The Experts below are selected from a list of 5205 Experts worldwide ranked by ideXlab platform

Ming-syan Chen - One of the best experts on this subject based on the ideXlab platform.

  • ICDM - Highly Parallel Sequential Pattern Mining on a Heterogeneous Platform
    2018 IEEE International Conference on Data Mining (ICDM), 2020
    Co-Authors: Yu-heng Hsieh, Chun-chieh Chen, Hong-han Shuai, Ming-syan Chen
    Abstract:

    Sequential pattern mining can be applied to various fields such as disease prediction and stock analysis. Many algorithms have been proposed for sequential pattern mining, together with acceleration methods. In this paper, we show that a Heterogeneous Platform with CPU and GPU is more suitable for sequential pattern mining than traditional CPU-based approaches since the support counting process is inherently succinct and repetitive. Therefore, we propose the PArallel SequenTial pAttern mining algorithm, referred to as PASTA, to accelerate sequential pattern mining by combining the merits of CPU and GPU computing. Explicitly, PASTA adopts the vertical bitmap representation of database to exploits the GPU parallelism. In addition, a pipeline strategy is proposed to ensure that both CPU and GPU on the Heterogeneous Platform operate concurrently to fully utilize the computing power of the Platform. Furthermore, we develop a swapping scheme to mitigate the limited memory problem of the GPU hardware without decreasing the performance. Finally, comprehensive experiments are conducted to analyze PASTA with different baselines. The experiments show that PASTA outperforms the state-of-the-art algorithms by orders of magnitude on both real and synthetic datasets.

  • Highly Parallel Sequential Pattern Mining on a Heterogeneous Platform
    2018 IEEE International Conference on Data Mining (ICDM), 2018
    Co-Authors: Yu-heng Hsieh, Chun-chieh Chen, Hong-han Shuai, Ming-syan Chen
    Abstract:

    Sequential pattern mining can be applied to various fields such as disease prediction and stock analysis. Many algorithms have been proposed for sequential pattern mining, together with acceleration methods. In this paper, we show that a Heterogeneous Platform with CPU and GPU is more suitable for sequential pattern mining than traditional CPU-based approaches since the support counting process is inherently succinct and repetitive. Therefore, we propose the PArallel SequenTial pAttern mining algorithm, referred to as PASTA, to accelerate sequential pattern mining by combining the merits of CPU and GPU computing. Explicitly, PASTA adopts the vertical bitmap representation of database to exploits the GPU parallelism. In addition, a pipeline strategy is proposed to ensure that both CPU and GPU on the Heterogeneous Platform operate concurrently to fully utilize the computing power of the Platform. Furthermore, we develop a swapping scheme to mitigate the limited memory problem of the GPU hardware without decreasing the performance. Finally, comprehensive experiments are conducted to analyze PASTA with different baselines. The experiments show that PASTA outperforms the state-of-the-art algorithms by orders of magnitude on both real and synthetic datasets.

Yu-heng Hsieh - One of the best experts on this subject based on the ideXlab platform.

  • ICDM - Highly Parallel Sequential Pattern Mining on a Heterogeneous Platform
    2018 IEEE International Conference on Data Mining (ICDM), 2020
    Co-Authors: Yu-heng Hsieh, Chun-chieh Chen, Hong-han Shuai, Ming-syan Chen
    Abstract:

    Sequential pattern mining can be applied to various fields such as disease prediction and stock analysis. Many algorithms have been proposed for sequential pattern mining, together with acceleration methods. In this paper, we show that a Heterogeneous Platform with CPU and GPU is more suitable for sequential pattern mining than traditional CPU-based approaches since the support counting process is inherently succinct and repetitive. Therefore, we propose the PArallel SequenTial pAttern mining algorithm, referred to as PASTA, to accelerate sequential pattern mining by combining the merits of CPU and GPU computing. Explicitly, PASTA adopts the vertical bitmap representation of database to exploits the GPU parallelism. In addition, a pipeline strategy is proposed to ensure that both CPU and GPU on the Heterogeneous Platform operate concurrently to fully utilize the computing power of the Platform. Furthermore, we develop a swapping scheme to mitigate the limited memory problem of the GPU hardware without decreasing the performance. Finally, comprehensive experiments are conducted to analyze PASTA with different baselines. The experiments show that PASTA outperforms the state-of-the-art algorithms by orders of magnitude on both real and synthetic datasets.

  • Highly Parallel Sequential Pattern Mining on a Heterogeneous Platform
    2018 IEEE International Conference on Data Mining (ICDM), 2018
    Co-Authors: Yu-heng Hsieh, Chun-chieh Chen, Hong-han Shuai, Ming-syan Chen
    Abstract:

    Sequential pattern mining can be applied to various fields such as disease prediction and stock analysis. Many algorithms have been proposed for sequential pattern mining, together with acceleration methods. In this paper, we show that a Heterogeneous Platform with CPU and GPU is more suitable for sequential pattern mining than traditional CPU-based approaches since the support counting process is inherently succinct and repetitive. Therefore, we propose the PArallel SequenTial pAttern mining algorithm, referred to as PASTA, to accelerate sequential pattern mining by combining the merits of CPU and GPU computing. Explicitly, PASTA adopts the vertical bitmap representation of database to exploits the GPU parallelism. In addition, a pipeline strategy is proposed to ensure that both CPU and GPU on the Heterogeneous Platform operate concurrently to fully utilize the computing power of the Platform. Furthermore, we develop a swapping scheme to mitigate the limited memory problem of the GPU hardware without decreasing the performance. Finally, comprehensive experiments are conducted to analyze PASTA with different baselines. The experiments show that PASTA outperforms the state-of-the-art algorithms by orders of magnitude on both real and synthetic datasets.

