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.

Dynamic group communication for large-scale parallel data mining

Exascale systems are the next frontier in high-performance computing and are expected to deliver a performance of the order of 10^18 operations per second using massive multicore processors. Very large- and extreme-scale parallel systems pose critical algorithmic challenges, especially related to concurrency, locality and the need to avoid global communication patterns. This work investigates a novel protocol for dynamic group communication that can be used to remove the global communication requirement and to reduce the communication cost in parallel formulations of iterative data mining algorithms. The protocol is used to provide a communication-efficient parallel formulation of the k-means algorithm for cluster analysis. The approach is based on a collective communication operation for dynamic groups of processes and exploits non-uniform data distributions. Non-uniform data distributions can be either found in real-world distributed applications or induced by means of multidimensional binary search trees. The analysis of the proposed dynamic group communication protocol has shown that it does not introduce significant communication overhead. The parallel clustering algorithm has also been extended to accommodate an approximation error, which allows a further reduction of the communication costs. The effectiveness of the exact and approximate methods has been tested in a parallel computing system with 64 processors and in simulations with 1024 processing elements.

The role of soil microbes in the global carbon cycle: tracking the below-ground microbial processing of plant-derived carbon for manipulating carbon dynamics in agricultural systems.

It is well known that atmospheric concentrations of carbon dioxide (CO2) (and other greenhouse gases) have increased markedly as a result of human activity since the industrial revolution. It is perhaps less appreciated that natural and managed soils are an important source and sink for atmospheric CO2 and that, primarily as a result of the activities of soil microorganisms, there is a soil-derived respiratory flux of CO2 to the atmosphere that overshadows by tenfold the annual CO2 flux from fossil fuel emissions. Therefore small changes in the soil carbon cycle could have large impacts on atmospheric CO2 concentrations. Here we discuss the role of soil microbes in the global carbon cycle and review the main methods that have been used to identify the microorganisms responsible for the processing of plant photosynthetic carbon inputs to soil. We discuss whether application of these techniques can provide the information required to underpin the management of agro-ecosystems for carbon sequestration and increased agricultural sustainability. We conclude that, although crucial in enabling the identification of plant-derived carbon-utilising microbes, current technologies lack the high-throughput ability to quantitatively apportion carbon use by phylogentic groups and its use efficiency and destination within the microbial metabolome. It is this information that is required to inform rational manipulation of the plant–soil system to favour organisms or physiologies most important for promoting soil carbon storage in agricultural soil.

Median architecture by accumulative parallel counters

The time to process each of W/B processing blocks of a median calculation method on a set of N W-bit integers is improved here by a factor of three compared to the literature. Parallelism uncovered in blocks containing B-bit slices are exploited by independent accumulative parallel counters so that the median is calculated faster than any known previous method for any N, W values. The improvements to the method are discussed in the context of calculating the median for a moving set of N integers for which a pipelined architecture is developed. An extra benefit of smaller area for the architecture is also reported.

Reliability Nonparametric Bayesian Estimation in Parallel Systems

Relevant results for (sub-)distribution functions related to parallel systems are discussed. The reverse hazard rate is defined using the product integral. Consequently, the restriction of absolute continuity for the involved distributions can be relaxed. The only restriction is that the sets of discontinuity points of the parallel distributions have to be disjointed. Nonparametric Bayesian estimators of all survival (sub-)distribution functions are derived. Dual to the series systems that use minimum life times as observations, the parallel systems record the maximum life times. Dirichlet multivariate processes forming a class of prior distributions are considered for the nonparametric Bayesian estimation of the component distribution functions, and the system reliability. For illustration, two striking numerical examples are presented.

Harmonic identification using parallel neural networks in single-phase systems

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

A logical approach to the management of object identifiers in non-conventional database management systems

Non-conventional database management systems are used to achieve a better performance when dealing with complex data. One fundamental concept of these systems is object identity (OID), because each object in the database has a unique identifier that is used to access and reference it in relationships to other objects. Two approaches can be used for the implementation of OIDs: physical or logical OIDs. In order to manage complex data, was proposed the Multimedia Data Manager Kernel (NuGeM) that uses a logical technique, named Indirect Mapping. This paper proposes an improvement to the technique used by NuGeM, whose original contribution is management of OIDs with a fewer number of disc accesses and less processing, thus reducing management time from the pages and eliminating the problem with exhaustion of OIDs. Also, the technique presented here can be applied to others OODBMSs. © 2011 IEEE.

