Share
Embed
Transcript
67 slides
-
ݠन۞ਸ ਤೠ ӝୡ ࣻ ಝࠁӝ MinJae Kwon (@mingrammer) 2017.08.12 (Getting
started to learn the linear algebra with python) -
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. ࢶഋࣻۆ? 4. ୶о ӝୡ ࣻ ಝࠁӝ Contents 5. ࢶഋ
ഥӈ ҳഅ೧ࠁӝ 1. ࣻ ਃࢿ 3. NumPy۽ ࢶഋࣻ ܞࠁӝ 6. Next (more LA and Mathematics) -
ࣻ ਃࢿ ߣࣁ࣌۽ӝୡ ࢶഋ ࣻਸܖ۽ ࣻधٜࣼೞঋٜ࠙ਸ࢚ਵ۽פ W ←W −α ∂L
∂W E = − 1 N t nk log(y nk ) k ∑ n ∑ d(u ! u ! T ) = 2u ! -
ࣻ ਃ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 −∞ ∞ ∫ -
ࣻ ਃ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 −∞ ∞ ∫ -
ࣻ ਃࢿ ߑޙࣻ҅ஏ ୶ୌঌҊ્ܻ ঐഐࢸ҅ ୷ঌҊ્ܻ %ݽ؛݂ ѱূ ୭ࣗ࠺ਊঌҊ્ܻ ۽ࣁझझாે݂
Ӓېܻ ഛܫݽ؛ ҊബਯҊࢿמ҅ࢸ҅ ݠन۞ ؘఠ߬झ नഐܻ -
ࣻ ਃࢿ ݠन۞ ߑޙࣻ҅ஏ ୶ୌঌҊ્ܻ ঐഐࢸ҅ ୷ঌҊ્ܻ %ݽ؛݂ ѱূ ୭ࣗ࠺ਊঌҊ્ܻ
۽ࣁझझாે݂ Ӓېܻ ഛܫݽ؛ ҊബਯҊࢿמ҅ࢸ҅ ؘఠ߬झ नഐܻ -
ࣻ ਃ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 -
ࣻ ਃ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 ࢶഋࣻ ࢶഋࣻ ࢶഋࣻ ഛܫҗా҅ ഛܫҗా҅ ࠙ 0 1 0 ... 0 ⎡ ⎣ ⎤ ⎦ -
ࣻ ਃ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 ࠙ ࢶഋࣻ ؘఠഅ ҅ബਯࢿ ୭ച ౠࢿ୶ ഛܫ࠙ನ ୶ۿ߂ஏ оࢸѨૐ ೧ࢳӔ ӝӝ҅ ಞ࠙ ӏച 0 1 0 ... 0 ⎡ ⎣ ⎤ ⎦ -
ࢶഋࣻۆ? ߭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 ࢶഋߑध ݠन۞ ঐഐച ࠁѨ࢝ ۽ࣁय ӏݽܻ -
ܻח ࢶഋࣻܳ ܖӝ ਤ೧ Numpyܳ ࢎਊפ ৈӝח݅ ࣽࣻ1ZUIPOખוܿ
/VNQZחযо$'PSUSBOӝ߈ۄࡅܴ ࡅܳࡺ݅ইפۄߓৌਸബਯਵ۽ܻ೧ݫݽܻب؏ࢎਊೣ пઙಞܻೠҊࣻળੋఠಕझ৬بҳٜਸઁҕ NumPy۽ ࢶഋࣻ ܞࠁӝ -
NumPy۽ ࢶഋࣻ ܞࠁӝ ߭ఠ (vector) x = x 1 x
2 ... x n ⎡ ⎣ ⎤ ⎦ x = x 1 x 2 ... x n ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ೯߭ఠ SPXWFDUPS ৌ߭ఠ DPMVNOWFDUPS -
NumPy۽ ࢶഋࣻ ܞࠁӝ ߭ఠ ো ↟ؔࣅ BEEJUJPO ↟ࡓࣅ TVCUSBDUJPO
↟झணۄғ 4DBMBS1SPEVDU ↟ࢿ࠙ғ &MFNFOUXJTF.VMUJQMJDBUJPO ↟ղ *OOFS1SPEVDU ߭ఠחোٜਸࣻ೯ೡࣻ -
NumPy۽ ࢶഋࣻ 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 ) -
NumPy۽ ࢶഋࣻ 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 ∑ -
NumPy۽ ࢶഋࣻ ܞࠁӝ ೯۳ ো ↟ؔࣅ BEEJUJPO ↟ࡓࣅ TVCUSBDUJPO
↟झணۄғ 4DBMBS1SPEVDU ↟ࢿ࠙ғ &MFNFOUXJTF.VMUJQMJDBUJPO ↟೯۳ғ .BUSJY.VMUJQMJDBUJPO ೯۳োٜਸࣻ೯ೡࣻ -
NumPy۽ ࢶഋࣻ 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 ⎡ ⎣ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ -
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۽ ࢶഋࣻ ܞࠁӝ -
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۽ ࢶഋࣻ ܞࠁӝ -
(Transpose) NumPy۽ ࢶഋࣻ 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 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ -
ױਤ ೯۳ (Identity Matrix) NumPy۽ ࢶഋࣻ ܞࠁӝ I n =
1 0 0 ... 0 0 1 0 ... 0 0 0 1 ... 0 ... ... ... ... 0 0 0 0 0 1 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ -
೯۳ (Inverse Matrix) XX−1 = I n = X−1X NumPy۽
