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

머신러닝을 위한 기초 수학 살펴보기

머신러닝을 위한 기초 수학 살펴보기

머신러닝을 위한 매우 기초적인 수학과 이를 응용한 선형 회귀 학습 예제

mingrammer

August 11, 2017
Tweet

More Decks by mingrammer

Other Decks in Technology

Transcript

  1. Name ӂ޹੤ (MinJae Kwon) Nickname @mingrammer Email [email protected] Who Game

    Server Engineer @ SundayToz Blog https://mingrammer.com Facebook https://facebook.com/mingrammer Github https://github.com/mingrammer Eng Blog https://medium.com/@mingrammer
  2. 2. ࢶഋ؀ࣻ೟੉ۆ? 4. ୶о ӝୡ ࣻ೟ ࢓ಝࠁӝ Contents 5. ࢶഋ

    ഥӈ ҳഅ೧ࠁӝ 1. ࣻ೟੄ ೙ਃࢿ 3. NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ 6. Next (more LA and Mathematics)
  3. ࣻ೟੄ ೙ਃࢿ ਢ ѐߊ জ ѐߊ API ѐߊ ࣻ೟ T(x)

    ∂ ∂θ f (x,θ)dx ∫ −log(t)y(t) ∑ x∇f (x) 1 σ 2π e −(x−µ)2 2σ 2 −∞ ∞ ∫
  4. ࣻ೟੄ ೙ਃࢿ ਢ ѐߊ জ ѐߊ API ѐߊ ࣻ೟ *UEFQFOETPOjCVU

    T(x) ∂ ∂θ f (x,θ)dx ∫ −log(t)y(t) ∑ x∇f (x) 1 σ 2π e −(x−µ)2 2σ 2 −∞ ∞ ∫
  5. ࣻ೟੄ ೙ਃࢿ ݠन۞׬ ߑޙ੗ࣻ҅ஏ ୶ୌঌҊ્ܻ ঐഐࢸ҅ ঑୷ঌҊ્ܻ %ݽ؛݂ ѱ੐ূ૓ ୭ࣗ࠺ਊঌҊ્ܻ

    ೐۽ࣁझझாે݂ Ӓې೗୊ܻ ഛܫݽ؛ ҊബਯҊࢿמ҅࢑ࢸ҅ ؘ੉ఠ߬੉झ नഐ୊ܻ
  6. ࣻ೟੄ ೙ਃࢿ 6 7 8 ... 5 7 5 6

    ... 3 8 4 1 ... 4 ... ... ... ... 2 0 1 5 4 3 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ W = W −α ∂L ∂W 0 0 1 ... 0 1 0 0 ... 0 0 1 0 ... 0 ... ... ... ... 0 0 0 0 0 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 0 1 0 ... 0 1 0 0 ... 0 0 1 0 ... 0 ... ... ... ... 0 0 0 1 0 0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 4 1 5 ... 8 ⎡ ⎣ ⎤ ⎦ 1 2 (y k − t k )2 ∑ y k = eak eai ∑ 0.5 0.4 ... ... 0.8 0.1 −0.3 ... ... 0.2 ... ... ... ... 0.1 ... ... ... ... 0.43 0.03 0.23 0.1 0.3 0.13 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 0 1 0 ... 0 ⎡ ⎣ ⎤ ⎦ Y = X ⋅W + B
  7. ࣻ೟੄ ೙ਃࢿ 6 7 8 ... 5 7 5 6

    ... 3 8 4 1 ... 4 ... ... ... ... 2 0 1 5 4 3 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ W = W −α ∂L ∂W 0 0 1 ... 0 1 0 0 ... 0 0 1 0 ... 0 ... ... ... ... 0 0 0 0 0 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 0 1 0 ... 0 1 0 0 ... 0 0 1 0 ... 0 ... ... ... ... 0 0 0 1 0 0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 4 1 5 ... 8 ⎡ ⎣ ⎤ ⎦ 1 2 (y k − t k )2 ∑ y k = eak eai ∑ 0.5 0.4 ... ... 0.8 0.1 −0.3 ... ... 0.2 ... ... ... ... 0.1 ... ... ... ... 0.43 0.03 0.23 0.1 0.3 0.13 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ Y = X ⋅W + B ࢶഋ؀ࣻ೟ ࢶഋ؀ࣻ೟ ࢶഋ؀ࣻ೟ ഛܫҗా҅ ഛܫҗా҅ ޷੸࠙೟ 0 1 0 ... 0 ⎡ ⎣ ⎤ ⎦
  8. ࣻ೟੄ ೙ਃࢿ 6 7 8 ... 5 7 5 6

