Filters








339 Hits in 5.6 sec

Reports of the AAAI 2010 Conference Workshops

David W. Aha, Mark Boddy, Vadim Bulitko, Artur S. D'Avila Garcez, Prashant Doshi, Stefan Edelkamp, Christopher Geib, Piotr Gmytrasiewicz, Robert P. Goldman, Pascal Hitzler, Charles Isbell, Darsana Josyula (+20 others)
2010 The AI Magazine  
The workshop included two invited talks. Hector Geffner spoke on plan recognition as planning.  ...  Dragos Margineantu gave an invited talk reporting on years of experience at Boeing and DARPA on testing adaptive and learning models.  ...  The aim and scope of this one-day workshop were similar to an independent and bigger symposium series called SARA (Symposium on Abstraction, Reformulation, and Approximation).  ... 
doi:10.1609/aimag.v31i4.2318 fatcat:niss5pwei5hytcof433r45okja

HL-LHC Analysis With ROOT [article]

Axel Naumann, Philippe Canal, Enric Tejedor, Enrico Guiraud, Lorenzo Moneta, Bertrand Bellenot, Olivier Couet, Alja Mrak Tadel, Matevz Tadel, Sergey Linev, Javier Lopez Gomez, Jonas Rembser (+5 others)
2022 arXiv   pre-print
It is evolving since 25 years, now providing the storage format for more than one exabyte of data; virtually all high energy physics experiments use ROOT.  ...  same straight-forward interfaces from C++ and Python.  ...  RDataFrame-based training on GPUs is completed by 2023 and ML-optimized RNTuple-to-GPU transfer by 2025.  ... 
arXiv:2205.06121v1 fatcat:f7rk3km77feifmiqz6dia4sq5y

Megatron-LM: Training Multi-Billion Parameter Language Models Using Model Parallelism [article]

Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper, Bryan Catanzaro
2020 arXiv   pre-print
Our BERT model achieves SOTA results on the RACE dataset (90.9% compared to SOTA accuracy of 89.4%).  ...  We illustrate this approach by converging transformer based models up to 8.3 billion parameters using 512 GPUs.  ...  one GPU.  ... 
arXiv:1909.08053v4 fatcat:hqdaodavsfb7fcobxrsodjkqam

Overview of the IBM Neural Computer Architecture [article]

Pritish Narayanan, Charles E. Cox, Alexis Asseman, Nicolas Antoine, Harald Huels, Winfried W. Wilcke, Ahmet S. Ozcan
2020 arXiv   pre-print
It consists of hundreds of programmable nodes, primarily based on Xilinx's Field Programmable Gate Array (FPGA) technology. The nodes are interconnected in a scalable 3d mesh topology.  ...  computations performed and in the available modes of communication, enabling new machine intelligence approaches and learning strategies not well suited to the matrix manipulation/SIMD libraries that GPUs  ...  Narayanan has presented one prior keynote (International Memory Workshop 2017) and a tutorial session (Device Research Conference 2017), in addition to several invited talks.  ... 
arXiv:2003.11178v1 fatcat:53sbuejfrbaqrocmok5e7l3gdy

D8.4.2: Final Refactoring Report

Claudio Gheller
2014 Zenodo  
, in order to effectively run on coming generations of supercomputing architectures, optimally exploiting their innovative features.  ...  The main outcomes of the overall WP8 work are discussed and best practices for the development of scientific numerical applications on HPC systems are presented.  ...  2014) with the talk "The Ramses code for GPU accelerated numerical cosmology" (Gheller).  ... 
doi:10.5281/zenodo.6572435 fatcat:2zxmz27swjebvf77uimfdn2tim

Communication-free and Parallel Simulation of Neutral Biodiversity Models [article]

Momo Langenstein
2021 arXiv   pre-print
The GPU implementation further outperforms all algorithms on the CPU by a factor ranging from ∼ 2 to ∼ 80, depending on the model parameterisation and the analysis that is performed.  ...  We evaluate our parallelisation approach by comparing three traditional simulation algorithms against a CPU and GPU implementation of our Independent algorithm.  ...  The root process uses immediate probing receives to capture and combine the progress updates, which it then forwards live to the reporter.  ... 
arXiv:2108.05815v1 fatcat:tnqnti3s6vaqxpljdkttq2l3wu

A high-level characterisation and generalisation of communication-avoiding programming techniques [article]

Tobias Weinzierl
2019 arXiv   pre-print
Today's hardware's explosion of concurrency plus the explosion of data we build upon in both machine learning and scientific simulations have multifaceted impact on how we write our codes.  ...  To do so, it has to manage to move the compute data into the compute facilities on time.  ...  Its idea is that we bring all the computations forward that feed into computations on another compute entity.  ... 
arXiv:1909.10853v2 fatcat:72wuro6bhjhmfnilmvulwx4eqm

CERN openlab annual report 2018

Purcell, Di Meglio, Carminati, Girone, Rademakers, Gunne, Baechle
2019 Zenodo  
On 14 and 16 August, the students presented their work in two dedicated public "lightning talk" sessions.  ...  Progress in 2018 In accordance with the concept of data-parallel distributed learning, we trained the GAN on a total of twelve GPUs, distributed over the three nodes that comprise the test Power cluster  ... 
doi:10.5281/zenodo.3234404 fatcat:qmxmccdrzfd55p5nmbzijp23i4

