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

AWS で加速する機械学習 / Accelerate Machine Learning with AWS

7dc8611c26c3ca62c551109c65d04270?s=47 Yuichiro Someya
March 14, 2018
Programming
1
290

AWS で加速する機械学習 / Accelerate Machine Learning with AWS

https://aws.amazon.com/jp/startup-day/tokyo/

7dc8611c26c3ca62c551109c65d04270?s=128

Yuichiro Someya

March 14, 2018
Tweet

More Decks by Yuichiro Someya

See All by Yuichiro Someya
にんげんがさき 基盤はあと / Developers over ML platform
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
0
9.4k
機械学習をスモールスタートさせる方法 / small machine learning
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
3
1.7k
アットホームな分析基盤の作り方 / Homemade Machine Learning Toolkits
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
1
730
サービス開発、機械学習、クラウド / the trinity of machine learning
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
0
2.7k
成長を止めない機械学習のやり方 / Don't stop 'til you get enough (data).
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
15
4.7k
クックパッドの機械学習基盤 2018 / Machine Learning Platform at Cookpad ~ 2018 ~
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
15
15k
PyTorchとCaffe2とONNXと深層学習モデルのデプロイについて
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
1
2.6k
クックパッドにおけるAWS GPUインスタンスの利用事例 / Powering by AWS GPU Instances in Cookpad Inc
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
0
330
How we use GPUs in Cookpad
7dc8611c26c3ca62c551109c65d04270?s=48 ayemos
0
88

Other Decks in Programming

See All in Programming
Angular‘s Future without NgModules: Architectures with Standalone Components @enterJS
15934fa2aa7b2ce21f091e9b7cffa856?s=48 manfredsteyer
PRO
0
230
JSのウェブフレームワークで高速なルーターを実装する方法
6e34bfefc3edbe746ac7a71ca6635309?s=48 usualoma
1
1.8k
チームでカレーを作ろう！アジャイルカレークッキング
D30220c903365b2170e791313296b2f9?s=48 akitotsukahara
0
830
オブジェクト指向で挫折する初学者へ
Dde759e35b9d62d448b3924502750a31?s=48 deepoil
0
170
マルチプロダクト×非構造化データ×機械学習を支えるデータ信頼性
9b95bbde8856da418c299333b312da3f?s=48 akino
0
150
Gitlab CIでMRを自動生成する
1a2f29252240936c0ac4db4edab36d41?s=48 forcia_dev_pr
0
120
Android Compose Component - mapping.
7589a5a8fec022e8af4e46525150a291?s=48 taehwandev
0
140
Scrum Fest Osaka 2022/5年で200人になったスタートアップの アジャイル開発の歴史とリアル
F0a8d07597a19192791bf663504d2a17?s=48 atamaplus
1
900
Node-RED 3.0 新機能紹介
A56269ca269966f71293d08b32ad53b3?s=48 utaani
0
140
GoogleI/O2022 LT報告会資料
2bd3b8c3550384ab61d671b673c119bb?s=48 shinsukefujita1126
0
340
模組化的Swift架構(二) DDD速成
A3d3b179955bdea9fc05b3251f1b1f10?s=48 haifengkao
0
390
RFC 9111: HTTP Caching
1ff811939fd0923df8321ec6d8bf9d4b?s=48 jxck
0
160