ࢶഋࣻ ܞࠁӝ ೯۳ ೦࢚ ઓೞח ঋ -
୶о ӝୡ ࣻ ಝࠁӝ ӏച 1 y i = x
i x i ∑ sumਵ۽ ա׃ਵ۽ॄ [0, 1] ҳрਵ۽ ӏച -
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 -
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 -
ӝఋ ӝୡ ࣻ ಝࠁӝ ಞ࠙ 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 ಞ࠙ ࢎਊ ∂L ∂X = ∂L ∂x 11 ... ∂L ∂x 1n ... ... ... ∂L ∂x m1 ... ∂L ∂x mn ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ -
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 -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ y = β 0 x 0 +
β 1 x 1 ױੌؘఠ -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ y = β 0 x 0 +
β 1 x 1 y i = x i0 β 0 + x i1 β 1 ױੌؘTఠ ؘఠJߣ૩ؘఠ -
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 ױੌؘఠ ؘఠJߣ૩ؘఠ Jߣ૩ؘఠ߭ఠഅ -
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 ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ੌ߈ചػؘఠ೯۳അ -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ 1.0 1.1 1.2 ... 10.0 ⎡ ⎣
⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ 1.0 1.1 1.2 ... 10.0 ⎡ ⎣
⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 0.1 0.11 0.12 ... 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ y = ax + b = b
+ ax y 1 = b + ax 1 y 2 = b + ax 2 -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ 1 0.1 1 0.11 1 0.12 1
... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ r 1 r 2 r 3 ...
r 100 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ 1 0.1 1 0.11 1 0.12 1 ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ -
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 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ Ӓۄ٣ ࢸ҅ 1 0.1 1 0.11 1
0.12 ... ... 1 1.0 ⎡ ⎣ ⎢ ⎢ ⎢ ⎢ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ ⎥ ⎥ ⎥ ⎥ w 0 w 1 ⎡ ⎣ ⎢ ⎢ ⎤ ⎦ ⎥ ⎥ -
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 -
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 ∑ -
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] -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ Ӓۄ٣ ࢸ҅ ∂(mse) ∂W ೯۳ಞ࠙→ ೯۳ п
ਗࣗী ೧ ಞ࠙ -
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 ∑ -
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] -
ࢶഋ ഥӈ ҳഅ೧ࠁӝ ୭ઙ ण Ѿҗ &SSPS β = −2.152736
11.075026 ⎡ ⎣ ⎢ ⎤ ⎦ ⎥ y k = x k T β → -
Next (more LA and Mathematics) ↟ࢿ࠙࠙ࢳ 1$" ↟ױੌч࠙೧ 47%
↟-6࠙೧ ↟ҊਬчҊਬ߭ఠ ↟ળച ↟j ↟߬উా҅ ↟ഛܫӏ ↟ࢠ݂ߑध ↟୭بஏ ↟j ↟೯۳࠙ ↟Ӓۄ٣ ↟j -
хࢎפ MinJae Kwon (@mingrammer) Getting started to learn the linear
algebra with python 2017.08.12