    ... 3 8 4 1 ... 4 ... ... ... ... 2 0 1 5 4 3 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ W = W −α ∂L ∂W 0 0 1 ... 0 1 0 0 ... 0 0 1 0 ... 0 ... ... ... ... 0 0 0 0 0 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 0 1 0 ... 0 1 0 0 ... 0 0 1 0 ... 0 ... ... ... ... 0 0 0 1 0 0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 4 1 5 ... 8 ⎡ ⎣ ⎤ ⎦ 1 2 (y k − t k )2 ∑ y k = eak eai ∑ 0.5 0.4 ... ... 0.8 0.1 −0.3 ... ... 0.2 ... ... ... ... 0.1 ... ... ... ... 0.43 0.03 0.23 0.1 0.3 0.13 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ Y = X ⋅W + B ഛܫҗా҅ ޷੸࠙೟ ࢶഋ؀ࣻ೟ ؘ੉ఠ಴അ ҅࢑੄ബਯࢿ ୭੸ച ౠࢿ୶୹ ഛܫ࠙ನ ୶ۿ߂৘ஏ оࢸѨૐ ೧ࢳ೟੸੽Ӕ ӝ਎ӝ҅࢑ ಞ޷࠙ ੿ӏച 0 1 0 ... 0 ⎡ ⎣ ⎤ ⎦
  9. ࢶഋ؀ࣻ೟੉ۆ? ߭ఠҕр x 11 x 12 x 13 x 14

    x 15 x 21 x 22 x 23 x 24 x 25 x 31 x 32 x 33 x 34 x 35 x 41 x 42 x 43 x 44 x 45 x 51 x 52 x 53 x 54 x 55 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ T y 1 y 2 y 3 y 4 y 5 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ −3 −2 6 ⎡ ⎣ ⎤ ⎦ 2 2 8 ⎡ ⎣ ⎤ ⎦ ߭ఠো࢑ ೯۳ো࢑ ରਗ ࢶഋߑ੿ध ݠन۞׬ ঐഐച ੿ࠁѨ࢝ ੉޷૑೐۽ࣁय ؀ӏݽ୊ܻ
  10. ਋ܻח ࢶഋ؀ࣻܳ ׮ܖӝ ਤ೧ Numpyܳ ࢎਊ೤פ׮   ৈӝח౵੉௑੉૑݅ ࣽࣻ1ZUIPO਷ખוܿ

     /VNQZח௏যо$'PSUSBOӝ߈੉ۄࡅܴ  ࡅܳࡺ݅ইפۄߓৌਸബਯ੸ਵ۽୊ܻ೧ݫݽܻب؏ࢎਊೣ  пઙಞܻೠҊࣻળੋఠಕ੉झ৬بҳٜਸઁҕ NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ
  11. NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ ߭ఠ (vector) x = x 1 x

    2 ... x n ⎡ ⎣ ⎤ ⎦ x = x 1 x 2 ... x n ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ೯߭ఠ SPXWFDUPS ৌ߭ఠ DPMVNOWFDUPS
  12. NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ ߭ఠ ো࢑ ↟ؔࣅ BEEJUJPO  ↟ࡓࣅ TVCUSBDUJPO

     ↟झணۄғ 4DBMBS1SPEVDU  ↟ࢿ࠙ғ &MFNFOUXJTF.VMUJQMJDBUJPO  ↟ղ੸ *OOFS1SPEVDU ߭ఠח׮਺੄ো࢑ٜਸࣻ೯ೡࣻ੓਺
  13. NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ ߭ఠ ো࢑ 1 ↟ؔࣅ BEEJUJPO  ↟ࡓࣅ