Featured

See All Featured
"I'm Feeling Lucky" - Building Great Search Experiences for Today's Users (#IAC19)
3d4519fd34b76fb265fc0237f3792bd4?s=48 danielanewman
212
20k
Building Adaptive Systems
06f8b41980eb4c577fa40c41d5030c19?s=48 keathley
25
1.1k
Creating an realtime collaboration tool: Agile Flush - .NET Oxford
Fd55de4174accaf3b4f030e43a8a70c6?s=48 marcduiker
4
510
Product Roadmaps are Hard
7cbd3f5f922d798cc312eaf596de7ae6?s=48 iamctodd
34
6.5k
Web development in the modern age
465724d73fe3a92c0879fdfb43a3a6f3?s=48 philhawksworth
197
9.3k
Reflections from 52 weeks, 52 projects
E43f8dfd01057f8304f5f8fb9a294d7d?s=48 jeffersonlam
337
17k
The Brand Is Dead. Long Live the Brand.
A415db3ac9e9668acbc1fb36eead84ac?s=48 mthomps
46
2.7k
Faster Mobile Websites
C620790ae5bf5b50c245b2e0ef95f338?s=48 deanohume
294
28k
CSS Pre-Processors: Stylus, Less & Sass
7559f6cff1f5efc2d210965febd4d71c?s=48 bermonpainter
349
27k
Building an army of robots
5f2da528927a2ec9ba4fec2069cbc958?s=48 kneath
299
40k
Ruby is Unlike a Banana
6804f1775cb4babfcc3851298566fbce?s=48 tanoku
91
9.2k
Agile that works and the tools we love
462dad0dd24c568a32dcdb0ae895ae1b?s=48 rasmusluckow
319
19k

Transcript

  1. "84ͰՃ଎͢Δػցֶश ΫοΫύουגࣜձࣾ ੕๺ేછ୩༔Ұ࿠ "844UBSUVQ%BZ5PLZP

  2. ੕๺ే ΄͠΄͘ͱ !LBOJ@C  ΫοΫύουגࣜձࣾ
 ΠϯϑϥετϥΫνϟʔ෦෦௕  43& ηΩϡϦςΟΤϯδχΞ 

    "84ೝఆ4" %FW0QTΤϯδχΞ 1SPGFTTJPOBM
  3. None

  4. Ϩγϐ਺ ສ඼Ҏ্ ࠃ಺ͷ݄ؒར༻ऀ਺໿ ສਓ

  5. ରԠݴޠ ݴޠΧࠃ ւ֎ͷ݄ؒར༻ऀ਺  ສਓҎ্

  6. ΫοΫύουͱ"84  ೥ʹ%$͔Β׬શҠߦ  BQOPSUIFBTU͓ΑͼVTFBTU
 Λओʹར༻  άϩʔόϧαʔϏε͸ओʹ64ͰՔಇ "NB[PO&$Πϯελϯε d

    ୆ఔ౓ "NB[PO4ΦϒδΣΫτ ԯҎ্  SFRTҎ্ ϐʔΫ࣌ಉ࣌ϦΫΤετ਺

  7. IUUQTTQFBLFSEFDLDPNLBOOZNJBPKJBOTIVNPGBMTFSPHVXPJJHBO[JOJTVSVBLJUFLVUJZB IUUQTTQFBLFSEFDLDPNBZFNPTNBDIJOFMFBSOJOHPQTBUDPPLQBE

