2.5 / August 3, 2018
(5.0/5) (12)

Description

This application lets you participate in experiment crowdsourcingto help researchers solve complex problems using open-sourceCollective Knowledge Engine available athttp://github.com/ctuning/ck . You can see all public crowd resultsat http://cknowledge.org/repo! You can also check out other ways toparticipate in experiment crowdsourcing (using laptops and datacenters) athttps://github.com/ctuning/ck/wiki/Crowdsourcing-optimization.Finally, you can get open sources of this Android app athttps://github.com/ctuning/crowdsource-experiments-using-android-devices. This development is coordinated by the non-profit cTuningfoundation: http://cTuning.org . Please, check out our otherinitiatives for open science athttp://cTuning.org/reproducibility-wiki .We have been strugglingwith a lack of computational resources and diverse workloads/datasets/hardware for our own research to make faster, smaller, morepower efficient reliable self-tuning software and hardware for morethan a decade! Indeed, computer systems are becoming veryinefficient - it is nowadays not uncommon to obtain 10x speedups,2x size reduction and 40% energy reduction for popular algorithms(DNN, BLAS, video processing) on latest hardware using so-calledautotuning of various algorithm parameters and compileroptimizations. However, this process is extremely time consumingdue to very large design and optimization spaces. With the help ofthe non-profit cTuning foundation, we have developed this opensource Collective Knowledge technology (CK) to let the communityshare workloads, data sets, tools and experimental workflows in anopen CK format via GitHub or BitBucket, crowdsource experimentsacross numerous devices provided by volunteers, classify solutionson the fly (active learning), apply predictive analytics, exchangeknowledge and reproduce results. Our current shared experimentalscenarios including GCC and LLVM compiler flag tuning, and we planto add automatic OpenCL/CUDA tuning, scalability testing, compilerbug detection and numerical stability analysis (see CK-basedinteractive report: http://cknowledge.org/interactive-report )!This application simply queries any available public CK server,obtains experimental pack for your device (such as several binariesof a shared benchmark with different optimizations), runs it onyour mobile device with some random data set and send statisticsback to CK server to be processed using active learning andpredictive analytics. Your participation supports our open scienceand reproducible research initiatives such as Artifact Evaluationat various conferences (sharing experimental workflows with allrelated artifacts and results along with publications in areproducible and reusable way to be validated by the community):*http://cTuning.org/ae* http://adapt-workshop.org*http://ctuning.org/reproducibility-wikiYou may read more about ourlong-term vision in the following publications:*http://arxiv.org/abs/1506.06256* http://bit.ly/ck-date16*http://hal.inria.fr/hal-01054763* http://arxiv.org/abs/1406.4020*https://hal.inria.fr/inria-00436029We would like to thank ourcurrent supporters:* http://cTuning.org (non-profit foundation)*http://dividiti.comWe have many new ideas and possible projectsbased on this technology. If you are interested to know more oreven participate in our activities, join our growing academic andindustrial consortium, optimize or share your realistic workloads,implement new experiment crowdsourcing scenarios, provide morecomputational resources, do not hesitate to get in touch with theauthor (Grigori Fursin, http://fursin.net) or via our publicmailing list:*https://groups.google.com/forum/#!forum/collective-knowledgeThankyou very much for participating in experiment crowdsourcing andenabling open science!

