Overview (Gaussian Process) GPLVM GPDM 2 / 59
|
|
|
- よりとし かくはり
- 7 years ago
- Views:
Transcription
1 ( ) 1 / 59
2 Overview (Gaussian Process) GPLVM GPDM 2 / 59
3 (Gaussian Process) y x x y (regressor) D = { (x (n), y (n) ) } N, n=1 x (n+1) y (n+1), ( ) 3 / 59
4 (Gaussian Process) y x x y (regressor) D = { (x (n), y (n) ) } N, n=1 x (n+1) y (n+1), ( ) 3 / 59
5 (Gaussian Process) y x x y (regressor) D = { (x (n), y (n) ) } N, n=1 x (n+1) y (n+1), ( ) 4 / 59
6 y = w 0 + w 1 x 1 + w 2 x 2 + ϵ = (w 0 w 1 w 2 ) 1 +ϵ }{{} w T x 1 = w T x + ϵ x 2 }{{} x ŵ = (X T X) 1 X T y ( ), 5 / 59
7 (GLM) y = w 0 + w 1 x + w 2 x 2 + w 3 x 3 + ϵ (1) = (w 0 w 1 w 2 w 3 ) 1 +ϵ (2) }{{} w T x x 2 x 3 }{{} = w T ϕ(x) + ϵ ϕ(x) (3) ϕ(x)! 6 / 59
8 (GLM) (2) ϕ(x) = ( (x µ 1) 2 2σ 2, (x µ 2) 2 2σ 2,, (x µ K) 2 ) 2σ 2 (4),! µ = (µ 1, µ 2,, µ K ) 7 / 59
9 x y R y = f(x) x = (x 1,, x d ) R d y = f(x), x ϕ(x) y = w T ϕ(x) (5) ϕ(x) = (ϕ 1 (x), ϕ 2 (x),, ϕ H (x)) T = (1, x 1,, x d, x 2 1,, x 2 d )T w = (w 0, w 1,, w 2d ) T, y = w T ϕ(x) = w 0 + w 1 x w d x d + w d+1 x w 2d x 2 d. 8 / 59
10 GP (1) y (1) y (N), y = Φw (Φ : ) ϕ 1 (x (1) ) ϕ H (x (1) ) w 1. = ϕ 1 (x (2) ) ϕ H (x (2) ) w 2... ϕ 1 (x (N) ) ϕ H (x (N) ). y (1) y (2) y (N) w H y Φ w w p(w) = N(0, α 1 I), y = Φw, 0, yy T = (Φw) (Φw) T = Φ ww T Φ T (7) = α 1 ΦΦ T (6) 9 / 59
11 GP (2) p(y) = N(y 0, α 1 ΦΦ T ) (8), {x n } N n=1 (x 1, x 2,, x N ), y = (y 1, y 2,, y N ), p(y). =, K = α 1 ΦΦ T k(x, x ) = α 1 ϕ(x) T ϕ(x ) (9) k(x, x ) x x ; x y 10 / 59
12 GP (3), ϵ { y = w T ϕ(x) + ϵ = p(y f) = N(w T ϕ(x), β 1 I) (10) ϵ N(0, β 1 I) f = w T ϕ(x) p(y x) = p(y f)p(f x)df (11) = N(0, C) (12) Gaussian, C : C(x i, x j ) = k(x i, x j ) + β 1 δ(i, j). (13) GP, k(x, x ) α, β. 11 / 59
13 y 0 y x Gaussian: exp( (x x ) 2 /l) x Exponential: exp( x x /l) (OU process) y 0 y x x Periodic: exp( 2 sin 2 ( x x 2 )/l 2 ) Periodic(L): exp( 2 sin 2 ( x x 2 )/(10l) 2 ) 12 / 59
14 Correlated Gaussian K = 13 / 59
15 (2) Correlated Gaussian K = 14 / 59
16 (3) Correlated Gaussian K = 15 / 59
17 Infinite dimensional Gaussian, (x 1, x 2,, x n ) y = (y 1, y 2,, y n ), y. (x 1, x 2,, x n ), ( ). K K ij = k(x i, x j ) k. 16 / 59
18 RBF ϕ(x) = exp((x h) 2 /r 2 ) 1, h k(x, x ) = σ 2 H h=1 ϕ h (x)ϕ h (x ) (14) (x h)2 exp ( r 2 ) exp ( (x h) 2 r 2 ) dh (15) = πr 2 exp ( (x x ) 2 ) 2r 2 θ 1 exp ( (x x ) 2 ) θ 2 2 (16) (x, x ) RBF, RBF. θ 1, θ 2 17 / 59
19 GP y new y Gaussian, p(y new x new, X, y, θ) = p((y, ynew ) (X, x new ), θ) p(y X, θ) [ exp 1 2 ([y, K ynew ] k T k k ] 1 [ ] y y new y T K 1 y) (17) (18) (19) N(k T K 1 y, k k T K 1 k). (20) K = [k(x, x )]. k = (k(x new, x 1 ),, k(x new, x N )). 18 / 59
20 GP SVR, Ridge, ARD (Cohn+ 2013) ( ) k(x, x ) = σf 2 exp 1 (x k x k )2 2 σk 2 k (21) Model MAE RMSE µ SVM Linear ARD Squared exp. Isotropic Squared exp. ARD Rational quadratic ARD Matern(5,2) Neural network / 59
21 GP SVR, Ridge, ARD (Cohn+ 2013) ( ) k(x, x ) = σf 2 exp 1 (x k x k )2 2 σk 2 k (22) Model MAE RMSE µ Independent SVMs EasyAdapt SVM Independent Pooled Pooled &{N} Combined / 59
22 GP>SVR,, (Cohn etc.)! 21 / 59
23 GP GP : / X K 1 O(N 3 ) N > 1000, : m X m, X m O(m 2 N) 22 / 59
24 Subset of Data : K K mm (23), m O(m 3 ), 23 / 59
25 Subset of Data : K K mm (24), m O(m 3 ), / 59
26 (2) Subset of Regressors (Silverman 1985) : m K K nm K 1 mmk mn = K (25) K nm : N m O(m 2 N) 25 / 59
27 (2) Subset of Regressors (Silverman 1985) : m K K nm K 1 mmk mn = K (26) K nm : N m O(m 2 N), / 59
28 K, (Quiñonero-Candela & Rasmussen 2005). 27 / 59