  8. "84SF*OWFOU IUUQTXXXTMJEFTIBSFOFU"NB[PO8FC4FSWJDFTDPODPOUBJOFSJ[FENBDIJOFMFBSOJOHPOBXT

  9. "84ͰՃ଎͢Δػցֶश

  10. ྉཧ͖Ζ͘ ৞ΈࠐΈχϡʔϥϧωοτϫʔΫ ʹΑΔྉཧը૾ͷࣗಈೝࣝ εϚʔτϑΥϯͷࣸਅͷ಺ɺ ྉཧࣸਅΛࣗಈతʹऩूه࿥

  11. %&.0

  12. None

  13. ৽͍͠௅ઓ  ॳͷػցֶशϓϩμΫτ  ηϛϦΞϧλΠϜͳը૾ೝࣝ  8FCΞϓϦέʔγϣϯͱҧ͏ϫʔΫϩʔυ  ඞཁͩͬͨ΋ͷ 

    ৽͍͠ϫʔΫϩʔυʹରͯ͠εέʔϧ͢ΔΠϯϑϥ  ࣦഊͯ͠΋طଘͷ؀ڥʹӨڹΛ༩͑ͳ͍ಠཱͨ͠؀ڥ  ૉૣ͘ࢼߦࡨޡΛ܁Γฦ͢ͷʹద࣮ͨ͠ݧ؀ڥ

  14. ৽͍͠௅ઓ  ॳͷػցֶशϓϩμΫτ  ηϛϦΞϧλΠϜͳը૾ೝࣝ  8FCΞϓϦέʔγϣϯͱҧ͏ϫʔΫϩʔυ  ඞཁͩͬͨ΋ͷ 

    ৽͍͠ϫʔΫϩʔυʹରͯ͠εέʔϧ͢ΔΠϯϑϥ  ࣦഊͯ͠΋طଘͷ؀ڥʹӨڹΛ༩͑ͳ͍ಠཱͨ͠؀ڥ  ૉૣ͘ࢼߦࡨޡΛ܁Γฦ͢ͷʹద࣮ͨ͠ݧ؀ڥ ৽͍͠αʔϏε࡞ΓΛ lՃ଎z͢Δ؀ڥ

  15. εέʔϥϏϦςΟͷ֬อ  ༧ଌͰ͖ͳ͍τϥϑΟοΫ େྔͷը૾ΞοϓϩʔυͳͲ   ಥવ57ʹ঺հ͞ΕΔ͜ͱ΋͋Δ͔΋͠Εͳ͍ ࣮ࡍ͋ͬͨ  

    ը૾ͷΞοϓϩʔυ΍൑ఆʹ͸͕͔͔࣌ؒΔ d਺ඦNT   ಉظॲཧͯ͠͠·͏ͱϢʔβମݧΛѱ͘͢Δ

  16. ඇಉظͳΞʔΩςΫνϟ  ը૾Ξοϓϩʔυͱը૾൑ఆΛݸผʹॲཧ  ͦΕͧΕΛεέʔϧ͠΍͍͢Α͏ͳߏ੒ʹ  "NB[PO4ͷQSFTJHOFE63-Ͱ௚઀Ξοϓϩʔυ  4OPUJpDBUJPOΛར༻͠ॲཧ͢΂͖ը૾ͷ৘ใ͸
 "NB[PO424΁

  17. ؀ڥͷ෼཭  ͜Ε·Ͱͷ8FCαʔό؀ڥͱҧ͏؀ڥ͕ٻΊΒΕͨ  ҧ͏ݴޠ ҧ͏ϫʔΫϩʔυ ҧ͏ϋʔυ΢ΣΞ (16  

    ίϯςφ؀ڥΛར༻  ϥϯλΠϜ΍(16υϥΠόɺઃఆΛ෼཭  "NB[PO&$4Λར༻ͯ͠؅ཧ  طʹଟ͘ͷαʔϏε͕ίϯςφ؀ڥͰՔಇ

  18. API Server "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ

  19. API Server "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ
 4όέοτʹΞοϓϩʔυ

  20. API Server "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ
 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕
 424ʹ౤ೖ͞ΕΔ {

    bucket: thumbnails key: 123.jpg }

  21. API Server "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ
 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕
 424ʹ౤ೖ͞ΕΔ {

    bucket: thumbnails key: 123.jpg } ϝοηʔδΛ
 424͔ΒऔΓग़͢ ECS Cluster

  22. API Server "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ
 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕
 424ʹ౤ೖ͞ΕΔ {

    bucket: thumbnails key: 123.jpg } ϝοηʔδΛ
 424͔ΒऔΓग़͢ ECS Cluster αϜωΠϧͷμ΢ϯϩʔυ

  23. API Server "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ
 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕
 424ʹ౤ೖ͞ΕΔ {

    bucket: thumbnails key: 123.jpg } ϝοηʔδΛ
 424͔ΒऔΓग़͢ ECS Cluster αϜωΠϧͷμ΢ϯϩʔυ ը૾ͷ൑ఆΛߦ͍݁ՌΛ
 "1*αʔόʹ௨஌

