Abstract: Frog population has been declining the past decade for habitat loss, invasive species, climate change, and so forth. Therefore, it is becoming ever more important to monitor the frog population. Recent advances in acoustic sensors make it possible to collect frog vocalizations over large spatio-temporal scale. Through the detection of frog calling activity with collected acoustic data, frog population can be predicted. In this paper we propose a novel method for detecting frog calling activity using acoustic event detection and multi-label learning. Here, frog calling activity consists of frog abundance and frog species richness, which denotes number of individual frog calls and number of frog species respectively. To be specific, each segmented recording is first transformed to a spectrogram. Then, acoustic event detection is used to calculate frog abundance. Meanwhile, those recordings without frog calls are filtered out. For frog species richness, three acoustic features, linear predictive coefficients, Mel-frequency Cepstral coefficients and wavelet-based features are calculated. Then, multi-label learning is used to predict frog species richness. Lastly, statistical analysis is used to reflect the relationship between frog calling activity (frog abundance and frog species richness) and weather variables. Experiment results show that our proposed method can accurately detect frog calling activity and reflect its relationship with weather variables.

Abstract: In the past few years massive amounts of data have been generated for genetic analysis. Existing solutions to analyze this data concerning genome-wide gene interactions are either not powerful enough or can barely be managed with standard computers due to the tremendous amount of statistical tests to be performed. Also, common approaches using cluster or cloud technologies for parallel analysis are operating at the edge of what is currently possible. This work demonstrates how FPGAs are able to address this problem. We present a highly parallel, hardware oriented solution for genome-wide association interaction studies (GWAIS) with MB-MDR and the maxT multiple testing correction on an FPGA-based architecture. We achieve a more than 300-fold speedup over an AMD Opteron cluster with 160 cores on an FPGA-system equipped with 128 Xilinx Spartan6 LX150 low-cost FPGAs when analyzing a WTCCC-like dataset with 500,000 markers and 5,000 samples. Furthermore, we are able to keep pace with a 256-core Intel Xeon cluster running MB-MDR~4.2.2 with an approximative version of maxT, while we achieve a 190-fold speedup over the sequential execution of this version on one Xeon core.

Abstract: Computation is often seen as information processing. Many biological systems may be investigated in terms of information storage, signaling, and data processing networks. Much of this data processing activity is embodied in structural transformations in spatial scales ranging from the molecular to cellular networks. The biomedical sciences make use of an increasingly powerful arsenal of tools and technologies for obtaining structural data as well as details of mass transport and the chemical and electrical signals that underlie these fundamental biological processes. For example, new staining techniques combined with computer-based electron microscope tomography, permit the clear imaging of chromatin filaments in the cell nucleus and filament networks in the cytoplasmic and extracellular space via the electron microscope. The application of tomographic reconstruction software developed at the National Center for Microscopy and Imaging Research (NCMIR) enables detailed 3D reconstructions of the relevant biological structures and processes. In order to deal with fundamental issues related to information processing in biological systems, new data processing methods as well as advances in chemically sensitive probes and imaging technology must be applied across a wide range of spatial and temporal scales. One class of increasingly useful tools for modeling biological systems, evaluating imaging technologies and characterizing the fidelity of digital processing has its roots in theoretical investigations in statistical mechanics, which arise from the concepts of information and entropy. We review how concepts of information and entropy may give new perspectives on the flow of information within biological systems, as well as our instrumentation and computer processing.

Abstract: In recent years increased the research interest in the development of different approaches to support the mobility of the visually impaired. The automatic detection of tactile paving surface is one important topic of research, not only to help the mobility of visually impaired persons, but also for use in the displacement of autonomous robots, providing a safely route and warnings. In this paper we propose an approach for tactile paving surface detection in real-time with the purpose to assist visually impaired persons. It uses computer vision algorithms combined with decision tree to eliminate some possible false alarms. We assume the visually impaired persons holds a smartphone, which is used to obtain images, as well as to assist him by audio feedback to keep it on the tactile paving surface. This problem is very challenging, mainly due to illumination changes, occlusion, image noise and resolution, as well as different possible colors of the tactile paving surfaces. Experimental results indicate that the proposed approach works well in low resolution images, effectively detecting the tactile paving surfaces in real test scenarios.