29 (Titsias 2009), Jensen : log p(x)f(x)dx p(x) log f(x)dx X m GP f m, log p(y) = log p(y, f, f m )dfdf m (27) = log q(f, f m ) p(y, f, f m) q(f, f m ) dfdf m (28) q(f, f m ) log p(y, f, f m) q(f, f m ) dfdf m (29), q(f, f m ) 28 / 59
30 (2) p(y, f, f m ) = p(y f)p(f f m )p(f m ), q(f, f m ) = p(f f m )q(f m ), log p(y) = = = q(f, f m ) log p(y, f, f m) q(f, f m ) dfdf m (30) p(f f m )q(f m ) log p(y f) p(f f m )p(f m ) dfdf m p(f f m )q(f m ) (31) p(f f m )q(f m ) log p(y f)p(f m) dfdf m q(f m ) (32) q(f m )[ p(f f m ) log p(y f)df } {{ } G(f m ) + log p(f ] m) df m q(f m ) (33) 29 / 59
31 (3) G(f m ), G(f m ) = p(f f m ) log p(y f)df (34) = p(f f m ) ( N2 ) (y log(2πσ2 f)2 ) 2σ 2 df (35) [ = p(f f m ) N 2 log(2πσ2 ) 1 ] 2σ 2 tr(yt y 2y T f +f T f) df = N 2 log(2πσ2 ) 1 [ y T 2σ 2 y 2y T α+α T α+tr ( K nn K nm K 1 ( α = E[f fm ] = K nm Kmmf 1 ) m = log N(y α, σ 2 I) 1 2σ 2 tr ( K nn K nn (36) mmk mn )] (37) ). (38) 30 / 59
32 (4), log p(y) = = [ q(f m ) q(f m ) G(f m ) + log p(f m) q(f m ) [ log N(y α, σ 2 I) 1 ] df m (39) 2σ 2 tr ( K nn K nn ) + log p(f m) q(f m ) ] df m [ q(f m ) log N(y α, σ2 I) + log p(f m ) q(f m ) Jensen bound, p(x) log f(x) dx log p(x) ] df m (40) 1 2σ 2 tr(k nn K nn) (41) f(x)dx (42) 31 / 59
33 (5), log N(y α, σ 2 I)p(f m )df m 1 2σ 2 tr(k nn K nn) (K nn = K nm K 1 mmk mn ) (43) α = E[f f m ] = K nm K 1 mmf m, N(y α, σ 2 I)p(f m )df m = N(y 0, σ 2 I + K nn) (44), log p(y) log N(y 0, σ 2 I + K nn) 1 2σ 2 tr(k nn K nn). (45) 32 / 59
34 (6) log N(y 0, σ 2 I + K nn) 1 2σ 2 tr(k nn K nn) = log N(y 0, σ 2 I + K nn) 1 2σ 2 tr(cov(f f m)) (46) 1 : f m 2 : f m K nn, / 59
35 GP SVM y = {+1, 1}, p(y f) = σ(y f) (logit) or Ψ(y f) (probit) minimize: log p(y f)p(f X) = 1 N 2 f T K 1 f log p(y i f i ) (47) i=1 SVM Kα = f, w = α i x i w 2 = α T Kα = f T K 1 f, i 1 N minimize: 2 w 2 C (1 y i f i ) + i=1 = 1 N 2 f T K 1 f C (1 y i f i ) +. (48) i=1, SVM hinge loss. 34 / 59
36 Loss functions Relationships between GPs and Other Models 2 log(1 + exp( z)) log Φ(z) max(1 z, 0) g ǫ(z) ǫ 0 ǫ z (a) (b) Figure 6.3: (a) A comparison of the hinge error, g λ and g Φ. (b) The ǫ-insensitive error function used in SVR. SVM ME, :, GP classifier ( ) 35 / 59
37 DP Gaussian process Dirichlet process [ ] GP: (x 1, x 2,, x ), (y 1, y 2,, y ) DP: (X 1, X 2,, X ), Dir(α(X 1 ), α(x 2 ),, α(x )), smoother 36 / 59
38
39 Probabilistic PCA (Tipping & Bishop 1999), { yn = Wx n + ϵ ϵ N(0, σ 2 I) (49) L = log p(y n ) = log N(Wx n, σ 2 I) (50) = N 2 ( log 2π + log C + tr(c 1 S) ) (51), C = WW T + σ 2 I (52) S = 1 N YYT. (53) 38 / 59
40 (2) L = 0, L W W Ŵ U q(λ q σ 2 I) 1 2 (σ 2 = 0 U q Λ 1 2 ) (54) Λ q, U q : YY T q σ 2 = 0 39 / 59
41 Gaussian Process Latent Variable Models (GPLVM) Probabilistic PCA (Tipping&Bishop 1999): p(y n W, β) = p(y n x n,w, β)p(x n )dx n (55) p(y W, β) = n p(y n W, β) W GPLVM (Lawrence, NIPS 2003): W prior p(w) = D N(w i 0, α 1 I) (56) i=1 p(y X, β) = p(y X, β)p(w)dw (57) ( 1 = (2π) DN/2 exp 1 ) K D/2 2 tr(k 1 YY T ) (58) 40 / 59
42 GPLVM (2): PPCA Dual log p(y X, β) = DN 2 log(2π) D 2 log K 1 2 tr(k 1 YY T ) (59) K = αxx T + β 1 I (60) X = [x 1,, x N ] T (61) X, L X = αk 1 YY T K 1 X αdk 1 X = 0 (62) X = 1 D YYT K 1 X X U Q LV T (63) U Q (N Q) : YY T Q λ 1 λ Q L = diag(l 1,, l Q ); l i = 1/ λi αd 1 αβ 41 / 59
43 GPLVM (3) : Kernel log p(y X, β) = DN 2 log(2π) D 2 log K 1 2 tr(k 1 YY T ) K = αxx T + β 1 I, (64) X = [x 1,, x N ] T (65) = K ( k(x n, x m ) = α exp γ 2 (x n x m ) 2) + δ(n, m)β 1 (66) L K = K 1 YY T K 1 DK 1 L = L K x n,j K x n,j Scaled Conjugate Gradient GPLVM in MATLAB: neill/gplvm/ 42 / 59
