision of financial datasets

Within the QUT Business School (QUTBS)– researchers across economics, finance and accounting depend on data driven research. They analyze historic and global financial data across a range of instruments to understand the relationships and effects between them as they respond to news and events in their region. Scholars and Higher Degree Research Students in turn seek out universities which offer these particular datasets to further their research. This involves downloading and manipulating large datasets, often with a focus on depth of detail, frequency and long tail historical data. This is stock exchange data and has potential commercial value therefore the license for access tends to be very expensive. This poster reports the following findings: •The library has a part to play in freeing up researchers from the burden of negotiating subscriptions, fundraising and managing the legal requirements around license and access. •The role of the library is to communicate the nature and potential of these complex resources across the university to disciplines as diverse as Mathematics, Health, Information Systems and Creative Industries. •Has demonstrated clear concrete support for research by QUT Library and built relationships into faculty. It has made data available to all researchers and attracted new HDRs. The aim is to reach the output threshold of research outputs to submit into FOR Code 1502 (Banking, Finance and Investment) for ERA 2015. •It is difficult to identify what subset of dataset will be obtained given somewhat vague price tiers. •The integrity of data is variable as it is limited by the way it is collected, this occasionally raises issues for researchers(Cook, Campbell, & Kelly, 2012) •Improved library understanding of the content of our products and the nature of financial based research is a necessary part of the service.

R1SVM: A randomised nonlinear approach to large-scale anomaly detection

The problem of unsupervised anomaly detection arises in a wide variety of practical applications. While one-class support vector machines have demonstrated their effectiveness as an anomaly detection technique, their ability to model large datasets is limited due to their memory and time complexity for training. To address this issue for supervised learning of kernel machines, there has been growing interest in random projection methods as an alternative to the computationally expensive problems of kernel matrix construction and sup-port vector optimisation. In this paper we leverage the theory of nonlinear random projections and propose the Randomised One-class SVM (R1SVM), which is an efficient and scalable anomaly detection technique that can be trained on large-scale datasets. Our empirical analysis on several real-life and synthetic datasets shows that our randomised 1SVM algorithm achieves comparable or better accuracy to deep auto encoder and traditional kernelised approaches for anomaly detection, while being approximately 100 times faster in training and testing.

Clustering Based Large Margin Classification: A Scalable Approach using SOCP Formulation

This paper presents a novel Second Order Cone Programming (SOCP) formulation for large scale binary classification tasks. Assuming that the class conditional densities are mixture distributions, where each component of the mixture has a spherical covariance, the second order statistics of the components can be estimated efficiently using clustering algorithms like BIRCH. For each cluster, the second order moments are used to derive a second order cone constraint via a Chebyshev-Cantelli inequality. This constraint ensures that any data point in the cluster is classified correctly with a high probability. This leads to a large margin SOCP formulation whose size depends on the number of clusters rather than the number of training data points. Hence, the proposed formulation scales well for large datasets when compared to the state-of-the-art classifiers, Support Vector Machines (SVMs). Experiments on real world and synthetic datasets show that the proposed algorithm outperforms SVM solvers in terms of training time and achieves similar accuracies.

Robust formulations for clustering-based large-scale classification

Chebyshev-inequality-based convex relaxations of Chance-Constrained Programs (CCPs) are shown to be useful for learning classifiers on massive datasets. In particular, an algorithm that integrates efficient clustering procedures and CCP approaches for computing classifiers on large datasets is proposed. The key idea is to identify high density regions or clusters from individual class conditional densities and then use a CCP formulation to learn a classifier on the clusters. The CCP formulation ensures that most of the data points in a cluster are correctly classified by employing a Chebyshev-inequality-based convex relaxation. This relaxation is heavily dependent on the second-order statistics. However, this formulation and in general such relaxations that depend on the second-order moments are susceptible to moment estimation errors. One of the contributions of the paper is to propose several formulations that are robust to such errors. In particular a generic way of making such formulations robust to moment estimation errors is illustrated using two novel confidence sets. An important contribution is to show that when either of the confidence sets is employed, for the special case of a spherical normal distribution of clusters, the robust variant of the formulation can be posed as a second-order cone program. Empirical results show that the robust formulations achieve accuracies comparable to that with true moments, even when moment estimates are erroneous. Results also illustrate the benefits of employing the proposed methodology for robust classification of large-scale datasets.

Large-Scale Elastic Net Regularized Linear Classification SVMs and Logistic Regression

Elastic Net Regularizers have shown much promise in designing sparse classifiers for linear classification. In this work, we propose an alternating optimization approach to solve the dual problems of elastic net regularized linear classification Support Vector Machines (SVMs) and logistic regression (LR). One of the sub-problems turns out to be a simple projection. The other sub-problem can be solved using dual coordinate descent methods developed for non-sparse L2-regularized linear SVMs and LR, without altering their iteration complexity and convergence properties. Experiments on very large datasets indicate that the proposed dual coordinate descent - projection (DCD-P) methods are fast and achieve comparable generalization performance after the first pass through the data, with extremely sparse models.