An analog VLSI implementation of one-class support vector machine for multiclass classification of highly dimensional vectors

Full text

(1)JAIST Repository https://dspace.jaist.ac.jp/. Title. An analog VLSI implementation of one-class support vector machine for multiclass classification of highly dimensional vectors. Author(s). Zhang, Renyuan; Kaneko, Mineo; Shibata, Tadashi. Citation. Japanese Journal of Applied Physics, 53(4S): 04EE03-1-04EE03-8. Issue Date. 2014-02-28. Type. Journal Article. Text version. author. URL. http://hdl.handle.net/10119/12155. Rights. This is the author's version of the work. It is posted here by permission of The Japan Society of Applied Physics. Copyright (C) 2014 The Japan Society of Applied Physics. Renyuan Zhang, Mineo Kaneko and Tadashi Shibata, Japanese Journal of Applied Physics, 53(4S), 2014, 04EE03-1-04EE03-8. http://dx.doi.org/10.7567/JJAP.53.04EE03. Description. Japan Advanced Institute of Science and Technology.

(2) .

(3) . .

(4)

(5)

(6)

(7)

(8)

(9) ½.

(10)

(11)

(12)

(13) ! ¾ "

(14)

(15)

(16)

(17) #

(18)

(19) $

(20) % &' '&() # * + ,-

(21) . .-.

(22) ! " # $ % " & . . " ! & ! " ' & % " # () " # & % . % . . . " * &. . + , ". .

(23)

(24)

(25)

(26)

(27)

(28) !

(29) "

(30) #

(31) " $ Æ

(32)

(33)

(34)

(35)

(36)

(37)

(38)

(39) %!

(40) &

(41) %!

(42)

(43) "

(44) '

(45)

(46)

(47) "

(48) " ! %! -.-/.

(49) . " "

(50) " ".

(51) "

(52) #

(53)

(54) # '

(55)

(56)

(57)

(58) #

(59)

(60) ! "

(61) ()

(62)

(63)

(64) " #

(65) "

(66) &*" !

(67)

(68) &*"

(69)

(70) "

(71) !

(72)

(73) (

(74)

(75)

(76) &*"

(77)

(78) "

(79) + &*"

(80) (

(81)

(82)

(83)

(84) &*"

(85) "

(86)

(87)

(88)

(89)

(90)

(91) *

(92)

(93) &*"

(94) "

(95)

(96) ,

(97)

(98)

(99)

(100)

(101) ' $-'

(102) "

(103) &*"

(104) . " /" ! %! " &*" 0 # %!

(105)

(106)

(107) &*"

(108)

(109)

(110) &

(111) &*"

(112) ' %!

(113)

(114) &*" " "

(115) "" 12"

(116)

(117) 3

(118)

(119) #

(120) 0

(121)

(122)

(123) 4

(124)

(125) '

(126) 5 1 $

(127)

(128)

(129) . 0.-/.

(130) . " "

(131) " " Learning samples … …. …. … …. … …. …. …. …. …. …. …. …. …. …. …. …. … … … …. …. Test patterns. …. … …. ….

(132) . …. ....... …. Multi-Class Recognition system. …. …. … Support Vectors. ?. . .

(133)

(134)

(135) . 0 4

(136) 0 6 ! 7

(137) &*"

(138) ! 8 &*" "

(139) #

(140)

(141) 2

(142)

(143)

(144) $ 9

(145) .

(146)

(147)

(148) . ! " %! Æ

(149)

(150)

(151) " /

(152)

(153)

(154) 0

(155) &

(156) /

(157)

(158) "

(159) + ( " "

(160) %!

(161)

(162) " "

(163)

(164)

(165) !

(166)

(167) " #

(168) +

(169) " " &*"

(170) 1.-/.

(171) . " "

(172) " ". Learning samples Class1. ... .... .... .... .... 㻌㻌㻌䈈䈈㻌 ClassN. .... .... .... .... .... ... ... .... .... .... ... .... .... Class2. Input broadcast. SVDD Chip1. .... Class1. SVDD Chip2. SVDD. 㻌㻌㻌䈈䈈㻌 ChipN. .... 㻌㻌㻌䈈䈈㻌 ClassN .... Class2. Recognition results. .

(173)

(174) .

(175)

(176)

(177)

(178) .

(179) '

(180) '

(181) " &*"

(182)

(183) : ;

(184)

(185) &*"

(186) $ <

(187) #

(188)

(189) &*"

(190)

(191)

(192)

(193)

(194) ' " 0 4 !