    TVCUSBDUJPO v + u = (v 1 + u 1 ,...,v n + u n ) v − u = (v 1 − u 1 ,...,v n − u n )
  14. NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ ߭ఠ ো࢑ 2 ↟झணۄғ 4DBMBS1SPEVDU  ↟ࢿ࠙ғ

    &MFNFOUXJTF.VMUJQMJDBUJPO  ↟ղ੸ *OOFS1SPEVDU v⊗u = (v 1 u 1 ,...,v n u n ) av = (av 1 ,...,av n ) v⋅u = v i u i i=1 n ∑
  15. NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ ೯۳ ো࢑ ↟ؔࣅ BEEJUJPO  ↟ࡓࣅ TVCUSBDUJPO

     ↟झணۄғ 4DBMBS1SPEVDU  ↟ࢿ࠙ғ &MFNFOUXJTF.VMUJQMJDBUJPO  ↟೯۳ғ .BUSJY.VMUJQMJDBUJPO ೯۳਷׮਺੄ো࢑ٜਸࣻ೯ೡࣻ੓਺
  16. NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ ೯۳ ো࢑ 1 ↟ؔࣅ BEEJUJPO  ↟ࡓࣅ

    TVCUSBDUJPO X +Y = x 11 + y 11 ... x 1n + y 1n ... ... ... x m1 + y m1 ... x mn + y mn ⎡ ⎣ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ X −Y = x 11 − y 11 ... x 1n − y 1n ... ... ... x m1 − y m1 ... x mn − y mn ⎡ ⎣ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥
  17. ೯۳ ো࢑ 2 ↟झணۄғ 4DBMBS1SPEVDU  ↟ࢿ࠙ғ &MFNFOUXJTF.VMUJQMJDBUJPO aX =

    ax 11 ... ax 1n ... ... ... ax m1 ... ax mn ⎡ ⎣ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ X ⊗Y = x 11 y 11 ... x 1n y 1n ... ... ... x m1 y m1 ... x mn y mn ⎡ ⎣ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ
  18. ೯۳ ো࢑ 3 x 11 x 12 x 13 x

    14 x 21 x 22 x 23 x 24 x 31 x 32 x 33 x 34 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ೯۳ғ .BUSJY.VMUJQMJDBUJPO = y 11 y 12 y 13 y 21 y 22 y 23 y 31 y 32 y 33 y 41 y 42 y 43 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ x 11 y 11 + x 12 y 21 + x 13 y 31 + x 14 y 41 ... ... ... ... ... ... ... ... ⎡ ⎣ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ Y Y Y = Y NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ
  19. ੹஖ (Transpose) NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ XT = x 11 x

    12 x 13 x 14 x 21 x 22 x 23 x 24 x 31 x 32 x 33 x 34 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ T = x 11 x 12 x 13 x 21 x 22 x 23 x 31 x 32 x 33 x 41 x 42 x 43 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ xT = x 1 x 2 ... x n ⎡ ⎣ ⎤ ⎦ T = x 1 x 2 ... x n ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥
  20. ױਤ ೯۳ (Identity Matrix) NumPy۽ ࢶഋ؀ࣻ೟ ׮ܞࠁӝ I n =

    1 0 0 ... 0 0 1 0 ... 0 0 0 1 ... 0 ... ... ... ... 0 0 0 0 0 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥
  21. ৉೯۳ (Inverse Matrix) XX−1 = I n = X−1X NumPy۽

    ࢶഋ؀ࣻ೟ ׮ܞࠁӝ ৉೯۳਷ ೦࢚ ઓ੤ೞ૑ח ঋ਺
  22. ୶о ӝୡ ࣻ೟ ࢓ಝࠁӝ ੿ӏച 1 y i = x

    i x i ∑ ੹୓ sumਵ۽ ա׃ਵ۽ॄ [0, 1] ҳрਵ۽ ੿ӏച
  23. ӝఋ ӝୡ ࣻ೟ ࢓ಝࠁӝ ੿ӏച 2 Xi = X i

    − E(X) σ E(X) = 1 N X i i=1 N ∑ σ = 1 N (X i − E(X))2 i=1 N ∑ ಴ળ ੿ӏ ࠙ನܳ ഝਊ೧ ಣӐ = 0, ࠙࢑ = 1۽ ੿ӏച
  24. ӝఋ ӝୡ ࣻ೟ ࢓ಝࠁӝ ޷࠙ ߂ Ӓۄ٣঱౟ 1 2 3

