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
多変量正規分布に従う確率変数の条件付き期待値・分散
Search
Sponsored
·
Your Podcast. Everywhere. Effortlessly.
Share. Educate. Inspire. Entertain. You do you. We'll handle the rest.
→
axjack
January 11, 2022
Science
1k
0
Share
多変量正規分布に従う確率変数の条件付き期待値・分散
多変量正規分布に従う確率変数の条件付き期待値・分散
axjack
January 11, 2022
More Decks by axjack
See All by axjack
実験計画法_フィッシャーの3原則
axjack
0
490
統計学実践ワークブック 第16章 重回帰分析 pp.125-127
axjack
0
3.1k
統計学実践ワークブック 第15章 確率過程の基礎 p.117のεiの分布を導出する
axjack
0
980
第14章マルコフ連鎖
axjack
0
150
修正項を用いて繰り返しのある二元配置分散分析の分散分析表を完成させる
axjack
0
350
Other Decks in Science
See All in Science
タンパク質間相互作⽤を利⽤した⼈⼯知能による新しい薬剤遺伝⼦-疾患相互作⽤の同定
tagtag
PRO
0
180
Accelerating operator Sinkhorn iteration with overrelaxation
tasusu
0
260
[Paper Introduction] From Bytes to Ideas:Language Modeling with Autoregressive U-Nets
haruumiomoto
0
220
コミュニティサイエンスの実践@日本認知科学会2025
hayataka88
0
150
Hakonwa-Quaternion
hiranabe
1
200
Text-to-SQLの既存の評価指標を問い直す
gotalab555
1
190
Performance Evaluation and Ranking of Drivers in Multiple Motorsports Using Massey’s Method
konakalab
0
170
データマイニング - コミュニティ発見
trycycle
PRO
0
230
シャボン玉の虹から原子も地震も重力も見える! 〜 物理の目「干渉縞」のすごい力 〜
syotasasaki593876
1
110
ハミルトン・ヤコビ方程式の解の性質と物理的意味
enakai00
0
190
データマイニング - ノードの中心性
trycycle
PRO
0
360
Optimization of the Tournament Format for the Nationwide High School Kyudo Competition in Japan
konakalab
0
170
Featured
See All Featured
Ruling the World: When Life Gets Gamed
codingconduct
0
190
Data-driven link building: lessons from a $708K investment (BrightonSEO talk)
szymonslowik
1
990
Designing Dashboards & Data Visualisations in Web Apps
destraynor
231
54k
DevOps and Value Stream Thinking: Enabling flow, efficiency and business value
helenjbeal
1
160
The untapped power of vector embeddings
frankvandijk
2
1.7k
First, design no harm
axbom
PRO
2
1.2k
How to Talk to Developers About Accessibility
jct
2
170
Groundhog Day: Seeking Process in Gaming for Health
codingconduct
0
140
Effective software design: The role of men in debugging patriarchy in IT @ Voxxed Days AMS
baasie
0
280
Templates, Plugins, & Blocks: Oh My! Creating the theme that thinks of everything
marktimemedia
31
2.7k
Impact Scores and Hybrid Strategies: The future of link building
tamaranovitovic
0
250
The #1 spot is gone: here's how to win anyway
tamaranovitovic
2
1k
Transcript
ଟมྔਖ਼نʹै͏֬มͷ ͖݅ظɾࢄ 4BUPBLJ/PHVDIJ BYKBDL!HNBJMDPN  1
ͱ͠ɺ9ฏۉЖɾࢄڞࢄߦྻЄ ͷଟมྔਖ਼ن ʹै͏ͱ͢Δɻ ͜͜Ͱɺ ɹɾ9Λׂ̎ ɹɾЖΛׂ̎ ɹɾЄΛׂ̐ ͓ͯ͘͠ɻ ४උ Λ֬มϕΫτϧ
ΛظϕΫτϧ Λࢄڞࢄߦྻ Σ = ( Σ11 Σ12 Σ21 Σ22 ) X μ Σ X = ( X1 X2 ) μ = ( μ1 μ2 ) X ∼ N(μ, Σ) μi = E[Xi ] ͨͩ͠ Σij = Cov[Xi , Xj ] ͨͩ͠  2
ެࣜ ͖݅֬มͷ ظɾࢄ E[X1 |X2 = x2 ] = μ1
+ Σ12 Σ22 −1(x2 − μ2 ) V[X1 |X2 = x2 ] = Σ11 − Σ12 Σ22 −1Σ21 X1 |X2 = x2 Λɺ9YͰ͚݅ͮͨ9ͷ֬มͱ͢Δɻ ͜ͷ࣌ɺ9c9YͷظɾࢄҎԼͰ͋Δɻ ˞ࢀߟɿʰຊ౷ܭֶձެࣜೝఆɹ౷ܭݕఆ̍ڃରԠɹ౷ܭֶʱຊ౷ܭֶձɹฤ Qఆཧ  3
ྫ ( X Y Z ) ∼ N (( 1
2 3 ) , ( 2 0 1 0 3 2 1 2 4 )) ( X Y Z ) ̏มྔ֬ม ̏มྔਖ਼ن ʹै͏ͱ͢Δɻ ͜ͷ࣌ɺ Z|X = x, Y = y X, Y|Z = z ʹ͓͚ΔɺظɾࢄΛٻΊΑɻ ˞ࢀߟ౷ܭݕఆ४̍ڃ݄  4
ͷղ μ = ( 3 1 2 ) Σ
= ( 4 1 2 1 2 0 2 0 3 ) μ1 = E[Z] μ2 = E[(X Y)′  ] Σ11 Σ12 Σ22 Σ21 ( X1 X2 ) ∼ N (( μ1 μ2 ), ( Σ11 Σ12 Σ21 Σ22 )) E[X1 |X2 = x2 ] = μ1 + Σ12 Σ22 −1(x2 − μ2 ) V[X1 |X2 = x2 ] = Σ11 − Σ12 Σ22 −1Σ21 ( Z X Y ) ∼ N (( 3 1 2 ) , ( 4 1 2 1 2 0 2 0 3 )) ΑΓɺ E[Z|(X = x, Y = y)] = μ1 + Σ12 Σ22 −1 ( x − 1 y − 2) = 3 + (1 2) ( 2 0 0 3) −1 ( x − 1 y − 2) V[Z |(X = x, Y = y)] = Σ11 − Σ12 Σ22 −1Σ21 = 4 − (1 2) ( 2 0 0 3) −1 ( 1 2)  5
ͷղ μ = ( 1 2 3 ) Σ
= ( 2 0 1 0 3 2 1 2 4 ) μ1 = E[(X Y)′  ] μ2 = E[Z] Σ11 Σ12 Σ22 Σ21 ( X1 X2 ) ∼ N (( μ1 μ2 ), ( Σ11 Σ12 Σ21 Σ22 )) E[X1 |X2 = x2 ] = μ1 + Σ12 Σ22 −1(x2 − μ2 ) V[X1 |X2 = x2 ] = Σ11 − Σ12 Σ22 −1Σ21 ( X Y Z ) ∼ N (( 1 2 3 ) , ( 2 0 1 0 3 2 1 2 4 )) ΑΓɺ E[X, Y |Z = z] = ( 1 2) + ( 1 2) 4−1 (z − 3) V[X, Y |Z = z] = Σ11 − Σ12 Σ22 −1Σ21 = ( 2 0 0 3) − ( 1 2) 4−1 (1 2)  6