(195)

(196)

(197) & %! " = $ 7

(198)

(199) !

(200) "

(201)

(202) # &*"

(203)

(204) "

(205) 3

(206) &*" "

(207) #

(208)

(209)

(210) &

(211) $

(212)

(213)

(214)

(215) ).-/.

(216) . " "

(217) " ". minimum volume. center volume support vector domain With “kernel trick”. . support vector. Initial kernel function feature. non-support vector After learning. . .

(218)

(219) !

(220)

(221) "

(222) " #$%. "

(223) %

(224) "

(225) & ' %! %

(226) " " %" % & (

(227) &

(228) ) * "+

(229) , .

(230) . !"

(231) #. &*" #

(232)

(233)

(234) &*"

(235)

(236) ) : ; :4 ;

(237)

(238) &*" #

(239) >

(240)

(241) $ 8 * Æ

(242)

(243) ? 2.-/.

(244) . " "

(245) " ". #

(246)

(247)

(248)

(249)

(250)

(251) +

(252) >

(253) :; ! &*" " : ;

(254) "

(255)

(256) @

(257) <

(258)

(259) ' #

(260)

(261)

(262)

(263) "

(264) 4 # . . . . . . . . . . @. A. . <. .

(265). .

(266) "/

(267)

(268)

(269)

(270) 4 4

(271)

(272) 0 .

(273). . . . Ì. . . . . . A. 7. . 3 &*"

(274) " B @

(275) (

(276)

(277) C-

(278) 6 .

(279) @ < . . . . . . . . . . . . . . . . . 8. . @<. 2. . 5 D %

(280) C

(281) =

(282) 4 D Ì # 4 ?

(283) 4 .

(284). . . < 7 . . . . . . A. . .

(285) (.-/. . . . . . 9. . . . .

(286) . " "

(287) " ".

(288) ?( 9

(289) 6 <A. . . . . . . @7. . . . . . . 1. !

(290) 4 #6. . . . . . . . . . . . . E. ! $ % # & '% #

(291) !"

(292) #. &*"

(293) C-

(294) ?( 8 D

(295)

(296) C-

(297)

(298) "

(299) $

(300)

(301)

(302)

(303)

(304)

(305)

(306)

(307) ( !

(308)

(309)

(310) (

(311)

(312)

(313) 0 &

(314)

(315)

(316) " !

(317) # C-

(318)

(319)

(320)

(321)

(322)

(323)

(324)

(325)

(326) Æ

(327)

(328) %! C-

(329) %

(330) 6 @ A < F @< A < (

(331)

(332) 0 6 @ 5 G < @ 5 (

(333) . . . . . . . . . . . . . . 3.-/.

(334) . " "

(335) " ". . 7 7. . . . 7. . . . . . A @5 A @ 5 . . . . A @ 5 A A A @< D ( ( . . . . . . . . . 7 7 7 . . . . . . <5. . . . . . . @ . A @5 A @ 5 . . . <<. A @5 @< A H

(336) ( <7 < A 3 #

(337)

(338) 5

(339) <8

(340) 5 < A "/

(341) < < C-

(342) ' $ 8 0

(343) " '

(344) 4

(345) 0 :. ;)

(346) >

(347)

(348)

(349)

(350) " . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

(351). .

(352) .

(353). . . /.-/.

(354) . " "

(355) " ". 䈈䈈. − ( Xi − X N ) 2. 䈈䈈. α N K ( X1 , X N ). Xi = X 2. α 1 K ( X 2 , X1 ). α 2 K (X 2 , X 2 ). 䈈䈈. α N K (X 2 , X N ). 䈈䈈. α N K (X N , X N ). α 1 K ( X N , X1 ). α 2 K (X N , X 2 ). α1 㼍㼐㼖㼡㼟㼠㼑㼞㻌. α 2 㼍㼐㼖㼡㼟㼠㼑㼞㻌. 䈈䈈. Xi = X N. α N 㼍㼐㼖㼡㼟㼠㼑㼞㻌. 㼁㼜㼐㼍㼠㼑㻌㼏㼛㼘㼡㼙㼚㻌㻺. . 㼁㼜㼐㼍㼠㼑㻌㼏㼛㼘㼡㼙㼚㻌㻟㼁㼜㼐㼍㼠㼑㻌㼏㼛㼘㼡㼙㼚㻌㻞. 䈈. K ( X i , X j ) = exp(−( X i − X j ) 2 / σ ). λ λ 㻌㻌㻌㻌㻌㼍㼐㼖㼡㼟㼠㼑㼞㻌. α 2 K ( X1 , X 2 ). 䈈䈈. α1 K ( X1 , X1 ). 䈈䈈. Xi = X1. 䈈䈈. − ( Xi − X 2 ) 2. 䈈䈈. − ( Xi − X1 ) 2. 䈈䈈. 㼁㼜㼐㼍㼠㼑㻌㼏㼛㼘㼡㼙㼚㻌㻝. -

(356)

(357) .