Abstract: Inspired by the social behavior of the bird flocking or fish schooling, the particle swarm optimization (PSO) is a population based stochastic optimization method developed by Eberhart and Kennedy in 1995. It has been used across a wide range of applications. Faure, Halton and Vander Corput sequences have been used for initializing the swarm in PSO. Quasirandom(or low-discrepancy) sequences such as Faure, Halton, Vander Corput etc are deterministic and suffers from correlations between radical inverse functions with different bases used for different dimensions. In this paper, we investigate the effect of initializing the swarm with scrambled optimal Halton sequence, which is a randomized quasirandom sequence. This ensures that we still have the uniformity properties of quasirandom sequences while preserving the stochastic behavior for particles in the swarm. Numerical experiments are conducted with benchmark objective functions with high dimensions to verify the convergence and effectiveness of the proposed initialization of PSO

Abstract: When writing parallel software for high performance computing, a common practice is to start from a sequential variant of a program that is consecutively enriched with parallelization directives. This process - progressive parallelization - has the advantage that, at every point in time, a correct version of the program exists. However, progressive parallelization leads to an entanglement of concerns, especially, if different variants of the same functional code have to be maintained and evolved concurrently. We propose orchestration style sheets (OSS) as a novel approach to separate parallelization concerns from problem-specific code by placing them in reusable style sheets, so that concerns for different platforms are always separated, and never lead to entanglement. A weaving process automatically generates platform-specific code for required target platforms, taking semantic properties of the source code into account. Based on a scientific computing case study for fluid mechanics, we show that OSS are an adequate way to improve maintainability and reuse of Fortran code parallelized for several different platforms.

Abstract: Increasing variety and affordability of multi- and many-core embedded architectures can pose both a challenge and opportunity to developers of high performance computing applications. In this paper we present a case study where we develop and evaluate a unified parallel approach to correlation signal correlation algorithm,currently in use in a commercial/industrial locating system. We utilize both HPX C++ and CUDA runtimes to achieve scalable code for current embedded multi- and many-core architectures (NVIDIA Tegra, Intel Broadwell M, Arm Cortex A-15). We also compare our approach onto traditional high-performance hardware as well as a native embedded many-core variant. To increase the accuracy of our performance analysis we introduce dedicated performance model. The results show that our approach is feasible and enables us to harness the advantages of modern micro-server architectures, but also indicates that there are limitations to some of the currently existing many-core embedded architectures, that can lead to traditional hardware being superior both in efficiency and absolute performance.

Abstract: An important challenge of dataflow programming is the problem of partitioning dataflow components onto a target architecture. A common objective function associated to this problem is to find the maximum data processing throughput. This NP-complete problem is very difficult to solve with high quality close-to-optimal solutions for the very large size of the design space and the possibly large variability of input data. This paper introduces four variants of the tabu search metaheuristic expressly developed for partitioning components of a dataflow program. The approach relies on the use of a simulation tool, capable of estimating the performance for any partitioning configuration exploiting a model of the target architecture and the profiling results. The partitioning solutions generated with tabu search are validated for consistency and high accuracy with experimental platform executions.

Abstract: The definition of an efficient scheduling policy is an important, difficult and open design problem for the implementation of applications based on dynamic dataflow programs for which optimal closed-form solutions do not exist. This paper describes an approach based on the study of the execution of a dynamic dataflow program on a target architecture with different scheduling policies. The method is based on a representation of the execution of a dataflow program with the associated dependencies, and on the cost of using scheduling policy, expressed as a number of conditions that need to be verified to have a successful execution within each partition. The relation between the potential gain of the overall execution satisfying intrinsic data dependencies and the runtime cost of finding an admissible schedule is a key issue to find close-to-optimal solutions for the scheduling problem of dynamic dataflow applications.

Abstract: Shared memory is a critical issue for large distributed systems. Despite several data consistency protocols have been proposed, the selection of the protocol that best suits to the application requirements and system constraints remains a challenge. The development of multi-consistency systems, where different protocols can be deployed during runtime, appears to be an interesting alternative. In order to explore the design space of the consistency protocols a fast and accurate method should be used. In this work we rely on a compilation toolchain that transparently handles data consistency decisions for a multi-protocol platform. We focus on the analytical evaluation of the consistency configuration that stands within the optimization loop. We propose to use a TLM NoC simulator to get feedback on expected network contentions. We evaluate the approach using five workloads and three different data consistency protocols. As a result, we are able to obtain a fast and accurate evaluation of the different consistency alternatives.