44 GPLVM (4): Figure 1: Visualisation of the Oil data with (a) PCA (a linear GPLVM) and (b) A GPLVM which uses an RBF kernel. Crosses, circles and plus signs represent stratifi ed, annular and homogeneous flows respectively. The greyscales in plot (b) indicate the precision with which the manifold was expressed in data-space for that latent point. The optimised parameters of the kernel were, and f. PPCA( ), GP-LVM( ), Confidence (O(N 3 )): active set ( ), 43 / 59
45 GPLVM (4): Caveat PCA, Neil Lawrence, 1e-2*randn(N,dims) Scaled conjugate gradient / 59
46 GPLVM (5): / 59
47 GPLVM (6): / 59
48 GPLVM (6): PCA 47 / 59
49 GPLVM (7): MCMC Local Global MCMC ( =0.2, 400 iteration) 0 ( GPDM) 48 / 59
50 GPLVM (8): MCMC (Oil Flow) Local Global MCMC, X, X 49 / 59
51 Gaussian Process Dynamical Model (Hertzmann 2005) jmwang/gpdm/ GPLVM, x n x n (GP ). ( ).? 50 / 59
52 GPDM (2): Formulation (1) { xt = f(x t 1 ; A) + ϵ x,t y t = g(x t ; B) + ϵ y,t, f GP(0, K x ) (67) g GP(0, K y ) (68) p(y, X α, β) = p(y X, β)p(x α). 1 W N ( p(y X, β) = (2π) ND/2 exp 1 ) K Y D/2 2 tr(k 1 Y YW2 Y T ) (69) GPLVM. K Y ( ) RBF 51 / 59
53 GPDM (3): Formulation (2) 2 Markov N p(x α) = p(x 1 ) p(x t x t 1, A, α) p(a α) da (70) }{{} t=2 Gaussian ( 1 = p(x 1 ) (2π) d(n 1)/2 K X d exp 1 ) 2 tr(k 1 X X X T ) (71) X = [x 2,, x N ] T K X x 1 x N 1 RBF+ ( k(x, x ) = α 1 exp α 2 2 x x 2) + α 3 x T x + α4 1 δ(x, x ). (72) 52 / 59
54 GPDM (4): Formulation(3) p(y, X, α, β) = p(y X, β)p(x α)p(α)p(β) (73) p(α) i αi 1, p(β) i β 1 i. (74) log p(y, X, α, β) = 1 2 tr(k 1 X X X T ) tr(k 1 Y YW2 Y T ) + d 2 log K X + D 2 log K Y ( ) log W + log α j + log β j j j }{{} ( ) (75). (76) 53 / 59
55 Gaussian Process Density Sampler (1) (a) l x =1, l y =1, α=1 (b) l x =1, l y =1, α=10 (c) l x =0.2, l y =0.2, α=5 (d) l x =0.1, l y =2, α=5 GP prior? p(x) = 1 Φ(f(x))π(x) (77) Z(f) f(x) GP(x) ; π(x) : Φ(x) [0, 1] : ex. Φ(x) = 1/(1 + exp( x)) 54 / 59
56 Gaussian Process Density Sampler (2) : Rejection sampling p(x) = 1 Φ(f(x))π(x) (78) Z(f) 1. Draw x π(x). 2. Draw r Uniform[0, 1]. 3. If r < Φ(g(x)) then accept x; else reject x Accept N, reject M ( ) Z(f), Φ(g(x)) MCMC! Infinite Mixture 55 / 59
57 Gaussian process,,,,, (GPLVM, GPDM) 56 / 59
58 Literature Gaussian Process Dynamical Models. J. Wang, D. Fleet, and A. Hertzmann. NIPS jmwang/gpdm/ Gaussian Process Latent Variable Models for Visualization of High Dimensional Data. Neil D. Lawrence, NIPS The Gaussian Process Density Sampler. Ryan Prescott Adams, Iain Murray and David MacKay. NIPS Archipelago: Nonparametric Bayesian Semi-Supervised Learning. Ryan Prescott Adams and Zoubin Ghahramani. ICML / 59
59 (Pattern Recognition and Machine Learning), Chapter 6. Christopher Bishop, Springer, Gaussian Processes for Machine Learning. Rasmussen and Williams, MIT Press, Gaussian Processes A Replacement for Supervised Neural Networks?. David MacKay, Lecture notes at NIPS Videolectures.net: Gaussian Process Basics. mackay gpb/ (1)., tmasada/ pdf 58 / 59
60 Codes GPML Toolbox (in MATLAB): GPy (in Python): 59 / 59
A
A 2563 15 4 21 1 3 1.1................................................ 3 1.2............................................. 3 2 3 2.1......................................... 3 2.2............................................
I A A441 : April 15, 2013 Version : 1.1 I Kawahira, Tomoki TA (Shigehiro, Yoshida )
I013 00-1 : April 15, 013 Version : 1.1 I Kawahira, Tomoki TA (Shigehiro, Yoshida) http://www.math.nagoya-u.ac.jp/~kawahira/courses/13s-tenbou.html pdf * 4 15 4 5 13 e πi = 1 5 0 5 7 3 4 6 3 6 10 6 17
30
3 ............................................2 2...........................................2....................................2.2...................................2.3..............................
