16,667 Hits in 5.9 sec

Performance Analysis of Deep Learning Workloads on a Composable System [article]

Kauotar El Maghraoui and Lorraine M. Herger and Chekuri Choudary and Kim Tran and Todd Deshane and David Hanson
2021 arXiv   pre-print
Our experimental evaluations on the composable system give insights into how the system works and evaluates the impact of various resource aggregations and reconfigurations on representative deep learning  ...  This design provides flexibility to serve a variety of workloads and provides a dynamic co-design platform that allows experiments and measurements in a controlled manner.  ...  ACKNOWLEDGEMENTS The authors would like to thank the following colleagues from IBM Research: I-Hsin Chung and Paul Crumley for their technical assistance with early design and experimentation on the composable  ... 
arXiv:2103.10911v1 fatcat:6sv7ndzhwbhw7mm3fc5fzfp4le

A Comparative Study of Learning Techniques with Convolutional Neural Network Based on HPC-Workload Dataset

Anupong Banjongkan, School of Computer Engineering, Suranaree University of Technology (SUT), Thailand, Watthana Pongsena, Nittaya Kerdprasop, Kittisak Kerdprasop
2020 International Journal of Machine Learning and Computing  
Therefore, managing the jobs or job scheduling is very important since it involves the overall system efficiency. The analysis of an HPC-workload log file is a solution to improve system efficiency.  ...  A huge amount of jobs from a large group of users prefer to complete their jobs in this kind of system.  ...  ACKNOWLEDGMENT The authors would like to acknowledge the "National e-Science Infrastructure Consortium" of NECTEC for providing the HPC-workload as a dataset that we use in this research (URL:  ... 
doi:10.18178/ijmlc.2020.10.1.891 fatcat:jqzfvbyjdfcz7bu2o6oupgdxci

HPC AI500: A Benchmark Suite for HPC AI Systems [article]

Zihan Jiang, Wanling Gao, Lei Wang, Xingwang Xiong, Yuchen Zhang, Xu Wen, Chunjie Luo, Hainan Ye, Yunquan Zhang, Shengzhong Feng, Kenli Li, Weijia Xu (+1 others)
2019 arXiv   pre-print
In recent years, with the trend of applying deep learning (DL) in high performance scientific computing, the unique characteristics of emerging DL workloads in HPC raise great challenges in designing,  ...  We propose a set of metrics for comprehensively evaluating the HPC AI systems, considering both accuracy, performance as well as power and cost.  ...  Acknowledgment This work is supported by the Standardization Research Project of Chinese Academy of Sciences No.BZ201800001.  ... 
arXiv:1908.02607v3 fatcat:bv4aia7yrbglrjtg47zetxmsnu

Autonomic Workload Performance Modeling for Large-Scale Databases and Data Warehouses through Deep Belief Network with Data Augmentation using Conditional Generative Adversarial Networks

Nusrat Shaheen, Basit Raza, Ahmad Raza Shahid, Ahmad Kamran Malik
2021 IEEE Access  
. 4) Deep Belief Network (DBN) DBN is a generative graphical model that is composed of multiple layers.  ...  PROPOSED OPTIMIZED GAN-BASED DEEP LEARNING (OGDL) MODEL We proposed an optimized GAN-based deep learning model for workload performance tuning.  ...  Her areas of research are databases, data mining, machine learning and autonomic workload management.  ... 
doi:10.1109/access.2021.3096039 fatcat:d5ogtlrtgrg2riivmmlrcuv7z4

Continuous EEG Decoding of Pilots' Mental States using Multiple Feature Block-based Convolutional Neural Network