  24. API Server "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ { bucket: thumbnails key: 123.jpg }

    ϝοηʔδΛ
 424͔ΒऔΓग़͢ ECS Cluster αϜωΠϧͷμ΢ϯϩʔυ ը૾ͷ൑ఆΛߦ͍݁ՌΛ
 "1*αʔόʹ௨஌ খ͍͞αϜωΠϧΛ
 4όέοτʹΞοϓϩʔυ 4OPUJpDBUJPO͕
 424ʹ౤ೖ͞ΕΔ ಛผͳίʔυ΍ΠϯϑϥΛ༻ҙͤͣ
 εέʔϥϒϧͳߏ੒ʹ

  25. API Server "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ
 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕
 424ʹ౤ೖ͞ΕΔ {

    bucket: thumbnails key: 123.jpg } αϜωΠϧͷμ΢ϯϩʔυ ը૾ͷ൑ఆΛߦ͍݁ՌΛ
 "1*αʔόʹ௨஌ ϝοηʔδΛ
 424͔ΒऔΓग़͢ ECS Cluster ECS Cluster 424ͷΩϡʔ௕ʹԠͯ͡ &$4λεΫ΋εέʔϧ

  26. "1*αʔόʹ
 4QSFTJHOFE63-Λཁٻ খ͍͞αϜωΠϧΛ
 4όέοτʹΞοϓϩʔυ TUIVNCOBJMTKQH 4OPUJpDBUJPO͕
 424ʹ౤ೖ͞ΕΔ { bucket: thumbnails

    key: 123.jpg } ϝοηʔδΛ
 424͔ΒऔΓग़͢ ECS Cluster αϜωΠϧͷμ΢ϯϩʔυ ը૾ͷ൑ఆΛߦ͍݁ՌΛ
 "1*αʔόʹ௨஌ API Server API Server &$4λεΫ͔Βͷ൑ఆ݁ՌΛ଴͚ͭͩ

  27. &$4্Ͱͷࣸਅ൑ఆ  ࣸਅ൑ఆʹ(16Λ࢖༻͍ͯ͠Δ  ൑ఆॲཧʹ(16ΞΫηϥϨʔγϣϯ͕ར༻Ͱ͖Δ  ಉՁ֨ଳͷ$16Πϯελϯεͱൺ΂ͯdഒͷੑೳࠩ  HΠϯελϯεͰࣸਅ൑ఆ༻ͷ&$4$MVTUFSΛߏ੒

  28. IBLPʹΑΔΞϓϦέʔγϣϯఆٛ  IBLPIUUQTHJUIVCDPNFBHMFUNUIBLP  &$4ޓ׵ͷίϯςφσϓϩΠπʔϧ  ZBNMKTPOOFUܗࣜͰͷΞϓϦέʔγϣϯఆٛ  "VUP4DBMFɺ-#ͷࢦఆͳͲ΋Մೳ 

    શ։ൃऀ͕ࣗ෼ͰఆٛΛॻ͍ͯσϓϩΠՄೳ scheduler: type: ecs region: ap-northeast-1 cluster: hako-production-g2 desired_count: 1 app: image: food-photo-classifier cpu: 128 memory: 3072 memory_reservation: 2048 env: AWS_REGION: ap-northeast-1 COOKPADNET_ENV: production ...

  29. ӡ༻࣮੷  ը૾ೝࣝϞσϧΛҰ౓Ξοϓσʔτ  Ξοϓσʔτ΋৐Γӽ͑ɺେ͖ͳ໰୊΋ͳ͘ӡ༻Ͱ͖͍ͯΔ ສਓҎ্ ͷྉཧࣸਅΛه࿥ ͷϢʔβʔ͕ར༻  ສຕҎ্