    0 0 1 ... 0 1 0 0 ... 0 0 1 0 ... 0 ... ... ... ... 0 0 0 0 0 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 0 1 0 ... 0 1 0 0 ... 0 0 1 0 ... 0 ... ... ... ... 0 0 0 1 0 0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ৘ஏ ݽ؛ীࢲ੄ Ѿҗчҗ पઁ Ѿҗчҗ੄ рӓ੄ ૑಴ੋ ର੉ “ࣚप"ਸ ઴੉ӝ ਤೣ ࣚप ೣࣻ Ӓې೐
  25. ӝఋ ӝୡ ࣻ೟ ࢓ಝࠁӝ ಞ޷࠙ f (x) = g(x 1

    )+...+ g(x n ) ∂ f (x) ∂x 1 = ∂g(x 1 ) ∂x 1 ׮߸ࣻ ҕр ഑਷ ೣࣻীࢲ ౠ੿ ߸ࣻܳ ؀࢚ਵ۽ ޷࠙ ݠन ۞׬ীࢶ ࠁా ೯۳ਸ ؀࢚ਵ۽ೠ ೯۳ ಞ޷࠙ ࢎਊ ∂L ∂X = ∂L ∂x 11 ... ∂L ∂x 1n ... ... ... ∂L ∂x m1 ... ∂L ∂x mn ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥
  26. ࢶഋ ഥӈ ҳഅ೧ࠁӝ ୭੸੄ ࢶഋ ࢚ҙ ҙ҅ܳ ಴അೞח ૒ࢶਸ ଺ਵ۰Ҋ

    ೣ → ୭੸੄ ߬ఋܳ ೟ण y = β 0 x 0 + β 1 x 1 y i = x i T β
  27. ࢶഋ ഥӈ ҳഅ೧ࠁӝ y = β 0 x 0 +

    β 1 x 1 y i = x i0 β 0 + x i1 β 1 ױੌؘ੉Tఠ ׮઺ؘ੉ఠ઺Jߣ૩ؘ੉ఠ
  28. ࢶഋ ഥӈ ҳഅ೧ࠁӝ y = β 0 x 0 +

    β 1 x 1 y i = x i T β y i = x i0 β 0 + x i1 β 1 ױੌؘ੉ఠ ׮઺ؘ੉ఠ઺Jߣ૩ؘ੉ఠ Jߣ૩ؘ੉ఠ߭ఠ಴അ
  29. ࢶഋ ഥӈ ҳഅ೧ࠁӝ Y = Xβ Y = y 1

    y 2 ... y n ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ X = x 1 T x 2 T ... x n T ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ β = β 0 β 1 ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ੌ߈ചػ׮઺ؘ੉ఠ੄೯۳಴അ
  30. ࢶഋ ഥӈ ҳഅ೧ࠁӝ 1.0 1.1 1.2 ... 10.0 ⎡ ⎣

    ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥
  31. ࢶഋ ഥӈ ҳഅ೧ࠁӝ 1.0 1.1 1.2 ... 10.0 ⎡ ⎣

    ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 0.1 0.11 0.12 ... 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥
  32. ࢶഋ ഥӈ ҳഅ೧ࠁӝ y = ax + b = b

    + ax y 1 = b + ax 1 y 2 = b + ax 2
  33. ࢶഋ ഥӈ ҳഅ೧ࠁӝ 1 0.1 1 0.11 1 0.12 1

    ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥
  34. ࢶഋ ഥӈ ҳഅ೧ࠁӝ r 1 r 2 r 3 ...

    r 100 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 1 0.1 1 0.11 1 0.12 1 ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥
  35. ࢶഋ ഥӈ ҳഅ೧ࠁӝ r 1 r 2 r 3 ...