meiji_resume_1.PDF
β β β (q 1,q,..., q n ; p 1, p,..., p n ) H(q 1,q,..., q n ; p 1, p,..., p n ) Hψ = εψ ε k = k +1/ ε k = k(k 1) (x, y, z; p x, p y, p z ) (r; p r ), (θ; p θ ), (ϕ; p ϕ ) ε k = 1/ k p i dq i E total = E
‚åŁÎ“·„´Šš‡ðŠp‡¢‡½‹âfi`fiI…A…‰…S…−…Y…•‡ÌMarkovŸA“½fiI›ð’Í
Markov 2009 10 2 Markov 2009 10 2 1 / 25 1 (GA) 2 GA 3 4 Markov 2009 10 2 2 / 25 (GA) (GA) L ( 1) I := {0, 1} L f : I (0, ) M( 2) S := I M GA (GA) f (i) i I Markov 2009 10 2 3 / 25 (GA) ρ(i, j), i, j I
ばらつき抑制のための確率最適制御
( ) http://wwwhayanuemnagoya-uacjp/ fujimoto/ 2011 3 9 11 ( ) 2011/03/09-11 1 / 46 Outline 1 2 3 4 5 ( ) 2011/03/09-11 2 / 46 Outline 1 2 3 4 5 ( ) 2011/03/09-11 3 / 46 (1/2) r + Controller - u Plant y
X G P G (X) G BG [X, BG] S 2 2 2 S 2 2 S 2 = { (x 1, x 2, x 3 ) R 3 x 2 1 + x 2 2 + x 2 3 = 1 } R 3 S 2 S 2 v x S 2 x x v(x) T x S 2 T x S 2 S 2 x T x S 2 = { ξ R 3 x ξ } R 3 T x S 2 S 2 x x T x S 2
tokei01.dvi
2. :,,,. :.... Apr. - Jul., 26FY Dept. of Mechanical Engineering, Saga Univ., JAPAN 4 3. (probability),, 1. : : n, α A, A a/n. :, p, p Apr. - Jul., 26FY Dept. of Mechanical Engineering, Saga Univ., JAPAN
…p…^†[…fiflF”¯ Pattern Recognition
Pattern Recognition Shin ichi Satoh National Institute of Informatics June 11, 2019 (Support Vector Machines) (Support Vector Machines: SVM) SVM Vladimir N. Vapnik and Alexey Ya. Chervonenkis 1963 SVM
1 1.1 ( ). z = a + bi, a, b R 0 a, b 0 a 2 + b 2 0 z = a + bi = ( ) a 2 + b 2 a a 2 + b + b 2 a 2 + b i 2 r = a 2 + b 2 θ cos θ = a a 2 + b 2, sin θ =
1 1.1 ( ). z = + bi,, b R 0, b 0 2 + b 2 0 z = + bi = ( ) 2 + b 2 2 + b + b 2 2 + b i 2 r = 2 + b 2 θ cos θ = 2 + b 2, sin θ = b 2 + b 2 2π z = r(cos θ + i sin θ) 1.2 (, ). 1. < 2. > 3. ±,, 1.3 ( ). A
.2 ρ dv dt = ρk grad p + 3 η grad (divv) + η 2 v.3 divh = 0, rote + c H t = 0 dive = ρ, H = 0, E = ρ, roth c E t = c ρv E + H c t = 0 H c E t = c ρv T
NHK 204 2 0 203 2 24 ( ) 7 00 7 50 203 2 25 ( ) 7 00 7 50 203 2 26 ( ) 7 00 7 50 203 2 27 ( ) 7 00 7 50 I. ( ν R n 2 ) m 2 n m, R = e 2 8πε 0 hca B =.09737 0 7 m ( ν = ) λ a B = 4πε 0ħ 2 m e e 2 = 5.2977
I A A441 : April 21, 2014 Version : Kawahira, Tomoki TA (Kondo, Hirotaka ) Google
I4 - : April, 4 Version :. Kwhir, Tomoki TA (Kondo, Hirotk) Google http://www.mth.ngoy-u.c.jp/~kwhir/courses/4s-biseki.html pdf 4 4 4 4 8 e 5 5 9 etc. 5 6 6 6 9 n etc. 6 6 6 3 6 3 7 7 etc 7 4 7 7 8 5 59
1 (Berry,1975) 2-6 p (S πr 2 )p πr 2 p 2πRγ p p = 2γ R (2.5).1-1 : : : : ( ).2 α, β α, β () X S = X X α X β (.1) 1 2
2005 9/8-11 2 2.2 ( 2-5) γ ( ) γ cos θ 2πr πρhr 2 g h = 2γ cos θ ρgr (2.1) γ = ρgrh (2.2) 2 cos θ θ cos θ = 1 (2.2) γ = 1 ρgrh (2.) 2 2. p p ρgh p ( ) p p = p ρgh (2.) h p p = 2γ r 1 1 (Berry,1975) 2-6
W u = u(x, t) u tt = a 2 u xx, a > 0 (1) D := {(x, t) : 0 x l, t 0} u (0, t) = 0, u (l, t) = 0, t 0 (2)
3 215 4 27 1 1 u u(x, t) u tt a 2 u xx, a > (1) D : {(x, t) : x, t } u (, t), u (, t), t (2) u(x, ) f(x), u(x, ) t 2, x (3) u(x, t) X(x)T (t) u (1) 1 T (t) a 2 T (t) X (x) X(x) α (2) T (t) αa 2 T (t) (4)
1 (1) () (3) I 0 3 I I d θ = L () dt θ L L θ I d θ = L = κθ (3) dt κ T I T = π κ (4) T I κ κ κ L l a θ L r δr δl L θ ϕ ϕ = rθ (5) l
1 1 ϕ ϕ ϕ S F F = ϕ (1) S 1: F 1 1 (1) () (3) I 0 3 I I d θ = L () dt θ L L θ I d θ = L = κθ (3) dt κ T I T = π κ (4) T I κ κ κ L l a θ L r δr δl L θ ϕ ϕ = rθ (5) l : l r δr θ πrδr δf (1) (5) δf = ϕ πrδr
211 [email protected] 1 R *1 n n R n *2 R n = {(x 1,..., x n ) x 1,..., x n R}. R R 2 R 3 R n R n R n D D R n *3 ) (x 1,..., x n ) f(x 1,..., x n ) f D *4 n 2 n = 1 ( ) 1 f D R n f : D R 1.1. (x,
1 Introduction 1 (1) (2) (3) () {f n (x)} n=1 [a, b] K > 0 n, x f n (x) K < ( ) x [a, b] lim f n (x) f(x) (1) f(x)? (2) () f(x)? b lim a f n (x)dx = b
![1 Introduction 1 (1) (2) (3) () {f n (x)} n=1 [a, b] K > 0 n, x f n (x) K < ( ) x [a, b] lim f n (x) f(x) (1) f(x)? (2) () f(x)? b lim a f n (x)dx = b](/thumbs/92/108011004.jpg "1 Introduction 1 (1) (2) (3) () {f n (x)} n=1 [a, b] K > 0 n, x f n (x) K < ( ) x [a, b] lim f n (x) f(x) (1) f(x)? (2) () f(x)? b lim a f n (x)dx = b")
1 Introduction 2 2.1 2.2 2.3 3 3.1 3.2 σ- 4 4.1 4.2 5 5.1 5.2 5.3 6 7 8. Fubini,,. 1 1 Introduction 1 (1) (2) (3) () {f n (x)} n=1 [a, b] K > 0 n, x f n (x) K < ( ) x [a, b] lim f n (x) f(x) (1) f(x)?