    ೥݄೔ݱࡏ

  30. ৽͍͠௅ઓ  ॳͷػցֶशϓϩμΫτ  ηϛϦΞϧλΠϜͳը૾ೝࣝ  8FCΞϓϦέʔγϣϯͱҧ͏ϫʔΫϩʔυ  ඞཁͩͬͨ΋ͷ 

    ৽͍͠ϫʔΫϩʔυʹରͯ͠εέʔϧ͢ΔΠϯϑϥ✅  طଘͷ؀ڥʹӨڹΛ༩͑ͳ͍ಠཱͨ͠؀ڥ✅  ࢼߦࡨޡΛ܁Γฦ͢ͷʹద࣮ͨ͠ݧ؀ڥ

  31. ࣮ݧ؀ڥͮ͘Γ  खΛಈ͔ͯ͠ػցֶशͷ஌ݟΛಘΔඞཁ͕͋ͬͨ  "84؀ڥ͸શͯΠϯϑϥ෦͕؅ཧ͍͕ͯͨ͠ʜ  Ϧιʔεͷ༻ҙ͕௥͍͔ͭͳ͍  ͍ΘΏΔຊ൪؀ڥͷӡ༻ʹ߹Θͳ͍ 

    ٻΊΒΕΔͷ͸αʔϏεʹඞཁͳ࣮ݧ݁Ռ  தԝ؅ཧʹݶք͕๚Εͨ ֤ࣄۀ෦ Πϯϑϥ෦ ٕज़ج൫ܥ෦ॺ ݚڀ ։ൃ෦

  32. ݚڀ։ൃج൫  ʮ͔ͤͬ͘৮ΕΔͷ͔ͩΒউखʹਐΊΒΕΕ͹ྑ͍ʯ  ݚڀ։ൃ෦ͷΤϯδχΞ͕ࣗ༝ʹखΛಈ͔ͤΔ৔ॴ  ຊ൪؀ڥͱ͸෼཭͞Ε͍ͯΔ  ؅ཧ͢΂͖෦෼͚͓ͩ͑͞ɺଟ͘Λ։ൃऀ΁Ҡৡ

  33. ΞΧ΢ϯτ෼཭  "84ΞΧ΢ϯτΛผʹ࡞੒ͯ͠෼཭  "840SHBOJ[BUJPOΛར༻ͯ͠ଞͷ༻్ʹ΋࡞੒  *".6TFS͸ຊ൪؀ڥͷݸਓϢʔβ͔Β"TTVNF3PMFPS440  $MPVE5SBJM΍$POpHΛ༗ޮʹͯ͠PVUQVUઌΛอޢ 

    lى͜Δલʹ๷͙zͰͳ͘l͙͢ؾ͚ͮΔzlޙ͔Β௥͑Δz

  34. ωοτϫʔΫ؀ڥ  ౿Έ୆44)αʔό͸ຊ൪ΞΧ΢ϯτͷ71$  Πϯϑϥ෦؅ཧ 71$1FFSJOHͰ઀ଓͯ͠ू໿  /BNFλάΛ࢖ͬͨ%/4ਖ਼Ҿ͖ɺٯҾ͖Λఏڙ  "84$POpH3VMFTͰηΩϡϦςΟάϧʔϓͷΠϯλʔωοτ

    ղ์ΛνΣοΫ
  35. None

  36. ίετ؅ཧ  ࡉ͔͍؅ཧ͸͠ͳ͍͕͋Δఔ౓͸೺Ѳ  ৵ೖͳͲͷΞϥʔτʹ΋ͳΔ  #VEHFUTػೳΛར༻  ༧ଌֹͳͲͰ4/4/PUJpDBUJPOΛड͚औΓରԠ

  37. None

  38. αϙʔτͷ׆༻  σϕϩούʔαϙʔτ͔ΒΤϯλʔϓϥΠζαϙʔτ΁ͷҠߦ  ࣌ؒͱ஌ࣝΛങ͏  ։ൃऀ͕ࣗ਎Ͱ໰୊ΛղܾͰ͖ΔΑ͏ʹ  ར༻ن໛ʹΑͬͯ͸ϏδωεͰ΋໰୊ͳ͍ͷͰཁݟੵ΋Γ

  39. ։ൃऀΞΧ΢ϯτ  ݚڀ։ൃ༻ͱಉ༷͕ͩશ։ൃऀ͕ࣗ༝ʹར༻Մೳ  ৭ʑͳ"84αʔϏεΛձࣾ࣋ͪͰݕূͰ͖Δ

  40. ར༻࣮੷  "84Λࣗ෼Ͱ৮ͬͯΈΔਓ͕ͱͯ΋૿͑ͨ  ༷ʑͳ"84αʔϏεͷར༻ͱऔΓࠐΈ  "1*(BUFXBZ -BNCEB $PHOJUP %ZOBNP%#