    r 100 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 1 x i1 1 x i2 1 x i3 ... ... 1 x i20 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 1 0.1 1 0.11 1 0.12 1 ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥
  36. ࢶഋ ഥӈ ҳഅ೧ࠁӝ Ӓۄ٣঱౟ ࢸ҅ 1 0.1 1 0.11 1

    0.12 ... ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ w 0 w 1 ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥
  37. ࢶഋ ഥӈ ҳഅ೧ࠁӝ Ӓۄ٣঱౟ ࢸ҅ 1 0.1 1 0.11 1

    0.12 ... ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ w 0 w 1 ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ y 1 y 2 y 3 ... y 20 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ − w 0 + 0.1w 1 w 0 + 0.11w 1 w 0 + 0.12w 1 ... w 0 +1.0w 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎛ ⎝ ⎜ ⎜ ⎜ ⎜ ⎜ ⎜ ⎞ ⎠ ⎟ ⎟ ⎟ ⎟ ⎟ ⎟ T
  38. ࢶഋ ഥӈ ҳഅ೧ࠁӝ Ӓۄ٣঱౟ ࢸ҅ 1 0.1 1 0.11 1

    0.12 ... ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ w 0 w 1 ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ y 1 y 2 y 3 ... y 20 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ − w 0 + 0.1w 1 w 0 + 0.11w 1 w 0 + 0.12w 1 ... w 0 +1.0w 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎛ ⎝ ⎜ ⎜ ⎜ ⎜ ⎜ ⎜ ⎞ ⎠ ⎟ ⎟ ⎟ ⎟ ⎟ ⎟ T 1 20 error i 2 i=1 20 ∑ = 1 20 (y i − e i )2 i=1 20 ∑
  39. ࢶഋ ഥӈ ҳഅ೧ࠁӝ Ӓۄ٣঱౟ ࢸ҅ 1 0.1 1 0.11 1

    0.12 ... ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ w 0 w 1 ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ y 1 y 2 y 3 ... y 20 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ − w 0 + 0.1w 1 w 0 + 0.11w 1 w 0 + 0.12w 1 ... w 0 +1.0w 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎛ ⎝ ⎜ ⎜ ⎜ ⎜ ⎜ ⎜ ⎞ ⎠ ⎟ ⎟ ⎟ ⎟ ⎟ ⎟ T 1 20 error i 2 i=1 20 ∑ = 1 20 (y i − e i )2 i=1 20 ∑ ∂(mse) ∂W = [error][x]
  40. ࢶഋ ഥӈ ҳഅ೧ࠁӝ Ӓۄ٣঱౟ ࢸ҅ ∂(mse) ∂w 0 = ∂(

    (y i − (w 0 + x i w 1 ))2 ) ∑ ∂w 0 = (y i − (w 0 + x i w 1 ))⋅1 ∑ ∂(mse) ∂w 1 = ∂( (y i − (w 0 + x i w 1 ))2 ) ∑ ∂w 1 = (y i − (w 0 + x i w 1 ))⋅ x i ∑
  41. ࢶഋ ഥӈ ҳഅ೧ࠁӝ Ӓۄ٣঱౟ ࢸ҅ ∂(mse) ∂W = [ y

    i − (w 0 + x i w 1 ) ∑ ] 1 0.1 1 0.11 1 0.12 ... ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ = [error][x]
  42. ࢶഋ ഥӈ ҳഅ೧ࠁӝ ୭ઙ ೟ण Ѿҗ &SSPS β = −2.152736

    11.075026 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ y k = x k T β →
  43. Next (more LA and Mathematics) ↟઱ࢿ࠙࠙ࢳ 1$"  ↟ױੌч࠙೧ 47%

     ↟-6࠙೧ ↟ҊਬчҊਬ߭ఠ ↟಴ળച ↟j ↟߬੉૑উా҅ ↟ഛܫӏ஗ ↟ࢠ೒݂ߑध ↟୭؀਋بஏ੿ ↟j ↟೯۳޷੸࠙೟ ↟Ӓۄ٣঱౟ ↟j