Abstract: Isoniazid is one of the front-line drugs in the treatment of tuberculosis for many years and the mechanism of the action of the drugs still not clear. Herein we report the synthesis, QSAR model study and biological effects of ester and thioester derivatives of isonicotinic acid (INA) on different strains of Mycobacterium tuberculosis in an attempt to establish the mode of action of the active form of Isoniazid (INH). The esters of INA is expected to show antibiotic activities against strains of Mycobacterium tuberculosis upon activation by intracellular non-specific esterases, while thioesters should not show antimicrobial effects and thus served as the negative control. The results indicated that the synthesized esters did not show antimicrobial activity. Hence, the conclusion drawn is that INA is not the active form of INH. Hence more studies are necessary to search for the possible active form of isoniazid.

Abstract: We will describe the science discovered by some of the community of researchers using the SEAGrid Science gateway using computational chemistry applications. Specific science projects to be discussed include calcium carbonate and bicorbonate hydrochemistry, graphene application modeling, photonic properties of some heavy metal complexes, mechanistic studies of redox proteins and diffraction modeling of metal and metal-oxide structures and interfaces. The modeling studies involve a variety of computational techniques and coupled execution of a workflows using specific set of applications enabled in the SEAGrid Science Gateway. The integration of applications and resources that enable workflows that integrate empirical, semi-empirical ab initio and DFT and perturbative ab initio techniques through a single point of access will be presented. SEAGrid gateway hitherto used Computational Chemistry Grid middlware infrastructure and provided GridChem desktop client for users to interact with the resources and services. This deployment suffers from maintainging unsupported service architectures and system specific scripts. Going forward, the services will be outsourced to the Apache Airavata infrastructure to gain from a sustainable and more easily maintaiable set of services. As part of the new deployment we will also provide a web browser based SEAGrid Portal in addition to the SEAGrid desktop application based on the previous GridChem client. We will elaborate the services and their enhancement in this process to exemplify how the new implementation will enhance the maintainability and sustainability. We will also provide exemplar science workflows and contrast how they are supported in the new deployment relative to the previous deployment to show case the adoptability and user support for services and resources.

Abstract: Algorithms with linear-scaling (O(N)) computational complexity for Kohn-Sham density functional theory (K-S DFT) is crucial for studying molecular systems beyond thousands of atoms. A typical cross-over point of an O(N) method with an cubic-scaling method lies between 600 − 1000 atoms. Of the O(N) methods that use a polynomial-based approximation of the density matrix, the linear-scaling spectral Gauss quadrature (LSSGQ) method (Suryanarayana, JMPS, 2013) has been shown to exhibit the fastest convergence. The LSSGQ method requires a Lanczos procedure at every node in a real-space mesh, leading to a large computational pre-factor. With the goal of reducing the computational pre-factor of the LSSGQ method, and consequently its cross-over point with an O(N3) method, we propose a new interpolation scheme specific to the LSSGQ method that lift the need to perform a Lanczos procedure at every node in the real-mesh. This interpolation will be referred to as subspace interpolation. The key idea behind subspace interpolation is that there is a large overlap in the Krylov-subspaces produced by the Lanczos procedures of nodes that are close in real-space. This overlap is more pronounced in a collection of nodes where the electron density of the system is not varying much spatially. The subspace interpolation scheme takes advantage of the block-Lanczos procedure to group the Krylov-subspaces from a few representative nodes to approximate the density matrix over a larger collection of nodes; reducing the number of nodes on which the Lanczos procedure has to be performed. Subspace interpolation outperforms cubic-spline interpolation by several orders of magnitude.

Abstract: Mining data streams is one of the most vital fields in the current era of big data. Continuously arriving data may pose various problems, connected to their volume, variety or velocity. In this paper we focus on two important difficulties embedded in the nature of data streams: non-stationary nature and skewed class distributions. Such a scenario requires a classifier that is able to rapidly adapt itself to concept drift and displays robustness to class imbalance problem. We propose to use online version of Extreme Learning Machine that is enhanced by an efficient drift detector and method to alleviate the bias towards the majority class. We investigate three approaches based on undersampling, oversampling and cost-sensitive adaptation. Additionally, to allow for a rapid updating of the proposed classifier we show how to implement online Extreme Learning Machines with the usage of GPU. The proposed approach allows for a highly efficient mining of high-speed, drifting and imbalanced data streams with significant acceleration offered by GPU processing.