,. Black-Scholes u t t, x c u 0 t, x x u t t, x c u t, x x u t t, x + σ x u t, x + rx ut, x rux, t 0 x x,,.,. Step 3, 7,,, Step 6., Step 4,. Step 5,,.
9 α ν β Ξ ξ Γ γ o δ Π π ε ρ ζ Σ σ η τ Θ θ Υ υ ι Φ φ κ χ Λ λ Ψ ψ µ Ω ω Def, Prop, Th, Lem, Note, Remark, Ex,, Proof, R, N, Q, C [a, b {x R : a x b} : a, b {x R : a < x < b} : [a, b {x R : a x < b} : a,
Dirichlet process mixture Dirichlet process mixture 2 /40 MIRU2008 :
Dirichlet Process : joint work with: Max Welling (UC Irvine), Yee Whye Teh (UCL, Gatsby) http://kenichi.kurihara.googlepages.com/miru_workshop.pdf 1 /40 MIRU2008 : Dirichlet process mixture Dirichlet process
y π π O π x 9 s94.5 y dy dx. y = x + 3 y = x logx + 9 s9.6 z z x, z y. z = xy + y 3 z = sinx y 9 s x dx π x cos xdx 9 s93.8 a, fx = e x ax,. a =
[ ] 9 IC. dx = 3x 4y dt dy dt = x y u xt = expλt u yt λ u u t = u u u + u = xt yt 6 3. u = x, y, z = x + y + z u u 9 s9 grad u ux, y, z = c c : grad u = u x i + u y j + u k i, j, k z x, y, z grad u v =
目次 ガウス過程 (Gaussian Process; GP) 序論 GPによる回帰 GPによる識別 GP 状態空間モデル 概括 GP 状態空間モデルによる音楽ムードの推定
公開講座 : ガウス過程の基礎と応用 05/3/3 ガウス過程の基礎 統計数理研究所 松井知子 目次 ガウス過程 (Gaussian Process; GP) 序論 GPによる回帰 GPによる識別 GP 状態空間モデル 概括 GP 状態空間モデルによる音楽ムードの推定 GP 序論 ノンパラメトリック予測 カーネル法の利用 参照文献 : C. E. Rasmussen and C. K. I. Williams
x V x x V x, x V x = x + = x +(x+x )=(x +x)+x = +x = x x = x x = x =x =(+)x =x +x = x +x x = x ( )x = x =x =(+( ))x =x +( )x = x +( )x ( )x = x x x R
V (I) () (4) (II) () (4) V K vector space V vector K scalor K C K R (I) x, y V x + y V () (x + y)+z = x +(y + z) (2) x + y = y + x (3) V x V x + = x (4) x V x + x = x V x x (II) x V, α K αx V () (α + β)x
2016 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 1 16 2 1 () X O 3 (O1) X O, O (O2) O O (O3) O O O X (X, O) O X X (O1), (O2), (O3) (O2) (O3) n (O2) U 1,..., U n O U k O k=1 (O3) U λ O( λ Λ) λ Λ U λ O 0 X 0 (O2) n =
知能科学：ニューラルネットワーク
2 3 4 (Neural Network) (Deep Learning) (Deep Learning) ( x x = ax + b x x x ? x x x w σ b = σ(wx + b) x w b w b .2.8.6 σ(x) = + e x.4.2 -.2 - -5 5 x w x2 w2 σ x3 w3 b = σ(w x + w 2 x 2 + w 3 x 3 + b) x,
2011de.dvi
211 ( 4 2 1. 3 1.1............................... 3 1.2 1- -......................... 13 1.3 2-1 -................... 19 1.4 3- -......................... 29 2. 37 2.1................................ 37
6 2 2 x y x y t P P = P t P = I P P P ( ) ( ) ,, ( ) ( ) cos θ sin θ cos θ sin θ, sin θ cos θ sin θ cos θ y x θ x θ P
6 x x 6.1 t P P = P t P = I P P P 1 0 1 0,, 0 1 0 1 cos θ sin θ cos θ sin θ, sin θ cos θ sin θ cos θ x θ x θ P x P x, P ) = t P x)p ) = t x t P P ) = t x = x, ) 6.1) x = Figure 6.1 Px = x, P=, θ = θ P
V(x) m e V 0 cos x π x π V(x) = x < π, x > π V 0 (i) x = 0 (V(x) V 0 (1 x 2 /2)) n n d 2 f dξ 2ξ d f 2 dξ + 2n f = 0 H n (ξ) (ii) H
199 1 1 199 1 1. Vx) m e V cos x π x π Vx) = x < π, x > π V i) x = Vx) V 1 x /)) n n d f dξ ξ d f dξ + n f = H n ξ) ii) H n ξ) = 1) n expξ ) dn dξ n exp ξ )) H n ξ)h m ξ) exp ξ )dξ = π n n!δ n,m x = Vx)
( 28 ) ( ) ( ) 0 This note is c 2016, 2017 by Setsuo Taniguchi. It may be used for personal or classroom purposes, but not for commercial purp
( 28) ( ) ( 28 9 22 ) 0 This ote is c 2016, 2017 by Setsuo Taiguchi. It may be used for persoal or classroom purposes, but ot for commercial purposes. i (http://www.stat.go.jp/teacher/c2epi1.htm ) = statistics
IA [email protected] Last updated: January,......................................................................................................................................................................................