Future Automotive HW/SW Platform Design (Dagstuhl Seminar 19502)

Dirk Ziegenbein, Selma Saidi, Xiaobo Sharon Hu, Sebastian Steinhorst
2020 Dagstuhl Reports  
and practitioners from academia and industry to discuss key industrial challenges, existing solutions and research directions in the design of future automotive HW/SW platforms, particularly focusing on  ...  Machine Learning in Cyber-Physical Systems We then started talking about uncertainty and the guarantees that can be provided either with machine learning or with other techniques, linking this to model  ...  Here the focus is on non-functional performance models of real-world systems, ranging from engine control [2] to automated driving [3].  ... 
doi:10.4230/dagrep.9.12.28 dblp:journals/dagstuhl-reports/ZiegenbeinSHS19 fatcat:gvrg2tj5enh3rnwtvquch54vo4

Proceedings of the 2020 Connecting the Dots Workshop

David Lange
2020 Zenodo  
The Connecting The Dots workshop series brings together experts on track reconstruction and other problems involving pattern recognition in sparsely sampled data.  ...  While the main focus will be on High Energy Physics (HEP) detectors, the Connecting The Dots workshop is intended to be inclusive across other scientific disciplines wherever similar problems or solutions  ...  ACKNOWLEDGEMENTS ACKNOWLEDGEMENTS Part of this work was conducted at "iBanks", the AI GPU cluster at Caltech.  ... 
doi:10.5281/zenodo.4088760 fatcat:kulhvq3t5fglnimqawge7lnt4q

Action Space Shaping in Deep Reinforcement Learning [article]

Anssi Kanervisto, Christian Scheller, Ville Hautamäki
2020 arXiv   pre-print
Currently, this is mostly done based on intuition, with little systematic research supporting the design decisions.  ...  In this work, we aim to gain insight on these action space modifications by conducting extensive experiments in video-game environments.  ...  https://github.com/minerllabs/baselines/tree/master/general/chainerrl 2 https://slideslive.com/38922867/invited-talk-reinforcement-learning-of-the-obstacle-tower-challenge  ... 
arXiv:2004.00980v2 fatcat:5v22qoedireprkbfkmevys5rsi

Roadmap for Reliable Ensemble Forecasting of the Sun-Earth System [article]

Gelu Nita, Rafal Angryk, Berkay Aydin, Juan Banda, Tim Bastian, Tom Berger, Veronica Bindi, Laura Boucheron, Wenda Cao, Eric Christian, Georgia de Nolfo, Edward DeLuca (+33 others)
2018 arXiv   pre-print
The workshop combined a set of plenary sessions featuring invited introductory talks and workshop progress reports, interleaved with a set of breakout sessions focused on specific topics of interest.  ...  The authors of this report met on 28-30 March 2018 at the New Jersey Institute of Technology, Newark, New Jersey, for a 3-day workshop that brought together a group of data providers, expert modelers,  ...  The data providers do need to provide tools, such as spectral synthesis methods, that allow people to use the data effectively for other purposes, including forward modeling the observations for one-to-one  ... 
arXiv:1810.08728v2 fatcat:i74ti544a5aiblsvitluop6buu

Report of the Third Global Experimentation for Future Internet (GEFI 2018) Workshop [article]

Mark Berman, Timur Friedman, Abhimanyu Gosain, Kate Keahey, Rick McGeer, Ingrid Moerman, Akihiro Nakao, Lucas Nussbaum, Kristin Rauschenbach, Violet Syrotiuk, Malathi Veeraraghavan, Naoaki Yamanaka
2019 arXiv   pre-print
This talk focuses on the challenges of integrating and harmonizing virtualization across multiple implementation technologies (CPU, GPU and FPGA).  ...  progress in our field.  ... 
arXiv:1901.02929v1 fatcat:nyzqaop4tjfo3djh4zmpakrepa

2019 DAC Roundtable

Giovanni De Micheli, Antun Domic, Massimiliano Di Ventra, Martin Roettler, Jason Cong
2020 IEEE design & test  
They have to answer this question all the time: "Shall I use GPU to accelerate or I use an FPGA to accelerate." We have made some good progress by now.  ...  Massimiliano Di Ventra: Thank you Nanni, for inviting me to this panel and thank you all for being here. So, I'll be talking about a new computing paradigm, we call mem (memory) computing.  ... 
doi:10.1109/mdat.2020.2968279 fatcat:p4mnyo7z6zhrbdihxlafdjfmvm

Browser-based Harnessing of Voluntary Computational Power

Tomasz Fabisiak, Arkadiusz Danilecki
2017 Foundations of Computing and Decision Sciences  
One way to utilize this public resource is via world wide web, where users can share their resources using nothing more except their browsers.  ...  From 1500 users invited, 274 sent back useable responses.  ...  By a job we understand a running application solving one particular problem instance (with particular set of parameters). One problem may concurrently be solved by several different jobs.  ... 
doi:10.1515/fcds-2017-0001 fatcat:mvtpvz4txvgphcfaqcdsfw6mgy
« Previous Showing results 1 — 15 out of 339 results