    &MFNFOUBM.FEJB 4FSWJDF (MVF ,JOFTJT 1PMMZ 4BHF.BLFS 93BZ -FY 1JOQPJOU  "QQ4ZODFUDʜ  ৽ػೳɺαʔϏεͷ૑ग़ΛՃ଎  ྉཧ͖Ζ͘ "MFYB4LJMM DPPLQBE57 FUD

  41. ֶΜͩ͜ͱ  ৽͍͜͠ͱΛૉૣ͘খ࢝͘͞ΊΔͨΊʹ͸ݖݶͱ੹೚͕ཁΔ  ҆৺ͯࣗ͠༝ʹ࢖͑Δ؀ڥͮ͘Γ͕ͦ͜ॏཁ  "84ʹ͸࣮ݱ͢ΔͨΊͷπʔϧ͕ἧ͍ͬͯΔ

  42. Ͳ͏࢖͍ͬͯͨͷ͔ʁ

  43. ࣗݾ঺հ  છ୩༔Ұ࿠<:VJDIJSP4PNFZB>  ΫοΫύουגࣜձࣾݚڀ։ൃ෦
 ϦαʔνΤϯδχΞ೥݄d  ʰػցֶशʹඞཁͳج൫Λ
 "84্ʹߏங͢Δʱ UXJUUFSDPN!BZFNPT@Z

    HJUIVCDPNBZFNPT IUUQTXXXBZFNPTNF

  44. εϞʔϧελʔτͰεέʔϧ͢Δ (16࣮ݧ؀ڥΛ"84্ʹߏங͢Δ

  45. എܠΫοΫύουݚڀ։ൃ෦  ೥݄ʹൃ଍  ॳظϝϯόʔ͸໊  ೥݄ݱࡏ͸
 ࠃ಺ʹ໊ ւ֎ʹ໊ 

    ෳ਺ͷਂ૚ֶशϓϩδΣΫτ  (16࣮ݧ؀ڥ͕ඞཁ ϢʔβʔϑΟʔυόοΫ ͷࣗಈλά෇͚ ը૾෼ੳʹΑΔϨγϐͷ෼ྨ Ϩγϐࡐྉ໊ͷࣗಈਖ਼نԽ

  46. (16࣮ݧ؀ڥͷཁ݅  ෳ਺ਓ͕ෳ਺ͷ࣮ݧΛߦ͏͜ͱ͕ग़དྷΔ

  47. &$ΠϯελϯεΛ࣮ݧ؀ڥͱͯ͠࢖͏  ΦϯσϚϯυΠϯελϯεͷར఺͕͋Δ  σϕϩούʔ਺΍λεΫʹԠͯ͡ඞཁͳ਺ͱεϖοΫ͕มԽ͢Δ  1/7*%*"(16࠷৽Ϟσϧͷαϙʔτ  ઐ༗Πϯελϯεͷར఺ 

    (16λεΫ͸جຊతʹഉଞͰɺଞͷ։ൃऀͱಉ࣌ʹ͸ར༻ͮ͠Β͍  ݚڀ։ൃ໨తͰ͸ɺࡶʹTVEPग़དྷΔಠཱͨ͠؀ڥ͕ศར

  48. (16࣮ݧ؀ڥ &$ ͕ग़དྷΔ·Ͱ  &$ΠϯελϯεΛ্ཱͪ͛ɺTTIΛૄ௨ͤ͞Δ  ඞཁͳιϑτ΢ΣΞͷΠϯετʔϧʗઃఆ  OWJEJBESJWFS $6%"