The embodied embedded character of system 1 processing

In the last thirty years, a relatively large group of cognitive scientists have begun characterising the mind in terms of two distinct, relatively autonomous systems. To account for paradoxes in empirical results of studies mainly on reasoning, Dual Process Theories were developed. Such Dual Process Theories generally agree that System 1 is rapid, automatic, parallel, and heuristic-based and System 2 is slow, capacity-demanding, sequential, and related to consciousness. While System 2 can still be decently understood from a traditional cognitivist approach, I will argue that it is essential for System 1 processing to be comprehended in an Embodied Embedded approach to Cognition.© MSM 2013.

Psychophysical analysis of contrast processing segregated into magnocellular and parvocellular systems in asymptomatic carriers of 11778 Leber`s hereditary optic neuropathy

We examined achromatic contrast discrimination in asymptomatic carriers of 11778 Leber`s hereditary optic neuropathy (LHON 18 controls) and 18 age-match were also tested. To evaluate magnocellular (MC) and Parvocellular (PC) contrast discrimination, we used a version of Pokorny and Smith`s (1997) Pulsed/steady-pedestal paradigms (PPP/SPP) thought to be detected via PC and MC pathways, respectively. A luminance pedestal (four 1 degrees x 1 degrees squares) was presented on a 12 cd/m(2) surround. The luminance of one of the squares (trial square, TS) was randomly incremented for either 17 or 133 ms. Observers had to detect the TS, in a forced-choice task, at each duration, for three pedestal levels: 7, 12, 19 cd/m(2). In the SPP, the pedestal was fixed, and the TS was modulated. For the PPP, all four pedestal squares pulsed for 17 or 133 ms, and the TS was simultaneously incremented or decremented. We found that contrast discrimination thresholds of LHON carriers were significantly higher than controls` in the condition with the highest luminance of both paradigms, implying impaired contrast processing with no evidence of differential sensitivity losses between the two systems. Carriers` thresholds manifested significantly longer temporal integration than controls in the SPP, consistent with slowed MC responses. The SPP and PPP paradigms can identify contrast and temporal processing deficits in asymptomatic LHON carriers, and thus provide an additional tool for early detection and characterization of the disease.

Polymorphic toxin systems: Comprehensive characterization of trafficking modes, processing, mechanisms of action, immunity and ecology using comparative genomics

Background: Proteinaceous toxins are observed across all levels of inter-organismal and intra-genomic conflicts. These include recently discovered prokaryotic polymorphic toxin systems implicated in intra-specific conflicts. They are characterized by a remarkable diversity of C-terminal toxin domains generated by recombination with standalone toxin-coding cassettes. Prior analysis revealed a striking diversity of nuclease and deaminase domains among the toxin modules. We systematically investigated polymorphic toxin systems using comparative genomics, sequence and structure analysis. Results: Polymorphic toxin systems are distributed across all major bacterial lineages and are delivered by at least eight distinct secretory systems. In addition to type-II, these include type-V, VI, VII (ESX), and the poorly characterized "Photorhabdus virulence cassettes (PVC)", PrsW-dependent and MuF phage-capsid-like systems. We present evidence that trafficking of these toxins is often accompanied by autoproteolytic processing catalyzed by HINT, ZU5, PrsW, caspase-like, papain-like, and a novel metallopeptidase associated with the PVC system. We identified over 150 distinct toxin domains in these systems. These span an extraordinary catalytic spectrum to include 23 distinct clades of peptidases, numerous previously unrecognized versions of nucleases and deaminases, ADP-ribosyltransferases, ADP ribosyl cyclases, RelA/SpoT-like nucleotidyltransferases, glycosyltranferases and other enzymes predicted to modify lipids and carbohydrates, and a pore-forming toxin domain. Several of these toxin domains are shared with host-directed effectors of pathogenic bacteria. Over 90 families of immunity proteins might neutralize anywhere between a single to at least 27 distinct types of toxin domains. In some organisms multiple tandem immunity genes or immunity protein domains are organized into polyimmunity loci or polyimmunity proteins. Gene-neighborhood-analysis of polymorphic toxin systems predicts the presence of novel trafficking-related components, and also the organizational logic that allows toxin diversification through recombination. Domain architecture and protein-length analysis revealed that these toxins might be deployed as secreted factors, through directed injection, or via inter-cellular contact facilitated by filamentous structures formed by RHS/YD, filamentous hemagglutinin and other repeats. Phyletic pattern and life-style analysis indicate that polymorphic toxins and polyimmunity loci participate in cooperative behavior and facultative 'cheating' in several ecosystems such as the human oral cavity and soil. Multiple domains from these systems have also been repeatedly transferred to eukaryotes and their viruses, such as the nucleo-cytoplasmic large DNA viruses. Conclusions: Along with a comprehensive inventory of toxins and immunity proteins, we present several testable predictions regarding active sites and catalytic mechanisms of toxins, their processing and trafficking and their role in intra-specific and inter-specific interactions between bacteria. These systems provide insights regarding the emergence of key systems at different points in eukaryotic evolution, such as ADP ribosylation, interaction of myosin VI with cargo proteins, mediation of apoptosis, hyphal heteroincompatibility, hedgehog signaling, arthropod toxins, cell-cell interaction molecules like teneurins and different signaling messengers.

