Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
ロジスティック回帰で お金もらえるし職務経歴書は倍になる
Search
yush1ga
August 06, 2018
Programming
1
3.2k
ロジスティック回帰で お金もらえるし職務経歴書は倍になる
yush1ga
August 06, 2018
Tweet
Share
More Decks by yush1ga
See All by yush1ga
日系大手と日系スタートアップと外資系で働いてみた
yush1ga
2
640
Other Decks in Programming
See All in Programming
Docker_OSS_ホスティング入門
satokoki645
0
110
PHPはいつから死んでいるかの調査
chiroruxx
2
430
Git Rebase
bkuhlmann
11
1.6k
Exploring the Implementation of “t.Run”, “t.Parallel”, and “t.Cleanup”
akarin
1
150
冗長なエラーログを削減し、スタックトレースを手に入れる / Reducing Verbose Error Logs and Obtaining Stack Traces
upamune
0
1.1k
JavaScript Closure
asoluka
0
1.3k
Implementing Design Systems in Swift
seyfoyun
2
500
パフォーマンスを求めてDBに機能を寄せる戦略
aoyagikouhei
0
110
SIMD Parallel Programming with the Vector API
josepaumard
0
240
Let's learn code review
riofujimon
2
620
Kotlin Multiplatform at Stable and Beyond (Android Makers 2024)
zsmb
0
550
Webアプリをできるだけコードを手書きしないで作ってみる
tomokusaba
2
190
Featured
See All Featured
Atom: Resistance is Futile
akmur
260
25k
Code Reviewing Like a Champion
maltzj
515
39k
Raft: Consensus for Rubyists
vanstee
133
6.3k
Thoughts on Productivity
jonyablonski
60
3.9k
ReactJS: Keep Simple. Everything can be a component!
pedronauck
660
120k
Scaling GitHub
holman
457
140k
A Modern Web Designer's Workflow
chriscoyier
689
190k
Making Projects Easy
brettharned
109
5.5k
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
34
6.1k
Keith and Marios Guide to Fast Websites
keithpitt
408
22k
[Rails World 2023 - Day 1 Closing Keynote] - The Magic of Rails
eileencodes
9
1.3k
Building Your Own Lightsaber
phodgson
100
5.7k
Transcript
ϩδεςΟοΫճؼͰ ͓ۚΒ͑Δ͠৬ܦྺॻഒʹͳΔ ࢤլ ༏ؽ
• େखWebܥاۀۈ (20164݄~, ৽ଔೖࣾ) • Apache Pulsar Committer (20176݄~) •
ϑΝΠϯσΟגࣜձࣾ ػցֶशΤϯδχΞ (20177݄~, ۀҕୗ) • IDAKS٬һݚڀһ (201810݄~༧ఆ) ࢤլ ༏ؽ
࣍ • ϩδεςΟοΫճؼͰ͓ۚΒ͑Δ • ৬ܦྺॻͷ • ͦͷଞΑ͔ͬͨ͜ͱͷ
ϩδεςΟοΫճؼͷ
͓ͼ • λΠτϧͪΐͬͱΓ • ϩδεςΟοΫճؼඇ Deep ͷϝλϑΝʔ
ػցֶशͱͷผΕ • େֶӃͰը૾ೝٕࣝज़Λར༻ͨ͠ݚڀ • ब৬ޙେنϛυϧΣΞ։ൃ • େنͳ։ൃָ͔͕ͬͨ͠ ػցֶश͔ΒΕΔ͜ͱʹ
͘झຯͰऔΓΉ • Kaggleʁ • উͪʹߦ͘ͳΒը૾ೝࣝܥ • Deep LearningΔͨΊͷGPU͕ͳ͍
͘झຯͰऔΓΉ • ͱΓ͋͑ͣجૅΛݻΊΔ • λΠλχοΫͷ٬ͷੜࢮ • ΞϠϝͷछྨ • 28×28ͷखॻ͖ͷࣈ Λ
ແҙຯʹࣝผ͢Δʑ
Findyͱͷग़ձ͍ ϢʔβΠϯλϏϡʔʹͯ 'JOEZͷ͋Δʁ ˓˓ͱ✗✗Ͱ͔͢Ͷ ͦΕͬͯղܾͰ͖Δ ͏ʔΜɺͦΕʜ
ϩδεςΟοΫճؼʂ
ελʔτΞοϓͷػցֶशࣄ • ͦΕ΄Ͳେنͳσʔλ͕ͳ͍߹͕ଟ͍ ※ → Deep Learning ͋·Γඞཁ͕ͳ͍ • ͘ઙ͘ඇ
Deep ͳࣝΛ ษڧ͓͍ͯͨ͠ͷ͕Α͔ͬͨ ※ ࢲͷؾ࣋ͪͰ͋Γɼ౷ܭత༗ҙੑ͋Γ·ͤΜ
৬ܦྺॻͷ
ࢦ໊ऩότϧ ٯٻਓͷస৬αΠτͰ ࢦ໊ऩότϧ͠ͳ͍ʁ 2ळ
ࢦ໊ऩότϧ ͑͑Ͱʂ ϥϯνΛ͔͚͍ͨઓ͍͕։࠵ ͑͑Ͱʂ ͑͑Ͱʂ ͑͑Ͱʂ
2ळͷ৬ܦྺॻ 14݄ (ೖࣾ) 210݄ 110݄ ݚम ϓϩδΣΫτ1 ී௨ʹಇ͍͍ͯΔͱ ৬ܦྺॻʹॻ͚Δͷ͜ͷ1͚ͭͩ
෭ۀ͕͋Δͱ͖ 14݄ (ೖࣾ) 210݄ 110݄ ݚम ϓϩδΣΫτ1 27݄ (෭ۀ։࢝)
ϓϩδΣΫτ2 2ݸॻ͚ΔͷͰ৬ܦྺॻ2ഒʂ
݁Ռ ࠷ߴࢦ໊ऩ: 1Ґ ฏۉࢦ໊ऩ: 1Ґ ɹ ࢦ໊: 2Ґ
ࢦ໊ͷ༁ ػցֶशΤϯδχΞɾσʔλαΠΤϯςΟετ ػցֶशΠϯϑϥɾόοΫΤϯυ ϑϧελοΫ ෆ໌
ࢦ໊ͷ༁ ػցֶशΤϯδχΞɾσʔλαΠΤϯςΟετ ػցֶशΠϯϑϥɾόοΫΤϯυ ϑϧελοΫ ෆ໌ ۀͰػցֶशͬͯͳ͍ͷʹʂ
ͦͷଞྑ͔ͬͨ͜ͱͷ
ຊۀػցֶशؔ࿈ʹ • R&Dҟಈ • ෭ۀͷ͓͔͛Ͱ ϒϥϯΫͦΕ΄Ͳ͖ͭ͘ͳ͍ • ຊۀͷ͓͔͛Ͱ ֶੜ࣌ΑΓwell-organizedͳίʔυ͕ॻ͚Δ
͔ΘΓʹفΔ͕࣌ؒ૿͑ͨ "EBN͞Μ ͓ئ͍͠·͢
·ͱΊ • ελʔτΞοϓͰ DeepLearning ΑΓ ඇ Deep ͳख๏ͷํཱ͕͔ͭ • ෭ۀΛ͢Δͱ৬ܦྺॻ͕Εͯ
స৬׆ಈʹཱ͔ͭ • ຊۀͷٕज़͕෭ۀʹɺ෭ۀͷٕज़͕ຊۀʹ ཱͭ͜ͱ͕͋Δ
EOP