Dae-Hyeok Lee, Ji-Hoon Jeong, Kiduk Kim, Baek-Woon Yu, Seong-Whan Lee
2020 IEEE Access  
Also, in the pseudo-online analysis, we obtained the detection accuracy of 0.72 (±0.20), 0.72 (±0.27), and 0.61 (±0.18) for fatigue, workload, and distraction, respectively.  ...  In the offline analysis, we confirmed the classification accuracy of 0.75 (±0.04).  ...  DISCUSSION In this paper, we demonstrated the feasibility of continuous decoding of various mental states (fatigue, workload, distraction, and the normal state) based on a deep learning method in a pilot  ... 
doi:10.1109/access.2020.3006907 fatcat:aggigogxurdndhmxswiibe7izy

Towards Robust Neuroadaptive HCI

Aurélien Appriou, Andrzej Cichocki, Fabien Lotte
2018 Extended Abstracts of the 2018 CHI Conference on Human Factors in Computing Systems - CHI '18  
Our results suggested that a shallow Convolutional Neural Network obtained the best performance in both conditions, outperforming state-of-the-art methods on the used data sets.  ...  This paper thus studies promising modern machine learning algorithms, including Riemannian geometry-based methods and Deep Learning, to estimate workload from EEG signals.  ...  Acknowledgements: This work received support from the European Research Council (grant ERC-2016-STG-714567) and the Japanese Society for the Promotion of Science (JSPS).  ... 
doi:10.1145/3170427.3188617 dblp:conf/chi/AppriouCL18 fatcat:kyglz2xlq5ck3grqmacp7iqxcm

Identifying Dwarfs Workloads in Big Data Analytics [article]

Wanling Gao, Chunjie Luo, Jianfeng Zhan, Hainan Ye, Xiwen He, Lei Wang, Yuqing Zhu, Xinhui Tian
2015 arXiv   pre-print
How can we construct a benchmark suite using a minimum set of units of computation to represent diversity of big data analytics workloads?  ...  One dwarf represents one unit of computation, and big data workloads are decomposed into one or more dwarfs.  ...  , deep learning, natural language processing, etc. 3) Large numbers of algorithms and the variants of these algorithms aggravate the difficulty of abstraction. 4) Not like traditional database systems,  ... 
arXiv:1505.06872v1 fatcat:5b7robneurfnlc75n7xe2a7oou

Location-aware Machine Learning Mechanisms (D4.3)

Roberto Bifulco, Giuseppe Siracusano
2019 Zenodo  
This deliverable comprehends the outcome of T4.4 and will report the implementation of the location-aware machine learning mechanisms.  ...  Deep Learning workload analysis In this section we analyse the processing required by NN during inference.  ...  Conclusion In this report, we presented an extended analysis of Deep Learning workloads, devising techniques to run neural networks in split mode and on heterogeneous devices.  ... 
doi:10.5281/zenodo.3752904 fatcat:32u3mz6e2zgkdduyemjodnceze

HPTMT Parallel Operators for High Performance Data Science Data Engineering [article]

Vibhatha Abeykoon, Supun Kamburugamuve, Chathura Widanage, Niranda Perera, Ahmet Uyar, Thejaka Amila Kanewala, Gregor von Laszewski, Geoffrey Fox
2021 arXiv   pre-print
They are comprised of a rich set of sub-domains such as data engineering, deep learning, and machine learning.  ...  This paper elaborates and illustrates this architecture using an end-to-end application with deep learning and data engineering parts working together.  ...  The uniqueness of this approach is the composition of a data engineering workload followed by a deep learning workload written in PyTorch.  ... 
arXiv:2108.06001v1 fatcat:qbnz7lk4mffc5mccq3xzntxkym

Cognitive Computing Safety: The New Horizon for Reliability / The Design and Evolution of Deep Learning Workloads