Part () () Γ Part ,
Contents a 6 6 6 6 6 6 6 7 7. 8.. 8.. 8.3. 8 Part. 9. 9.. 9.. 3. 3.. 3.. 3 4. 5 4.. 5 4.. 9 4.3. 3 Part. 6 5. () 6 5.. () 7 5.. 9 5.3. Γ 3 6. 3 6.. 3 6.. 3 6.3. 33 Part 3. 34 7. 34 7.. 34 7.. 34 8. 35
6kg 1.1m 1.m.1m.1 l λ ϵ λ l + λ l l l dl dl + dλ ϵ dλ dl dl + dλ dl dl 3 1. JIS 1 6kg 1% 66kg 1 13 σ a1 σ m σ a1 σ m σ m σ a1 f f σ a1 σ a1 σ m f 4
35-8585 7 8 1 I I 1 1.1 6kg 1m P σ σ P 1 l l λ λ l 1.m 1 6kg 1.1m 1.m.1m.1 l λ ϵ λ l + λ l l l dl dl + dλ ϵ dλ dl dl + dλ dl dl 3 1. JIS 1 6kg 1% 66kg 1 13 σ a1 σ m σ a1 σ m σ m σ a1 f f σ a1 σ a1 σ m
2000年度『数学展望 I』講義録
2000 I I IV I II 2000 I I IV I-IV. i ii 3.10 (http://www.math.nagoya-u.ac.jp/ kanai/) 2000 A....1 B....4 C....10 D....13 E....17 Brouwer A....21 B....26 C....33 D....39 E. Sperner...45 F....48 A....53
(3) (2),,. ( 20) ( s200103) 0.7 x C,, x 2 + y 2 + ax = 0 a.. D,. D, y C, C (x, y) (y 0) C m. (2) D y = y(x) (x ± y 0), (x, y) D, m, m = 1., D. (x 2 y
[ ] 7 0.1 2 2 + y = t sin t IC ( 9) ( s090101) 0.2 y = d2 y 2, y = x 3 y + y 2 = 0 (2) y + 2y 3y = e 2x 0.3 1 ( y ) = f x C u = y x ( 15) ( s150102) [ ] y/x du x = Cexp f(u) u (2) x y = xey/x ( 16) ( s160101)
v er.1/ c /(21)
12 -- 1 1 2009 1 17 1-1 1-2 1-3 1-4 2 2 2 1-5 1 1-6 1 1-7 1-1 1-2 1-3 1-4 1-5 1-6 1-7 c 2011 1/(21) 12 -- 1 -- 1 1--1 1--1--1 1 2009 1 n n α { n } α α { n } lim n = α, n α n n ε n > N n α < ε N {1, 1,
http://www.ns.kogakuin.ac.jp/~ft13389/lecture/physics1a2b/ pdf I 1 1 1.1 ( ) 1. 30 m µm 2. 20 cm km 3. 10 m 2 cm 2 4. 5 cm 3 km 3 5. 1 6. 1 7. 1 1.2 ( ) 1. 1 m + 10 cm 2. 1 hr + 6400 sec 3. 3.0 10 5 kg
I, II 1, A = A 4 : 6 = max{ A, } A A 10 10%
1 2006.4.17. A 3-312 tel: 092-726-4774, e-mail: [email protected], http://www.math.kyushu-u.ac.jp/ hara/lectures/lectures-j.html Office hours: B A I ɛ-δ ɛ-δ 1. 2. A 1. 1. 2. 3. 4. 5. 2. ɛ-δ 1. ɛ-n
1 1 x y = y(x) y, y,..., y (n) : n y F (x, y, y,..., y (n) ) = 0 n F (x, y, y ) = 0 1 y(x) y y = G(x, y) y, y y + p(x)y = q(x) 1 p(x) q(
1 1 y = y() y, y,..., y (n) : n y F (, y, y,..., y (n) ) = 0 n F (, y, y ) = 0 1 y() 1.1 1 y y = G(, y) 1.1.1 1 y, y y + p()y = q() 1 p() q() (q() = 0) y + p()y = 0 y y + py = 0 y y = p (log y) = p log
20 9 19 1 3 11 1 3 111 3 112 1 4 12 6 121 6 122 7 13 7 131 8 132 10 133 10 134 12 14 13 141 13 142 13 143 15 144 16 145 17 15 19 151 1 19 152 20 2 21 21 21 211 21 212 1 23 213 1 23 214 25 215 31 22 33
医系の統計入門第 2 版 サンプルページ この本の定価 判型などは, 以下の URL からご覧いただけます. このサンプルページの内容は, 第 2 版 1 刷発行時のものです.
医系の統計入門第 2 版 サンプルページ この本の定価 判型などは, 以下の URL からご覧いただけます. http://www.morikita.co.jp/books/mid/009192 このサンプルページの内容は, 第 2 版 1 刷発行時のものです. i 2 t 1. 2. 3 2 3. 6 4. 7 5. n 2 ν 6. 2 7. 2003 ii 2 2013 10 iii 1987
/ 2 n n n n x 1,..., x n 1 n 2 n R n n ndimensional Euclidean space R n vector point R n set space R n R n x = x 1 x n y = y 1 y n distance dx,
1 1.1 R n 1.1.1 3 xyz xyz 3 x, y, z R 3 := x y : x, y, z R z 1 3. n n x 1,..., x n x 1. x n x 1 x n 1 / 2 n n n n x 1,..., x n 1 n 2 n R n n ndimensional Euclidean space R n vector point 1.1.2 R n set
simx simxdx, cosxdx, sixdx 6.3 px m m + pxfxdx = pxf x p xf xdx = pxf x p xf x + p xf xdx 7.4 a m.5 fx simxdx 8 fx fx simxdx = πb m 9 a fxdx = πa a =
II 6 [email protected] 6.. 5.4.. f Rx = f Lx = fx fx + lim = lim x x + x x f c = f x + x < c < x x x + lim x x fx fx x x = lim x x f c = f x x < c < x cosmx cosxdx = {cosm x + cosm + x} dx = [
x, y x 3 y xy 3 x 2 y + xy 2 x 3 + y 3 = x 3 y xy 3 x 2 y + xy 2 x 3 + y 3 = 15 xy (x y) (x + y) xy (x y) (x y) ( x 2 + xy + y 2) = 15 (x y)
x, y x 3 y xy 3 x 2 y + xy 2 x 3 + y 3 = 15 1 1977 x 3 y xy 3 x 2 y + xy 2 x 3 + y 3 = 15 xy (x y) (x + y) xy (x y) (x y) ( x 2 + xy + y 2) = 15 (x y) ( x 2 y + xy 2 x 2 2xy y 2) = 15 (x y) (x + y) (xy
08-Note2-web
r(t) t r(t) O v(t) = dr(t) dt a(t) = dv(t) dt = d2 r(t) dt 2 r(t), v(t), a(t) t dr(t) dt r(t) =(x(t),y(t),z(t)) = d 2 r(t) dt 2 = ( dx(t) dt ( d 2 x(t) dt 2, dy(t), dz(t) dt dt ), d2 y(t) dt 2, d2 z(t)
