Upgrade to Pro — share decks privately, control downloads, hide ads and more …

reinvent-ml-mini-con

ryo nakamaru
December 09, 2016

 reinvent-ml-mini-con

JAWS-UG AI 支部 #2 での登壇資料です

ryo nakamaru

December 09, 2016
Tweet

More Decks by ryo nakamaru

Other Decks in Technology

Transcript

  1. AWS Ͱ࢝ΊΔ DeepLearning re:Invent 2016 Machine Learning Mini Con ࢀՃใࠂ

    JAWS-UG AI ࢧ෦ @ 2016.12.09
  2. @pottava SUPINF Inc.

  3. ͍͖ͳΓͰ͕͢

  4. FizzBuzz ஌͍ͬͯΔਓʙʁ

  5. FizzBuzz ॻ͚Δਓʙʁ

  6. ɹfor i in range(1,101): ɹ if i % 15 ==

    0: ɹ print 'FizzBuzz' ɹ elif i % 3 == 0: ɹ print 'Fizz' ɹ elif i % 5 == 0: ɹ print 'Buzz' ɹ else: ɹ print i ɹͰ͢Ͷɺྫ͑͹ɻ
  7. Ͱ͸

  8. ػցֶशͰ FizzBuzz ղ͚Δਓʙʁ

  9. ʁʁʁ

  10. ίϯϐϡʔλʹσʔλΛ౉ͯ͠ ύλʔϯΛݟ͚ͭͤ͞Δ

  11. Ξϓϩʔν Ͳ͏ղ͔͘ ໋ྩత ɹfor i in range(1,101): ɹ if i

    % 15 == 0: ɹ print 'FizzBuzz' ɹ elif i % 3 == 0: ɹ print 'Fizz' ɹ elif i % 5 == 0: ɹ print 'Buzz' ɹ else: ɹ print i ػցֶश େྔͷσʔλΛ౉͠... ༧ଌਫ਼౓ΛߴΊΔ... parameter result 6 Fizz 7 7 10 Buzz 30 FizzBuzz …
  12. Ҏ্ɺͱ͋ΔϫʔΫγϣοϓͰͷ ΞΠεϒϨΠΫͰͨ͠ɻ

  13. ຊ୊

  14. AWS Ͱ࢝ΊΔ DeepLearning

  15. ࠓ೔ͷ࿩୊ ɾ࠷ۙ͋ͬͨ Amazon ػցֶशܥχϡʔεͱ DL ɾre:Invent Machine Learning Mini Con

    ใࠂ ɾMXNet ʹ͍ͭͯ ɾϫʔΫγϣοϓͷ༷ࢠͱ࣮ྫ
  16. Amazon ػցֶशܥχϡʔε

  17. ɾAmazon Echo ɾAmazon Go AI ΧϯύχʔɺAmazon.com

  18. Amazon Echo ɾAmazonʮୈ࢛ͷऩӹͷபʯ͸ 2020 ೥·Ͱʹ 110 ԯυϧՔ͙ ɹਓ޻஌ೳΞγελϯτʮAlexaʯͱ ɹԻ੠ίϯτϩʔϥʔͷʮEchoʯ ɹhttp://thebridge.jp/2016/09/amazon-echo-alexa-add-11-billion-in-revenue-by-2020-2016-9-pickupnews

    ɾhttps://www.amazon.jobs/en/teams/alexa ɾre:Invent Ͱ͸ echo dot ͕ࢀՃऀʹ഑ΒΕ·ͨ͠
  19. Amazon Echo

  20. Amazon Echo Alexa ʹԻ੠Ͱ͓ئ͍ɾ࣭໰͢ΔͨΊͷσόΠεɻ ʢAlexa ͸ Amazon ͕։ൃͨ͠ AI ʣ

    ʮΞϨΫαɺUber ΛݺΜͰɻࠓ೔ͷఱؾ͸ʁ ʯ ʮΞϨΫαɺ͜ͷۂͷԋ૗ऀ͸୭ʁԻྔΛ্͛ͯʯ
  21. Amazon Echo 1. Ի੠Λฉ͖औΓ 2. ԿΒ͔ͷॲཧΛͯ͠ 3. Ի੠Λฦ͢

  22. Amazon Echo 1. Ի੠Λฉ͖औΓ 2. ԿΒ͔ͷॲཧΛͯ͠ 3. Ի੠Λฦ͢ Amazon Lex

    Amazon Polly
  23. ɾAmazon Echo ɾAmazon Go AI ΧϯύχʔɺAmazon.com

  24. Amazon Go

  25. Amazon Go 1. ೖళ࣌ɺήʔτʹεϚϗΞϓϦΛ͔͟͢ 2. ΄͍͠΋ͷΛόοάʹೖΕΔ 3. ͓ళΛग़Δ Coming early

    2017 !! 2131 7th Ave Seattle, Washington
  26. Amazon Rekognition ଞࣾͰ΋੝Μͳ Computer vision API ͷҰछɻ