    DV%//ͳͲ݁ߏେม  ".* "NB[PO.BDIJOF*NBHF ͷར༻

  49. ιϑτ΢ΣΞɺϛυϧ΢ΣΞͷ
 Πϯετʔϧͱઃఆ ".*࡞੒ͱ؅ཧ

  50. ιϑτ΢ΣΞɺϛυϧ΢ΣΞͷ
 Πϯετʔϧͱઃఆ εφοϓγϣοτ
 ".*ͷ࡞੒ ".*࡞੒ͱ؅ཧ

  51. ౳ ιϑτ΢ΣΞɺϛυϧ΢ΣΞͷ
 Πϯετʔϧͱઃఆ ίʔυԽͯ͠όʔδϣϯ؅ཧ εφοϓγϣοτ
 ".*ͷ࡞੒ ".*࡞੒ͱ؅ཧ

  52. IUUQTEPDTBXTBNB[PODPNKB@KQEMBNJMBUFTUEFWHVJEFXIBUJTEMBNJIUNM "84%FFQ-FBSOJOH".*  ਂ૚ֶशʹ ຆͲͷ৔߹ڞ௨ͯ͠ ඞཁͳιϑτ΢ΣΞ͕ηοτ Ξοϓ͞ΕͨϚωʔδυͳ".* l/7*%*"$6%"ɺ/7*%*"DV%//ɺ͓Αͼ࠷΋ਓؾ͕ߴ͍σΟʔϓϥʔχϯά
 ϑϨʔϜϫʔΫͷ࠷৽ϦϦʔεΛ࢖༻͢ΔΑ͏ʹࣄલʹઃఆ͞Εͯఏڙ͞Ε·͢ɻz 

  53. ࣗ࡞".* ౳ ίʔυԽͯ͠όʔδϣϯ؅ཧ εφοϓγϣοτ
 ".*ͷ࡞੒ "84 %FFQ-FBSOJOH ".* FD3VO*OTUBODFT

  54. ࣗ࡞".* ౳ ίʔυԽͯ͠όʔδϣϯ؅ཧ εφοϓγϣοτ
 ".*ͷ࡞੒ "84 %FFQ-FBSOJOH ".* FD3VO*OTUBODFT ".*Λར༻ͯ͠ɺ؀ڥΛؚΊͯΦϯσϚϯυͳ


    (16࣮ݧ؀ڥΛ࣮ݱ

  55.  ".*Λར༻͢Δ͜ͱʹΑΓιϑτ΢ΣΞͷΠϯετʔϧͱઃఆ ࡞ۀΛল͘͜ͱ͕ग़དྷͨ  71$ 4FDVSJUZ(SPVQͳͲͷઃఆ߲໨͕࢒͍ͬͯΔ  $IBUCPUӽ͠ʹطఆ஋Λར༻ͯ͠ΠϯελϯεΛ࡞Δ ΋͏ͻͱ޻෉ ʙ$IBUCPU

    "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ

  56. ΋͏ͻͱ޻෉ ʙ$IBUCPU "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ

  57. DSFBUFXPSLCFODI 0VUHPJOH8FCIPPLT ΋͏ͻͱ޻෉ ʙ$IBUCPU "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ

  58. DSFBUFXPSLCFODI 0VUHPJOH8FCIPPLT TMBDL@VJE6  DPNNBOEDSFBUFXPSLCFODI MBNCEB*OWPLF FD3VO*OTUBODFT 71$ 4VCOFU /BNFλά౳


    ؀ڥಛ༗ͷઃఆ΋͜͜Ͱ΍Δ ΋͏ͻͱ޻෉ ʙ$IBUCPU "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ

  59. *ODPNJOH8FCIPPLT ATTIBZFNPTXPSLCFODIEOTDPNQBOZA TTI ΋͏ͻͱ޻෉ ʙ$IBUCPU "84-BNCEBʹΑΔ&$ͷૢ࡞ʙ

  60. ͞Βʹ΋͏ͻͱ޻෉ ʙ"84-BNCEBΛར༻ͨ͠ΞΠυϧΠϯελϯεͷ؂ࢹʙ  ࢖͍ͬͯͳ͍Πϯελϯεͷམͱ͠๨ΕͳͲ͕ൃੜ  අ༻໘Ͱେ͖ͳ໰୊Ͱ͸ͳ͍͕ɺ͍ۙকདྷ૊৫֦େͷোนʹ
 ͳΓ͔Ͷͳ͍  ར༻୯Ձͷߴ͍Πϯελϯεʹ͍ͭͯɺΞΠυϧঢ়ଶΛ؂ࢹ

