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
teratail_study
Search
tkow
December 19, 2016
Technology
0
360
teratail_study
first step of understanding machine_learning
tkow
December 19, 2016
Tweet
Share
More Decks by tkow
See All by tkow
React NativeでCross Platform対応を頑張る話
tkow
1
690
Flipperを使ってみた rntb in Japan 2020/05/27
tkow
1
250
開発環境の切り分け方のベストプラクティスが分からない -andfactory20190917-
tkow
0
200
react-beginner
tkow
0
710
tkow-Angular2入門者の会発表資料
tkow
1
13k
Other Decks in Technology
See All in Technology
TypeScript開発にモジュラーモノリスを持ち込む
sansantech
PRO
2
700
小学3年生夏休みの自由研究「夏休みに Copilot で遊んでみた」
taichinakamura
0
190
なぜCodeceptJSを選んだか
goataka
0
180
Server-Side Engineer of LINE Sukimani
lycorp_recruit_jp
0
370
KnowledgeBaseDocuments APIでベクトルインデックス管理を自動化する
iidaxs
1
280
型情報を用いたLintでコード品質を向上させる
sansantech
PRO
2
150
AWS re:Invent 2024 ふりかえり勉強会
yhana
0
500
エンジニアカフェ忘年会2024「今年やらかしてしまったこと!」
keropiyo
0
110
Work as an App Engineer
lycorp_recruit_jp
0
380
効率的な技術組織が作れる!書籍『チームトポロジー』要点まとめ
iwamot
2
120
watsonx.ai Dojo #5 ファインチューニングとInstructLAB
oniak3ibm
PRO
0
200
株式会社ログラス − エンジニア向け会社説明資料 / Loglass Comapany Deck for Engineer
loglass2019
3
32k
Featured
See All Featured
Art, The Web, and Tiny UX
lynnandtonic
298
20k
Designing for Performance
lara
604
68k
The Power of CSS Pseudo Elements
geoffreycrofte
73
5.4k
Side Projects
sachag
452
42k
The Art of Delivering Value - GDevCon NA Keynote
reverentgeek
8
1.2k
RailsConf & Balkan Ruby 2019: The Past, Present, and Future of Rails at GitHub
eileencodes
132
33k
Understanding Cognitive Biases in Performance Measurement
bluesmoon
26
1.5k
Building a Modern Day E-commerce SEO Strategy
aleyda
38
7k
Java REST API Framework Comparison - PWX 2021
mraible
28
8.3k
The Web Performance Landscape in 2024 [PerfNow 2024]