A S hara/lectures/lectures-j.html ϵ-n 1 ϵ-n lim n a n = α n a n α 2 lim a n = 0 1 n a k n n k= ϵ
A S1-20 http://www2.mth.kyushu-u.c.jp/ hr/lectures/lectures-j.html 1 1 1.1 ϵ-n 1 ϵ-n lim n n = α n n α 2 lim n = 0 1 n k n n k=1 0 1.1.7 ϵ-n 1.1.1 n α n n α lim n n = α ϵ N(ϵ) n > N(ϵ) n α < ϵ (1.1.1)
& 3 3 ' ' (., (Pixel), (Light Intensity) (Random Variable). (Joint Probability). V., V = {,,, V }. i x i x = (x, x,, x V ) T. x i i (State Variable),
.... Deeping and Expansion of Large-Scale Random Fields and Probabilistic Image Processing Kazuyuki Tanaka The mathematical frameworks of probabilistic image processing are formulated by means of Markov
,, Poisson 3 3. t t y,, y n Nµ, σ 2 y i µ + ɛ i ɛ i N0, σ 2 E[y i ] µ * i y i x i y i α + βx i + ɛ i ɛ i N0, σ 2, α, β *3 y i E[y i ] α + βx i
![,, Poisson 3 3. t t y,, y n Nµ, σ 2 y i µ + ɛ i ɛ i N0, σ 2 E[y i ] µ * i y i x i y i α + βx i + ɛ i ɛ i N0, σ 2, α, β *3 y i E[y i ] α + βx i](/thumbs/87/95421301.jpg ",, Poisson 3 3. t t y,, y n Nµ, σ 2 y i µ + ɛ i ɛ i N0, σ 2 E[y i ] µ * i y i x i y i α + βx i + ɛ i ɛ i N0, σ 2, α, β *3 y i E[y i ] α + βx i")
Armitage.? SAS.2 µ, µ 2, µ 3 a, a 2, a 3 a µ + a 2 µ 2 + a 3 µ 3 µ, µ 2, µ 3 µ, µ 2, µ 3 log a, a 2, a 3 a µ + a 2 µ 2 + a 3 µ 3 µ, µ 2, µ 3 * 2 2. y t y y y Poisson y * ,, Poisson 3 3. t t y,, y n Nµ,
Note.tex 2008/09/19( )
1 20 9 19 2 1 5 1.1........................ 5 1.2............................. 8 2 9 2.1............................. 9 2.2.............................. 10 3 13 3.1.............................. 13 3.2..................................
4. ϵ(ν, T ) = c 4 u(ν, T ) ϵ(ν, T ) T ν π4 Planck dx = 0 e x 1 15 U(T ) x 3 U(T ) = σt 4 Stefan-Boltzmann σ 2π5 k 4 15c 2 h 3 = W m 2 K 4 5.
A 1. Boltzmann Planck u(ν, T )dν = 8πh ν 3 c 3 kt 1 dν h 6.63 10 34 J s Planck k 1.38 10 23 J K 1 Boltzmann u(ν, T ) T ν e hν c = 3 10 8 m s 1 2. Planck λ = c/ν Rayleigh-Jeans u(ν, T )dν = 8πν2 kt dν c
液晶の物理１：連続体理論（弾性，粘性）
The Physics of Liquid Crystals P. G. de Gennes and J. Prost (Oxford University Press, 1993) Liquid crystals are beautiful and mysterious; I am fond of them for both reasons. My hope is that some readers
II 2 3.,, A(B + C) = AB + AC, (A + B)C = AC + BC. 4. m m A, m m B,, m m B, AB = BA, A,, I. 5. m m A, m n B, AB = B, A I E, 4 4 I, J, K
II. () 7 F 7 = { 0,, 2, 3, 4, 5, 6 }., F 7 a, b F 7, a b, F 7,. (a) a, b,,. (b) 7., 4 5 = 20 = 2 7 + 6, 4 5 = 6 F 7., F 7,., 0 a F 7, ab = F 7 b F 7. (2) 7, 6 F 6 = { 0,, 2, 3, 4, 5 },,., F 6., 0 0 a F
No δs δs = r + δr r = δr (3) δs δs = r r = δr + u(r + δr, t) u(r, t) (4) δr = (δx, δy, δz) u i (r + δr, t) u i (r, t) = u i x j δx j (5) δs 2
No.2 1 2 2 δs δs = r + δr r = δr (3) δs δs = r r = δr + u(r + δr, t) u(r, t) (4) δr = (δx, δy, δz) u i (r + δr, t) u i (r, t) = u i δx j (5) δs 2 = δx i δx i + 2 u i δx i δx j = δs 2 + 2s ij δx i δx j
II (Percolation) ( 3-4 ) 1. [ ],,,,,,,. 2. [ ],.. 3. [ ],. 4. [ ] [ ] G. Grimmett Percolation Springer-Verlag New-York [ ] 3
![II (Percolation) ( 3-4 ) 1. [ ],,,,,,,. 2. [ ],.. 3. [ ],. 4. [ ] [ ] G. Grimmett Percolation Springer-Verlag New-York [ ] 3](/thumbs/94/121840826.jpg "II (Percolation) ( 3-4 ) 1. [ ],,,,,,,. 2. [ ],.. 3. [ ],. 4. [ ] [ ] G. Grimmett Percolation Springer-Verlag New-York [ ] 3")
II (Percolation) 12 9 27 ( 3-4 ) 1 [ ] 2 [ ] 3 [ ] 4 [ ] 1992 5 [ ] G Grimmett Percolation Springer-Verlag New-York 1989 6 [ ] 3 1 3 p H 2 3 2 FKG BK Russo 2 p H = p T (=: p c ) 3 2 Kesten p c =1/2 ( )
,,,17,,, ( ),, E Q [S T F t ] < S t, t [, T ],,,,,,,,
![,,,17,,, ( ),, E Q [S T F t ] < S t, t [, T ],,,,,,,,](/thumbs/91/105754403.jpg ",,,17,,, ( ),, E Q [S T F t ] < S t, t [, T ],,,,,,,,")
14 5 1 ,,,17,,,194 1 4 ( ),, E Q [S T F t ] < S t, t [, T ],,,,,,,, 1 4 1.1........................................ 4 5.1........................................ 5.........................................