  27. Amazon Rekognition ਂ૚ֶशϕʔεͷը૾ೝࣝ APIɻ ɾҰൠ෺ମ / ৘ܠݕग़ ɾද৘෼ੳ ɾإͷྨࣅ౓൑ఆ

  28. Amazon Rekognition ΍ͬͯΈͨ

  29. Amazon Rekognition ฐࣾ༐ऀͷ ྨࣅ౓൑ఆɻ

  30. re:Invent Machine Learning Mini Con

  31. Machine Learning Mini Con ɾػցֶशܥͷηογϣϯ / ϫʔΫγϣοϓ ɾhttp://bit.ly/reinvent-2016-ml ɾࠓ೥͸ 17

    ηογϣϯ ɾϫʔΫγϣοϓҎ֎͸ YouTube ͰݟΕ·͢
  32. ೖ໳ฤ ɾMAC201: Amazon Mechanical Turk Λ࢖ͬͯҰൠతಛ௃Λ͔ͭΉ ɾMAC202: Alexa ʹ͓͚Δਂ૚ֶश ɾMAC203:

    Amazon Rekognition ͷ͝঺հ ɾMAC204: Amazon Polly ͷ͝঺հ ɾMAC205: Ϋϥ΢υΒ͘͠εέʔϧ͢Δਂ૚ֶश: ɹɹɹɹɹ AWS Ͱ Caffe ΛεέʔϧΞοϓͯ͠ϏσΦݕࡧΛվળ͢Δ ɾMAC206: ػցֶशͷݱঢ়
  33. தڃฤ ɾMAC301: ਂ૚ֶशͰ޻৔ͷϓϩηεΛม͍͑ͯ͘ ɾMAC302: ෆಈ࢈Ͱͷઓུత༏ҐͷͨΊʹ Amazon ML, Redshift, S3 σʔλϨΠΫΛ׆༻͢Δ

    ɾMAC303: Amazon EMR ͱ Apache Spark ͰΫϥε෼ྨͱ ϨίϝϯσʔγϣϯΤϯδϯΛ։ൃ͢Δ ɾMAC304: Amazon Lex ͷ͝঺հ ɾMAC306: MXNet Λ࢖ͬͯϨίϝϯσʔγϣϯϞσϧΛߏங͢Δ ɾMAC306-R: MXNet Λ࢖ͬͨਂ૚ֶश
  34. தڃฤ ɾMAC307: Predicting Customer Churn with Amazon ML ɾMAC308: ϫʔΫγϣοϓ:

    Amazon Lex, Amazon Polly ͦͯ͠ Amazon Rekognition Λ࢖ͬͨϋϯζΦϯ ɾMAC309: Amazon Polly ͱ Amazon Lex ͷ͝঺հ
  35. ্ڃฤ ɾMAC401: Scalable Deep Learning Using MXNet ɾMAC403: Automatic Grading

    of Diabetic Retinopathy ɹɹɹɹɹ through Deep Learning
  36. ৄࡉ͸ YouTube ͱ Slideshare Ͱ

  37. ϐοΫΞοϓ ɾMAC201: Amazon Mechanical Turk Λ࢖ͬͯҰൠతಛ௃Λ͔ͭΉ ɾMAC206: ػցֶशͷݱঢ় ɾMAC306: MXNet

    Λ࢖ͬͯϨίϝϯσʔγϣϯϞσϧΛߏங͢Δ ɾMAC401: Scalable Deep Learning Using MXNet
  38. MAC201 Mechanical Turk Ͱػցֶश༻σʔλΛ࡞Δ ɾhttps://www.youtube.com/watch?v=vRtLdeNl7Tg ɾେྔͷɺߴ඼࣭ͳσʔληοτ͸ूΊʹ͍͘ ɾϝΧχΧϧλʔΫʹͦͷ࡞੒Λґཔ͢Δ

  39. MAC206 Amazon ૑ۀ͔Β࠷৽ AI αʔϏε·Ͱ঺հ ɾhttps://www.youtube.com/watch?v=HqsUfyu0XJc ɾDeep Learning AMI, MXNet,

    Alexa ͳͲͳͲ.. ɾޙ൒͸ܯ࡯ʹαʔϏεఏڙ͢ΔϞτϩʔϥͷࣄྫ
  40. MXNet

  41. ֶशϑϨʔϜϫʔΫ ͲΕ͕͓޷ΈͰ͔͢ɾɾʁ MXNet / TensorFlow / Caffe / Chainerɻ ɾͲͷχϡʔϥϧωοτ࢖͏ͷʁCNNʁRNNʁ