tammyeverts
2
290
JavaScript: Past, Present, and Future - NDC Porto 2020
reverentgeek
47
5.1k
The Success of Rails: Ensuring Growth for the Next 100 Years
eileencodes
44
6.9k
Transcript
ػցֶशʹຊ֨తೖ͢Δ࣌ ʹ࠷ॳʹԡ͓͖͍͑ͯͨ͜͞ͱ !ULPX
ࣗݾհ ૣҴాେֶઌਐཧֶݚڀՊम࢜ ͖ͬͯͨ͜ͱ ࣗવݴޠॲཧɾ҉߸ཧͳͲ ϨόϨδʔζʹདྷ݄̐ೖࣾ༧ఆ
ϑϧεΫϥονͰΞϧΰϦζϜΛ࣮͍ͯ͠·͢
ػցֶशͰ Կ͕Ͱ͖Δ͔ΛΔ
ҟৗݕ ղ૾ ࠷దઓུ ࣝผ ମೝࣝ ༧ଌ ΫϥελϦϯά ͳͲͳͲʜ
ͦͦԿΛػցʹ͍ͤͯΔͷ͔ ୡͯ͠ཉ͍͠తΛϓϩάϥϛϯά͢Δɻ ίϯϐϡʔλతΛઃఆ͢Δ͜ͱ͕ۤखͳͷͰ ਓ͕ؒࣜͱͯ͠ઃܭͯ͋͛͠Δɻ ਓ͕ؒڭ͑ͳֶͯ͘ͿڧԽֶश༗Γ·͕͢ ༩͑Δਓ͕ؒղ͚ͳͯ͘ྑ͍ɻ ݟ͚͍ͭͨؔੑʹ͍ͭͯίϯϐϡʔλ͕উखʹ ղ͍ͯ͘ΕΔɻ ղ͚ͳ͍͜ͱ͋Δ͚Ͳ
ԿΛษڧ͢Εྑ͍ͷ͔ ϕΠζͷఆཧ ઢܗճؼϞσϧ ֬ ڞࣄલ ࢦ ΧςΰϦΧϧ
χϡʔϥϧωοτ ͜ΕΒΛ͔ͬ͠Γԡ͑͞Δ͚ͩͰݟ௨͕͠ྑ͘ͳΓ·͢ɻ ػցֶशͰ࠷ॳʹԡ͓͖͑ͯࣝ͘͞
ϕΠζͷఆཧ جຊతͳΈ p ( x | ✓ ) ԾఆВͷݩͰY͕ى͜Δ֬
ϕΠζͷఆཧ p(✓| x ) = p( x |✓)p(✓) R
p( x |✓)p(✓)d✓ ԾఆВ ͋ΔԾఆͷ ݩͰಘΒΕͨ σʔλY ಘΒΕͨσʔλ9͔ΒഎޙͷԾఆВΛ ਪఆ͢Δ جຊతͳΈ
ϕΠζͷఆཧ جຊతͳΈ ಘΒΕͨσʔλू߹ͷ͔Β࠷దͳΛਪఆ͢Δ ࠷ ਪఆ ࣄલΛԾఆͯ͠ಘΒΕͨσʔλू߹͔Βࣄલͷ࠷ దͳύϥϝʔλΛਪఆ͢Δ ."1ਪఆ
ಘΒΕͨσʔλू߹͔ΒύϥϝʔλΛपลԽ ϕΠζਪఆ p(✓| x ) = p( x |✓)p(✓) R p( x |✓)p(✓)d✓
ͳͥେࣄ͔ͲΜͳؔͰύϥϝʔλΛ૿ͤۙࣅՄ ೳɻҰԠղੳղ͕ग़ͤΔࣄ͕ଟ͍ ݱ࣮࣌ؒͰղ͚Δͱ ݴ͍ͬͯͳ͍ ɻ ඇઢܗͷϞσϧΛද͢ͷʹઢܗϞσϧ͕ΘΕΔ χϡʔϥ ϧωοτͰɺγάϞΠυؔͳͲͷύϥϝʔλΛઢܗϞ σϧʹಥͬࠐΜͰ͍Δ ɻ
ઢܗϞσϧ y = w0 + w1 1( x ) + · · · + wn n( x )
ϦϯΫؔ ઢܗϞσϧ f (x) = w0 + w1 1(x)
+ . . . + wn n(x) f (x) ϦϯΫؔͱɺઢܗճؼϞσϧΛΈ߹ΘͤΔͱɺϕΠζ ͷఆཧʹͯΊͯɺ࠷ਪఆ."1ਪఆ͕Ͱ͖Δɻ ྫ ɹϦϯΫؔʹMPHJUؔΛͯΊͯ࠷ਪఆΛղ͘ͱϩ δεςΟοΫճؼʹͳΔͳͲ
ͷಛΛԡ͑͞Δͱ͍ΖΜͳσʔλʹର͢Δ༧ଌਪ ఆʹʹཱͪ·͢ ֬ ϕϧψʔΠίΠϯͷදཪͷग़Δ֬ ϕʔλίΠϯͷදཪͷग़Δ͕֬ͲΕ͘Β͍ʹ ͳΔ͔ͷ֬ ΨϯϚਖ਼ͷͷग़͢͞ΛௐઅͰ͖Δ ଟ߲αΠίϩͷͷͲΕ͕ͰΔ͔ͷ σΟϦΫϨαΠίϩͷͷ͕֬ͲΕ͘Β͍ʹ
ͳΔ͔ͷ֬
ͷಛΛԡ͑͞Δͱ͍ΖΜͳσʔλʹର͢Δ༧ଌਪ ఆʹʹཱͪ·͢ ֬ ϕʔλ֬ͷग़͢͞Λௐઅ͢Δ֬
ڞࣄલಛఆͷؔͷੵʹରͯ͠ɺࣄޙͷࣜ ͷܗ͕ಉ͡ࣄલͷ͜ͱɻ ڞࣄલº͕ɺࣄޙͷܗͱಉ͡ʹͳΔ͜ͱΛ ར༻͢Δ͜ͱͰɺࣄޙฏۉͷެࣜΛద༻ͨ͠ΓɺϥϯμϜ αϯϓϦϯάͱݺΕΔख๏Λద༻͘͢͠Δ͜ͱ͕Ͱ͖ Δ σʔλΛਫ૿ͯ͠͠શମͷΛਪఆ͢Δํ๏ ɻ ֬
p ( ✓ | x ) / p ( x | ✓ ) p ( ✓ ) ࣄલ ࣄޙ ϕʔλ㲌ϕϧψʔΠºϕʔλͳͲ
جຊతͳΈ ͋Δϊʔυʹ͓͚ΔೖྗͷΛ׆ੑԽؔͱݺΕΔҙ ͷؔʹΑͬͯɺม͠ɺ࣍ͷೖྗ ग़ྗ ͱ͢Δ ͋ΔϊʔυʹมԽ͕ى͖ͨ࣌ʹશͯͷύϥϝʔλʹٴ΅͢ ӨڹΛࠩͷ࿈Ͱද͢ɻ χϡʔϥϧωοτ x1
x2 x3 f (x)=w1x1+w2x2+w3x3 y= ( f ( x )) y w1 w2 w3
None
None
None
None
༗໊ͳຊ
ࣗવݴޠॲཧ͓͢͢Ίຊ
࣮͓͢͢Ίຊ