  61. ͞Βʹ΋͏ͻͱ޻෉ ʙ"84-BNCEBΛར༻ͨ͠ΞΠυϧΠϯελϯεͷ؂ࢹʙ ;[[

  62. ͞Βʹ΋͏ͻͱ޻෉ ʙ"84-BNCEBΛར༻ͨ͠ΞΠυϧΠϯελϯεͷ؂ࢹʙ IPVSMZ FD4UPQ*OTUBODFT ;[[

  63. ͞Βʹ΋͏ͻͱ޻෉ ʙ"84-BNCEBΛར༻ͨ͠ΞΠυϧΠϯελϯεͷ؂ࢹʙ IPVSMZ FD4UPQ*OTUBODFT ;[[ ར༻ऀʹ௨஌ʗࣗಈͰ4UPQ 

  64. (16࣮ݧ؀ڥ·ͱΊ  ".*ͷར༻ʹΑΔଈ࣌ར༻Մೳͳ  &$ΠϯελϯεΛ4MBDL͔Βૢ࡞͢Δ  ΠϯελϯεΛ্ཱͪ͛Δࡍͷઃఆ஋Λڞ௨Խ  4MBDLͷ*ODPNJOH8FCIPPL0VUHPJOH8FCIPPLͱ
 "NB[PO"1*(BUFXBZ"84-BNCEBͷ૊Έ߹Θͤ

     ҆ՁͰαʔόʔϨεͳ$IBU#PU͕࣮૷Ͱ͖Δ

  65. ࣮ݧ͔Βຊ൪΁

  66. ػցֶशϞσϧͷ։ൃͱσϓϩΠ  ػցֶशϞσϧ͸σϓϩΠ͞ΕͯॳΊͯՁ஋ΛੜΉ  FHྉཧ͖Ζ͘  ϞσϧͱΞϓϦέʔγϣϯຖʹઃܭͱσϓϩΠ༻։ൃ࡞ۀ͕ੜ͡Δ  ʮػցֶशΞϓϦέʔγϣϯઃܭ։ൃ͸୭͕΍Δͷ͔ʁʯ 

    εϞʔϧελʔτ΍૊৫ͷεέʔϥϏϦςΟʹڞʹࢧো͕ग़Δ

  67. "NB[PO4BHF.BLFS  "84ͷϚωʔδυػցֶशϓϥοτϑΥʔϜ  "84SF*OWFOUͰൃද  ػցֶशϞσϧͷֶशͱσϓϩΠΛαϙʔτ  ϓϥοτϑΥʔϜ೚ͤͰεέʔϥϒϧͳσϓϩΠख๏ 

    ϞχλϦϯά΍"#ςετͳͲϓϥοτϑΥʔϜͳΒͰ͸ͷ
 ػೳ΋ࠓޙʹظ଴

  68. ·ͱΊ

  69. ͢͹΍͘खΛಈ͔͠αʔϏεΛͭ͘ΔͨΊʹ ηϧϑαʔϏεԽ શͯΛ؅ཧ͢ΔͷͰͳ͘
 τϨʔεͰ͖Δ࢓૊Έͮ͘Γ $MPVE5SBJM $POGJH ΤϯλʔϓϥΠζαϙʔτͷ׆༻ ྉཧ͖Ζ͘ͷެ։

  70. (16࣮ݧ؀ڥ ".*Λར༻ͨ͠ (16࣮ݧ؀ڥͷ੔උ ".* Πϯελϯεͷ
 མͱ͠๨Ε؂ࢹ Πϯελϯε؅ཧ$IBU#PU -BNCEB $MPVE8BUDI ΦϯσϚϯυͳ(16࣮ݧ؀ڥΛ࣮ݱ

  71. 8F`SFIJSJOH IUUQTDPPLQBEKPCT