Yuhao Zhu, Vijay Janapa Reddi, Robert Adolf, Saketh Rama, Brandon Reagen, Gu-Yeon Wei, David Brooks
2017 IEEE Micro  
Among all the issues, a rapidly rising and critical challenge to address is the practice of safe cognitive computingthat is, how to architect machine learning-based systems to be robust against uncertainty  ...  To push such cognitive services closer to reality, recent research has focused extensively on improving the performance, energy efficiency, privacy, and security of cognitive computing platforms.  ...  Many of the primitives were implemented in only one library, and most of the analysis tools constructed to characterize the Fathom workloads would have been substantially more difficult.  ... 
doi:10.1109/mm.2017.2 fatcat:kgopzne2r5h75ckxwpabkxusr4

Deep Reinforcement Agent for Scheduling in HPC [article]

Yuping Fan, Zhiling Lan, Taylor Childers, Paul Rich, William Allcock, Michael E. Papka
2021 arXiv   pre-print
However, the increasing complexity of computing systems and the highly dynamic nature of application workloads have placed tremendous burden on manually designed and tuned scheduling heuristics.  ...  Existing cluster scheduling heuristics are developed by human experts based on their experience with specific HPC systems and workloads.  ...  The learning rate α is set to 0.001. We use 100 jobsets composed of 320,000 jobs for DRAS training on Theta.  ... 
arXiv:2102.06243v2 fatcat:rz7fzegm45fhbaprkiuvptyxda

A Workflow-Forecast Approach To The Task Scheduling Problem In Distributed Computing Systems [article]

Andrey Gritsenko
2013 arXiv   pre-print
The aim of this paper is to provide a description of deep-learning-based scheduling approach for academic-purpose high-performance computing systems.  ...  The core of this approach is to predict the future workflow of the system depending on the previously submitted tasks using deep learning algorithm.  ...  The Lomonosov workload was gained gathering information on submitted tasks via monitoring the official site that shows current state of the cluster system Lomonosov. [16] .  ... 
arXiv:1310.1553v1 fatcat:xuai6qmbjjctna46ct7fgkagba

Enhancing the Analysis of Software Failures in Cloud Computing Systems with Deep Learning [article]

Domenico Cotroneo, Luigi De Simone, Pietro Liguori, Roberto Natella
2021 arXiv   pre-print
The approach leverages Deep Embedded Clustering (DEC), a family of unsupervised clustering algorithms based on deep learning, which uses an autoencoder to optimize data dimensionality and inter-cluster  ...  for deep domain knowledge and reducing the effort to perform the analysis.  ...  Acknowledgements This work has been partially supported by the University of Naples Federico II in the frame of the Programme F.R.A., project id OSTAGE.  ... 
arXiv:2106.15182v1 fatcat:n77ysqsxybhybnuomnre5edsv4

Machine Learning in Manufacturing Ergonomics: Recent Advances, Challenges, and Opportunities

Sujee Lee, Li Liu, Robert Radwin, Jingshan Li
2021 IEEE Robotics and Automation Letters  
The rapid development of machine learning (ML) technology has introduced substantial impact on ergonomics research in manufacturing.  ...  ergonomics, and manufacturing systems perspectives.  ...  classifiers, and observes that multiple sensors and ensemble deep learning methods can achieve superior performances. 2) Motion Analysis Through Videos: In addition to sensing data, cameras and videos  ... 
doi:10.1109/lra.2021.3084881 fatcat:hqdsvn3v7vcepfv5orvqlp57lu


Shi Dong, David Kaeli
2017 Proceedings of the General Purpose GPUs on - GPGPU-10  
One commonly used class of deep learning techniques is deep neural networks (DNNs). ey are composed of a massive number of arti cial neurons and many hidden layers.  ...  Deep learning algorithms have been growing in popularity in the machine learning community based on their ability to accurately perform clustering and classi cation in a number of domains.  ...  However, if we want to tune performance of DNN computations on a GPU platform, there is a void of tools available. While there are a number of existing deep learning frameworks (e.g.  ... 
doi:10.1145/3038228.3038239 dblp:conf/ppopp/DongK17 fatcat:kmjirkvchbaizg5dfpwrmh7vey
« Previous Showing results 1 — 15 out of 16,667 results