³ÎΨÏÀ
2017 12 12 Makoto Nakashima 2017 12 12 1 / 22 2.1. C, D π- C, D. A 1, A 2 C A 1 A 2 C A 3, A 4 D A 1 A 2 D Makoto Nakashima 2017 12 12 2 / 22 . (,, L p - ). Makoto Nakashima 2017 12 12 3 / 22 . (,, L p
p.2/76
[email protected] p.1/76 p.2/76 ( ) (2001). (2006). (2002). p.3/76 N n, n {1, 2,...N} 0 K k, k {1, 2,...,K} M M, m {1, 2,...,M} p.4/76 R =(r ij ), r ij = i j ( ): k s r(k, s) r(k, 1),r(k, 2),...,r(k,
M3 x y f(x, y) (= x) (= y) x + y f(x, y) = x + y + *. f(x, y) π y f(x, y) x f(x + x, y) f(x, y) lim x x () f(x,y) x 3 -
M3............................................................................................ 3.3................................................... 3 6........................................... 6..........................................
20 4 20 i 1 1 1.1............................ 1 1.2............................ 4 2 11 2.1................... 11 2.2......................... 11 2.3....................... 19 3 25 3.1.............................
入試の軌跡
4 y O x 4 Typed by L A TEX ε ) ) ) 6 4 ) 4 75 ) http://kumamoto.s.xrea.com/plan/.. PDF) Ctrl +L) Ctrl +) Ctrl + Ctrl + ) ) Alt + ) Alt + ) ESC. http://kumamoto.s.xrea.com/nyusi/kumadai kiseki ri i.pdf
() n C + n C + n C + + n C n n (3) n C + n C + n C 4 + n C + n C 3 + n C 5 + (5) (6 ) n C + nc + 3 nc n nc n (7 ) n C + nc + 3 nc n nc n (
3 n nc k+ k + 3 () n C r n C n r nc r C r + C r ( r n ) () n C + n C + n C + + n C n n (3) n C + n C + n C 4 + n C + n C 3 + n C 5 + (4) n C n n C + n C + n C + + n C n (5) k k n C k n C k (6) n C + nc
N cos s s cos ψ e e e e 3 3 e e 3 e 3 e
3 3 5 5 5 3 3 7 5 33 5 33 9 5 8 > e > f U f U u u > u ue u e u ue u ue u e u e u u e u u e u N cos s s cos ψ e e e e 3 3 e e 3 e 3 e 3 > A A > A E A f A A f A [ ] f A A e > > A e[ ] > f A E A < < f ; >
ii 3.,. 4. F. (), ,,. 8.,. 1. (75%) (25%) =7 20, =7 21 (. ). 1.,, (). 3.,. 1. ().,.,.,.,.,. () (12 )., (), 0. 2., 1., 0,.
24(2012) (1 C106) 4 11 (2 C206) 4 12 http://www.math.is.tohoku.ac.jp/~obata,.,,,.. 1. 2. 3. 4. 5. 6. 7.,,. 1., 2007 (). 2. P. G. Hoel, 1995. 3... 1... 2.,,. ii 3.,. 4. F. (),.. 5... 6.. 7.,,. 8.,. 1. (75%)
x () g(x) = f(t) dt f(x), F (x) 3x () g(x) g (x) f(x), F (x) (3) h(x) = x 3x tf(t) dt.9 = {(x, y) ; x, y, x + y } f(x, y) = xy( x y). h (x) f(x), F (x
[ ] IC. f(x) = e x () f(x) f (x) () lim f(x) lim f(x) x + x (3) lim f(x) lim f(x) x + x (4) y = f(x) ( ) ( s46). < a < () a () lim a log xdx a log xdx ( ) n (3) lim log k log n n n k=.3 z = log(x + y ),
p = mv p x > h/4π λ = h p m v Ψ 2 Ψ
II p = mv p x > h/4π λ = h p m v Ψ 2 Ψ Ψ Ψ 2 0 x P'(x) m d 2 x = mω 2 x = kx = F(x) dt 2 x = cos(ωt + φ) mω 2 = k ω = m k v = dx = -ωsin(ωt + φ) dt = d 2 x dt 2 0 y v θ P(x,y) θ = ωt + φ ν = ω [Hz] 2π
DVIOUT
A. A. A-- [ ] f(x) x = f 00 (x) f 0 () =0 f 00 () > 0= f(x) x = f 00 () < 0= f(x) x = A--2 [ ] f(x) D f 00 (x) > 0= y = f(x) f 00 (x) < 0= y = f(x) P (, f()) f 00 () =0 A--3 [ ] y = f(x) [, b] x = f (y)
09 8 9 3 Chebyshev 5................................. 5........................................ 5.3............................. 6.4....................................... 8.4...................................
() x + y + y + x dy dx = 0 () dy + xy = x dx y + x y ( 5) ( s55906) 0.7. (). 5 (). ( 6) ( s6590) 0.8 m n. 0.9 n n A. ( 6) ( s6590) f A (λ) = det(a λi)
0. A A = 4 IC () det A () A () x + y + z = x y z X Y Z = A x y z ( 5) ( s5590) 0. a + b + c b c () a a + b + c c a b a + b + c 0 a b c () a 0 c b b c 0 a c b a 0 0. A A = 7 5 4 5 0 ( 5) ( s5590) () A ()
Microsoft PowerPoint - SSII_harada pptx
The state of the world The gathered data The processed data w d r I( W; D) I( W; R) The data processing theorem states that data processing can only destroy information. David J.C. MacKay. Information