    ɾGPU ࢖͏ͷʁCPU ͚ͩʁෳ਺ϊʔυ࢖͏ʁ ɾࠃ࢈ΛԠԉʁ
  42. AWS ͸ MXNet Ұ୒ײ͋Δ ɾ͑ɺAmazon DSSTNE ɾɾ ɾͱ͸͍͑ଞͷ΋ݕ౼͍ͨ͠ํ͸ͪ͜Β ɹ CMP314:

    Bringing Deep Learning to the Cloud with Amazon EC2 https://www.youtube.com/watch?v=34Xorby_pyw
  43. MAC306 Netflix ͷϨίϝϯυྫΛ௨ͯ͡ DL / MXNet Λৄઆ ɾhttps://www.youtube.com/watch?v=cftJAuwKWkA ɾDeep Learning

    ͷॳา͔Βɻͱͯ΋෼͔Γ΍͍͢ ɾGitHub ͷ MXNet ϦϙδτϦʹ͋ΔαϯϓϧΛσϞ https://github.com/dmlc/mxnet/tree/master/example/recommenders
  44. ϫʔΫγϣοϓͷ༷ࢠͱ࣮ྫ

  45. ϫʔΫγϣοϓʁ ϋϯζΦϯܗ͕ࣜଟ͍ɻάϧʔϓϫʔΫ΋͋ͬͨΓɻ ɾ࣮ࡍʹखΛಈ͔͢ͷͰͱͯ΋ཧղ͕ਐΉ ɾ·ΘΓͷࢀՃऀͱͷίϛϡχέʔγϣϯ .. !! ɾre:Invent ʹߦ͘ͳΒ௨ৗηογϣϯΑΓΦεεϝ

  46. MAC401 ECS ্Ͱ MXNet ʹΑΔ DL ͷֶशɾਪ࿦Λମݧ ɾECS ͷ Runtask

    + CPU ͷΈ ɾGitHub ͷ awslabs ϦϙδτϦΛར༻ https://github.com/awslabs/ecs-deep-learning-workshop/
  47. ࢼ͢ͷ͸ͱͯ΋؆୯ CloudFormation ʹΑΔ EC2 / ECS ౳ੜ੒ɻͦͷޙ.. ɾLab 3: ECS

    Ͱ MXNet ͷ Jupyter notebook ىಈ ɾLab 4: MXNet ʹΑΔը૾ͷΫϥε෼ྨ ɾLab 5: ECS λεΫͱͯ͠ը૾ΛΫϥε෼ྨ
  48. Deep Learning AMI http://qiita.com/pottava/items/c79117089be2406b127f

  49. ͓஌Βͤ

  50. དྷि͸ JAWS-UG ίϯςφࢧ෦

  51. ECS Λத৺ʹɺίϯςφ·ΘΓͷ࠷৽৘ใΛ͓ಧ͚ʂ http://jawsug-container.connpass.com/

  52. Amazon ECS ɾࠓ೥͸ ECS ͰδϣϒΛ૸ΒͤΔηογϣϯ͕ෳ਺ ɾMXNet on ECS ͷϫʔΫγϣοϓ͸ੈքͰ΋ਓؾ ɾECS

    Ϋϥελ্Ͱ MXNet ͷֶशɾਪ࿦
  53. AWS Batch ɾECS ্ʹ HPC ۀքͷҙຯʹ͍ۙΫϥελΛߏஙɻ ɾδϣϒεέδϡʔϥ ≠ ίϯςφք۾ͷεέδϡʔϥ ɾECS

    ্ͳͷͰɺ࣮͸ίϯςφϕʔε ɾGlue ΍ EFS ͱͷ૊Έ߹Θͤ௒ॏཁ
  54. ͓ΘΓ

  55. גࣜձࣾεϐϯϑ ΞΠσΟΞΛ͔ͨͪʹʂ +

  56. http://prtimes.jp/main/html/rd/p/000000007.000007768.html Comfy for Docker ϓϩδΣΫτ΁ͷ Docker ಋೖɾ։ൃࢧԉɾӡ༻؂ࢹ୅ߦΛ͍ͨ͠·͢ɻ ʢGCP / Azure

    ΋΋ͪΖΜରԠ͍ͯ͠·͢ɾɾʣ https://www.supinf.co.jp/service/dockersupport/
  57. ͝૬ஊ͸͓ؾܰʹͪ͜Β·Ͱ.. 57 <Thank you !! https://www.supinf.co.jp/service/dockersupport/