John E Bowers - One of the best experts on this subject based on the ideXlab platform.

  • Heterogeneous integration of lithium niobate and silicon nitride waveguides for wafer scale photonic integrated circuits on silicon
    Optics Letters, 2017
    Co-Authors: Lin Chang, Martin H P Pfeiffer, Nicolas Volet, Michael Zervas, Jon Peters, Costanza L Manganelli, Eric J Stanton, Yifei Li, Tobias J Kippenberg, John E Bowers
    Abstract:

    An ideal photonic integrated circuit for nonlinear photonic applications requires high optical nonlinearities and low loss. This work demonstrates a Heterogeneous Platform by bonding lithium niobate (LN) thin films onto a silicon nitride (Si3N4) waveguide layer on silicon. It not only provides large second- and third-order nonlinear coefficients, but also shows low propagation loss in both the Si3N4 and the LN-Si3N4 waveguides. The tapers enable low-loss-mode transitions between these two waveguides. This Platform is essential for various on-chip applications, e.g., modulators, frequency conversions, and quantum communications.

  • iii v silicon photonics for on chip and intra chip optical interconnects
    Laser & Photonics Reviews, 2010
    Co-Authors: Gunther Roelkens, B. R. Koch, Liu Liu, Di Liang, Richard Jones, Alexander W Fang, John E Bowers
    Abstract:

    In this paper III-V on silicon-on-insulator (SOI) het- erogeneous integration is reviewed for the realization of near infrared light sources on a silicon waveguide Platform, suitable for inter-chip and intra-chip optical interconnects. Two bonding technologies are used to realize the III-V/SOI integration: one based on molecular wafer bonding and the other based on DVS- BCB adhesive wafer bonding. The realization of micro-disk lasers, Fabry-Perot lasers, DFB lasers, DBR lasers and mode- locked lasers on the III-V/SOI material Platform is discussed. Artist impression of a multi-wavelength laser based on micro- disk cavities realized on a III-V/SOI Heterogeneous Platform and a microscope image of a realized structure.

Antonin Hermanek - One of the best experts on this subject based on the ideXlab platform.

  • Heterogeneous Platform for Stream Based Applications on FPGAs
    2011 21st International Conference on Field Programmable Logic and Applications, 2011
    Co-Authors: Jan Kloub, Tomas Mazanec, Antonin Hermanek
    Abstract:

    The complexity of embedded systems is ever increasing, to support wide range of application domains. Dedicated hardware provides improved performance but with limited resources reusability and application scalability. Using higher abstraction methods for hardware development shortens the time to market. The concept of hardware objects (HWO) allows for more efficient hardware resource reuse and scalability of the target application. A single task of the application can be mapped on an appropriate type of hardware structure. A single hardware structure can be shared in time by several application tasks, multiple instances of the same HW structure can be used for concurrent computation. Uniform hardware object structure shortens design time and simplifies design of target system independently on application domain. The system could contain a Heterogeneous set of HWO instances to satisfy all application domains.

Hong-han Shuai - One of the best experts on this subject based on the ideXlab platform.

  • ICDM - Highly Parallel Sequential Pattern Mining on a Heterogeneous Platform
    2018 IEEE International Conference on Data Mining (ICDM), 2020
    Co-Authors: Yu-heng Hsieh, Chun-chieh Chen, Hong-han Shuai, Ming-syan Chen
    Abstract:

    Sequential pattern mining can be applied to various fields such as disease prediction and stock analysis. Many algorithms have been proposed for sequential pattern mining, together with acceleration methods. In this paper, we show that a Heterogeneous Platform with CPU and GPU is more suitable for sequential pattern mining than traditional CPU-based approaches since the support counting process is inherently succinct and repetitive. Therefore, we propose the PArallel SequenTial pAttern mining algorithm, referred to as PASTA, to accelerate sequential pattern mining by combining the merits of CPU and GPU computing. Explicitly, PASTA adopts the vertical bitmap representation of database to exploits the GPU parallelism. In addition, a pipeline strategy is proposed to ensure that both CPU and GPU on the Heterogeneous Platform operate concurrently to fully utilize the computing power of the Platform. Furthermore, we develop a swapping scheme to mitigate the limited memory problem of the GPU hardware without decreasing the performance. Finally, comprehensive experiments are conducted to analyze PASTA with different baselines. The experiments show that PASTA outperforms the state-of-the-art algorithms by orders of magnitude on both real and synthetic datasets.

  • Highly Parallel Sequential Pattern Mining on a Heterogeneous Platform
    2018 IEEE International Conference on Data Mining (ICDM), 2018
    Co-Authors: Yu-heng Hsieh, Chun-chieh Chen, Hong-han Shuai, Ming-syan Chen
    Abstract:

    Sequential pattern mining can be applied to various fields such as disease prediction and stock analysis. Many algorithms have been proposed for sequential pattern mining, together with acceleration methods. In this paper, we show that a Heterogeneous Platform with CPU and GPU is more suitable for sequential pattern mining than traditional CPU-based approaches since the support counting process is inherently succinct and repetitive. Therefore, we propose the PArallel SequenTial pAttern mining algorithm, referred to as PASTA, to accelerate sequential pattern mining by combining the merits of CPU and GPU computing. Explicitly, PASTA adopts the vertical bitmap representation of database to exploits the GPU parallelism. In addition, a pipeline strategy is proposed to ensure that both CPU and GPU on the Heterogeneous Platform operate concurrently to fully utilize the computing power of the Platform. Furthermore, we develop a swapping scheme to mitigate the limited memory problem of the GPU hardware without decreasing the performance. Finally, comprehensive experiments are conducted to analyze PASTA with different baselines. The experiments show that PASTA outperforms the state-of-the-art algorithms by orders of magnitude on both real and synthetic datasets.