App Information Crowdsourcing experiments

  • App Name
    Crowdsourcing experiments
  • Package Name
    openscience.crowdsource.experiments
  • Updated
    August 3, 2018
  • File Size
    1.7M
  • Requires Android
    Android 2.3.3 and up
  • Version
    2.5
  • Developer
    Grigori Fursin
  • Installs
    500+
  • Price
    Free
  • Category
    Tools
  • Developer
  • Google Play Link
Grigori Fursin Show More...
Collective Mind Node 1.8.redirect APK
THIS APPLICATION IS NOW DEPRECATED! Download brand new applicationfor universal experiment crowdsourcing (including GCC and LLVMcrowd-tuning) here:*https://play.google.com/store/apps/details?id=openscience.crowdsource.experimentsThenew version is based on open Collective Knowledge research SDK:*http://github.com/ctuning/ckYou can see all public optimizationresults here:*http://cTuning.org/crowd-results=======================================================================Tuningapplications for rapidly evolving hardware or improvingoptimization heuristic of a new compiler is becoming complex,ad-hoc, costly and error prone due to an enormous number ofavailable optimization choices combined with the complexinteractions between all software and hardware components. Since2008, we are developing a methodology and supporting technology tocrowdsource compiler auto-tuning (iterative compilation) acrossmobile devices and make it practical using our open-source cTuningtechnology (http://cTuning.org). In our approach, multi-objectivecodelet, program and architecture tuning to balance performance,power consumption, compilation time, code size and any otherimportant metric is continuously distributed among multipleparticipants. We hope that our approach will help to automaticallyimprove default optimization heuristics of latest compilersincluding GCC, LLVM, Open64 and any other, or predict optimizationsfor new programs using collected knowledge and machine learning tomaximize utilization of modern computer devices. Interestingly, wealso hope that it will help to improve academic research and switchfocus from publishing numerous non-reproducible papers to sharingresearch material and experimental results that can becollaboratively validated and improved by the community. Thisapplication, supported by non-profit cTuning foundation, iscurrently a research proof-of-concept to support and validate ouridea. It helps us collect various performance statistics fromparticipants to validate machine learning techniques. This is anon-going project, so please be patient! Feedback and suggestionsare welcome!More info:* Vision publications:http://arxiv.org/abs/1506.06256 ,https://hal.inria.fr/hal-01054763* Community website:http://cTuning.org* Live Repositories: http://c-mind.org/repo* Ourrelated artifact evaluation initiative (to start collecting morepublic benchmarks and data sets): http://cTuning.org/ae* Authorpage: http://fursin.netWe are currently raising further funding tocontinue this project and related public research activities. Ifyou are interested, do not hesitate to contact Grigori Fursin(gfursin@gmail.com) for more details!
Research paper title generator 1.3 APK
News: * 2016.02.29: We released new CK-based engine to crowdsourceexperiments and focus open, collaborative and reproducible sciencerather than just publishing: http://cTuning.org/crowd-results *2015.11.11: CK engine for paper title generator has been improved(adding probabilities to select parts of the title) Academicpromotion is often based on a number of publications rather thanquality, novelty and reproducibility of the research results.Therefore, we created this application to automatically generatetitles for possible research papers in computer engineering to helpour fellow researchers ;) ... Alternatively, you may be interestedto check out our long-term initiatives to enable trulycollaborative, systematic and reproducible computer systems'research and experimentation with a focus on artifact sharing andpublic discussions: * Artifact evaluation for conference andjournals (sharing and validating code and data along withpublications): ** http://ctuning.org/ae * Open SDK forcollaborative and reproducible experimentation & interactivearticles: ** http://github.com/ctuning/ck **http://cknowledge.org/repo * Our examples of live and reproduciblearticles (gradually updated based on the community feedback): **http://cknowledge.org/repo/web.php?wcid=29db2248aba45e59:cd11e3a188574d80**http://cknowledge.org/repo/web.php?wcid=29db2248aba45e59:6f40bc99c4f7df58* New publication model with community-driven reviewing of papersand artifacts: ** http://adapt-workshop.org **http://arxiv.org/abs/1406.4020 **http://ctuning.org/reproducibility-wiki
Crowdsourcing experiments 2.5 APK
This application lets you participate in experiment crowdsourcingto help researchers solve complex problems using open-sourceCollective Knowledge Engine available athttp://github.com/ctuning/ck . You can see all public crowd resultsat http://cknowledge.org/repo! You can also check out other ways toparticipate in experiment crowdsourcing (using laptops and datacenters) athttps://github.com/ctuning/ck/wiki/Crowdsourcing-optimization.Finally, you can get open sources of this Android app athttps://github.com/ctuning/crowdsource-experiments-using-android-devices. This development is coordinated by the non-profit cTuningfoundation: http://cTuning.org . Please, check out our otherinitiatives for open science athttp://cTuning.org/reproducibility-wiki .We have been strugglingwith a lack of computational resources and diverse workloads/datasets/hardware for our own research to make faster, smaller, morepower efficient reliable self-tuning software and hardware for morethan a decade! Indeed, computer systems are becoming veryinefficient - it is nowadays not uncommon to obtain 10x speedups,2x size reduction and 40% energy reduction for popular algorithms(DNN, BLAS, video processing) on latest hardware using so-calledautotuning of various algorithm parameters and compileroptimizations. However, this process is extremely time consumingdue to very large design and optimization spaces. With the help ofthe non-profit cTuning foundation, we have developed this opensource Collective Knowledge technology (CK) to let the communityshare workloads, data sets, tools and experimental workflows in anopen CK format via GitHub or BitBucket, crowdsource experimentsacross numerous devices provided by volunteers, classify solutionson the fly (active learning), apply predictive analytics, exchangeknowledge and reproduce results. Our current shared experimentalscenarios including GCC and LLVM compiler flag tuning, and we planto add automatic OpenCL/CUDA tuning, scalability testing, compilerbug detection and numerical stability analysis (see CK-basedinteractive report: http://cknowledge.org/interactive-report )!This application simply queries any available public CK server,obtains experimental pack for your device (such as several binariesof a shared benchmark with different optimizations), runs it onyour mobile device with some random data set and send statisticsback to CK server to be processed using active learning andpredictive analytics. Your participation supports our open scienceand reproducible research initiatives such as Artifact Evaluationat various conferences (sharing experimental workflows with allrelated artifacts and results along with publications in areproducible and reusable way to be validated by the community):*http://cTuning.org/ae* http://adapt-workshop.org*http://ctuning.org/reproducibility-wikiYou may read more about ourlong-term vision in the following publications:*http://arxiv.org/abs/1506.06256* http://bit.ly/ck-date16*http://hal.inria.fr/hal-01054763* http://arxiv.org/abs/1406.4020*https://hal.inria.fr/inria-00436029We would like to thank ourcurrent supporters:* http://cTuning.org (non-profit foundation)*http://dividiti.comWe have many new ideas and possible projectsbased on this technology. If you are interested to know more oreven participate in our activities, join our growing academic andindustrial consortium, optimize or share your realistic workloads,implement new experiment crowdsourcing scenarios, provide morecomputational resources, do not hesitate to get in touch with theauthor (Grigori Fursin, http://fursin.net) or via our publicmailing list:*https://groups.google.com/forum/#!forum/collective-knowledgeThankyou very much for participating in experiment crowdsourcing andenabling open science!
Loading...