A novel strategy for building interoperable MPI environment in heterogeneous high performance systems

Breakthrough advances in microprocessor technology and efficient power management have altered the course of development of processors with the emergence of multi-core processor technology, in order to bring higher level of processing. The utilization of many-core technology has boosted computing power provided by cluster of workstations or SMPs, providing large computational power at an affordable cost using solely commodity components. Different implementations of message-passing libraries and system softwares (including Operating Systems) are installed in such cluster and multi-cluster computing systems. In order to guarantee correct execution of message-passing parallel applications in a computing environment other than that originally the parallel application was developed, review of the application code is needed. In this paper, a hybrid communication interfacing strategy is proposed, to execute a parallel application in a group of computing nodes belonging to different clusters or multi-clusters (computing systems may be running different operating systems and MPI implementations), interconnected with public or private IP addresses, and responding interchangeably to user execution requests. Experimental results demonstrate the feasibility of this proposed strategy and its effectiveness, through the execution of benchmarking parallel applications.

An online data access prediction and optimization approach for distributed systems

Current scientific applications have been producing large amounts of data. The processing, handling and analysis of such data require large-scale computing infrastructures such as clusters and grids. In this area, studies aim at improving the performance of data-intensive applications by optimizing data accesses. In order to achieve this goal, distributed storage systems have been considering techniques of data replication, migration, distribution, and access parallelism. However, the main drawback of those studies is that they do not take into account application behavior to perform data access optimization. This limitation motivated this paper which applies strategies to support the online prediction of application behavior in order to optimize data access operations on distributed systems, without requiring any information on past executions. In order to accomplish such a goal, this approach organizes application behaviors as time series and, then, analyzes and classifies those series according to their properties. By knowing properties, the approach selects modeling techniques to represent series and perform predictions, which are, later on, used to optimize data access operations. This new approach was implemented and evaluated using the OptorSim simulator, sponsored by the LHC-CERN project and widely employed by the scientific community. Experiments confirm this new approach reduces application execution time in about 50 percent, specially when handling large amounts of data.

Design of alternative energy storage systems for hybrid vehicles based on statistical processing of driving cycles information

Hybrid vehicles represent the future for automakers, since they allow to improve the fuel economy and to reduce the pollutant emissions. A key component of the hybrid powertrain is the Energy Storage System, that determines the ability of the vehicle to store and reuse energy. Though electrified Energy Storage Systems (ESS), based on batteries and ultracapacitors, are a proven technology, Alternative Energy Storage Systems (AESS), based on mechanical, hydraulic and pneumatic devices, are gaining interest because they give the possibility of realizing low-cost mild-hybrid vehicles. Currently, most literature of design methodologies focuses on electric ESS, which are not suitable for AESS design. In this contest, The Ohio State University has developed an Alternative Energy Storage System design methodology. This work focuses on the development of driving cycle analysis methodology that is a key component of Alternative Energy Storage System design procedure. The proposed methodology is based on a statistical approach to analyzing driving schedules that represent the vehicle typical use. Driving data are broken up into power events sequence, namely traction and braking events, and for each of them, energy-related and dynamic metrics are calculated. By means of a clustering process and statistical synthesis methods, statistically-relevant metrics are determined. These metrics define cycle representative braking events. By using these events as inputs for the Alternative Energy Storage System design methodology, different system designs are obtained. Each of them is characterized by attributes, namely system volume and weight. In the last part the work, the designs are evaluated in simulation by introducing and calculating a metric related to the energy conversion efficiency. Finally, the designs are compared accounting for attributes and efficiency values. In order to automate the driving data extraction and synthesis process, a specific script Matlab based has been developed. Results show that the driving cycle analysis methodology, based on the statistical approach, allows to extract and synthesize cycle representative data. The designs based on cycle statistically-relevant metrics are properly sized and have satisfying efficiency values with respect to the expectations. An exception is the design based on the cycle worst-case scenario, corresponding to same approach adopted by the conventional electric ESS design methodologies. In this case, a heavy system with poor efficiency is produced. The proposed new methodology seems to be a valid and consistent support for Alternative Energy Storage System design.