$G=$;NO@J8!&2]Bj8&5f(B $BEE;R%U%!%$%k$HF1$8>l=j$KCV$$$F2<$5$$!#$h$m$7$/$*4j$$$7$^$9!#(B ->
$BK\8&5f$G$O(B,$B?M$H%m%\%C%H$K$h$kBn5e%3%_%e%K%1!<%7%g%s%?%9%/$K$*$$$F=EMW(B
$B$JMWAG$N0l$D$G$"$kBG5eM=B,5!G=$K>GE@$rEv$F(B,$BB?MM$J%9%T%s$N$+$+$C$?BG5e$K(B $B$D$$$F9b@:EY$JBG5eM=B,$r
$B$=$3$G(B,$BK\8&5f$G$O(B,$BHf3SE*9b@:EY$G7WB,$,MF0W$J%W%l%$%d!<$N%9%$%s%0F0:n(B
$B$KCeL\$7(B,$B%9%$%s%0F0:n$K4^$^$l$k%9%T%s$KBP$7$FFCD'E*$J>pJs$rBG5eM=B,$KMx(B $BMQ$9$k%"%W%m!<%A$rDs0F$9$k(B.$B6qBNE*$K$O(B,$B%b!<%7%g%s%-%c%W%A%c$K$h$C$FF@$i(B $B$l$kA4?H$N4X@a>pJs$+$i(B
CCA(Canonical Correlation
Analysis) $B$rMQ$$$F%9%T%s(B $B$K4XO"$9$kFCD'E*$J>pJs$rCj=P$7(B,$B%+%a%i$+$iF@$i$l$k5e$N0LCV>pJs$K2C$($F(B,
$BCj=P$7$?>pJs$rMxMQ$7(B,GPR(Gaussian Process Regression) $B$K$h$C$FBG5e0LCV$r(B $BM=B,$9$k!#(B