Abstract: In this paper new methods for fast computation of the chordal kernels are proposed. Two versions of the chordal kernels for tensor data are discussed. These are based on different projectors of the flattened matrices obtained from the input tensors. A direct transformation of multidimensional objects into the kernel feature space leads to better data separation which can result with a higher classification accuracy. Our approach to more efficient computation of the chordal distances between tensors is based on an analysis of the tensor projectors which exhibit different properties. Thanks to this an efficient eigen-decomposition becomes possible which is done with a version of the fixed-point algorithm. Experimental results show that our method allows significant speed-up factors, depending mostly on tensor dimensions.

Abstract: In this paper we explore a new model focused on integrating two classifiers; Convolutional Neural Network (CNN) and Support Vector Machine (SVM) for offline Arabic handwriting recognition (OAHR) on which the dropout technique was applied. The suggested system altered the trainable classifier of the CNN by the SVM classifier. A convolutional network is beneficial for extracting features information and SVM functions as a recognizer. It was found that this model both automatically extracts features from the raw images and performs classification. Additionally, we protected our model against over-fitting due to the powerful performance of dropout. In this work, the recognition on the handwritten Arabic characters was evaluated; the training and test sets were taken from the HACDB and IFN/ENIT databases. Simulation results proved that the new design based-SVM of the CNN classifier architecture with dropout performs significantly more efficiently than CNN based-SVM model without dropout and the standard CNN classifier. The performance of our model is compared with character recognition accuracies gained from state-of-the-art Arabic Optical Character Recognition, producing favorable results.

Abstract: Contemporary classification systems have to make a decision not only on the basis of the static data, but on the data in motion as well. Objects being recognized may arrive continuously to a classifier in the form of data stream. Usually, we would like to start exploitation of the classifier as soon as possible, the models which can improve their models during exportation are very desirable. Basically, we produce the model on the basis a few object learning objects and then we use and improve the classifier when new data comes. This concept is still vibrant and may be used in the plethora of practical cases. Constructing such a system we have to realize that we have the limited resources (as memory and computational power) at our disposal. Nevertheless, during the exploitation of a classifier system the chosen characteristic of the classifier model may change within a time. This phenomena is called \textit{concept drift} and may lead the deep deterioration of the classification performance. This work deals with the data stream classification with the presence of \textit{concept drift}. We propose a novel classifier training algorithm based on the sliding windows approach which allows us to implement forgetting mechanism, i.e., that old objects came from outdated model will not be taken into consideration during the classifier updating and on the other hand we assume that all arriving examples can not be labeled, because we assume that we have a limited budget for labeling. We will employ active learning paradigm to choose an "interesting" object to be be labeled. The proposed approach has been evaluated on the basis of the computer experiments carried out on the data streams. Obtained results confirmed the usability of proposed method to the smoothly drifted data stream classification.

Abstract: Mass gatherings emerging both for specific occasions and spontaneously, are naturally associated with the risk of stampedes and crowd clashes that may trigger dramatic consequences in shorter or longer term perspectives. In order to address such issues the present paper suggests the application of the agent-based modeling approach to short-term predictions of future states of large congregates of people. The latter is of prime value for practitioners who seek to identify the potentially dangerous areas where the risk of stampede-induced injuries is assumed the highest based on the estimations of the crowd pressure at a given spot. In this paper, we outline the algorithm for generating forecasts and on its basis, propose a system of decision support. The test of system applicability has been performed based on the 2018 World Football Championship stadium use case. The object under investigation is expected to be put into operation in 2018.

Abstract: This paper focuses on techniques for dealing with imperfect data in a frame of early warning system (EWS). Despite the fact that data may be technically damaged by presenting noise, outliers or missing values, met-ocean simulation systems have to deal with them to provide data transaction between models, real time data assimilation, calibration, etc. In this context data quality-control becomes one of the most important parts of EWS. St. Petersburg FWS was considered as an example of EWS. Quality control in St. Petersburg FWS contains blocks of technical control, human mistakes control, statistical control of simulated fields, statistical control and restoration of measurements and control using alternative models. Domain specific quality control was presented as two types of procedures based on theoretically proved methods were applied. The first procedure is based on probabilistic model of dynamical system, where processes are spatially interrelated and could be implemented in a form of multivariate regression (MRM). The second procedure is based on principal component analysis extended for taking into account temporal relations in data set (ePCA).