(358) . . ' ""

(359)

(360) &*"

(361) 5 <F *& 0 $ 2

(362) $ 9 " 4

(363) 4 ' "

(364) # = "

(365) 4 )

(366) . "

(367) !

(368)

(369) " " " "

(370) ?

(371)

(372)

(373)

(374) &

(375)

(376)

(377) = # 4

(378) ?( E '

(379) (

(380)

(381) 0 . 4.-/.

(382) . " "

(383) " ". Euclidean calculation. 1100 x 640 um. DAC WTA. Learners. Exp. array 400 x 320 um. .

(384)

(385) % " . Iλ. . ∑I row. Ic. Iα. Ic. 䈈䈈.

(386) ". . I unit. . out _ i. ∑I. array. out _ i. Iλ. 䈈䈈.

(387) %" ". . " " D '

(388)

(389)

(390) ,

(391) @ @

(392) . . . . -5.-/. . . . .

(393) . " "

(394) " ". . . @ 7 .

(395) Î Î ¾. . <2. . . <9 A + " " & > ?( <8

(396)

(397)

(398) 4

(399)

(400) ' "

(401) " 4 $ 1 3 #

(402) " ?( <8 > "/

(403) "

(404)

(405) 4 $ E

(406) 4

(407)

(408) 4 > :<; ?( <5

(409) @ '

(410)

(411)

(412) &*" "

(413) #

(414) . . @. . . . . . . . .

(415). .

(416).

(417). . .

(418)

(419)

(420)

(421) &*" ' &*" 7"=

(422) &*" 12"=

(423)

(424) #

(425)

(426)

(427) &*" (.

(428)

(429)

(430) #

(431) .

(432)

(433) $

(434)

(435)

(436)

(437)

(438)

(439) . --.-/.

(440) . " "

(441) " " Alpha values (presented by Ic) during learning. Gaussian kernel feature. domain description 㻝㻚㻞. Vbias=0.8V Current (A). Current (A). number of SVs is 6. 㻝. support vectors. nonsupport vectors. 㻜㻚㻤㻜㻚㻢㻜㻚㻠㻜㻚㻞㻜㻜. input voltage (V). 㻜㻚㻞. 㻜㻚㻠. 㻜㻚㻢. 㻜㻚㻤. 㻝. 㻝㻚㻞. 㻜㻚㻤. 㻝. 㻝㻚㻞. Time(s). . Alpha values (presented by Ic) during learning. Gaussian kernel feature. domain description 㻝. Current (A). Vbias=1.2V Current (A). 㻜㻚㻥. number of SVs is 10 support vectors. 㻜㻚㻤. nonsupport vectors. 㻜㻚㻣㻜㻚㻢㻜㻚㻡㻜㻚㻠㻜㻚㻟㻜㻚㻞㻜㻚㻝㻜㻜. input voltage (V). 㻜㻚㻞. 㻜㻚㻠. 㻜㻚㻢. Time (s). . .

(442) &

(443) "

(444) "

(445)

(446)

(447) $ * $ !. * *" "

(448) , .

(449) "/

(450) %

(451) " " %" % &.

(452) &

(453) ' %! *

(454) * "

(455) , .

(456) "/ %

(457) " " %" %

(458) &

(459) . ( !)

(460)

(461)

(462)

(463) !" . '

(464)

(465)

(466) "" ?

(467) . $

(468) @ 5 1 5 1 @ 5 1 5 1 !

(469) " <1 <1

(470) +-!*?

(471) $ G / !

(472)

(473) 5 5F <5 5 F $

(474)

(475) #

(476) 5 1 D

(477) $ F

(478)

(479) # &*"

(480)

(481)

(482) !

(483) 4

(484) . . . . . .

(485). . . . -0.-/. .

(486) . " "

(487) " ". ...... ....... . learning SVDD classify Chip.

(488)

(489) 01 2.

(490)

(491) . . $ F > ' #

(492) $ F +

(493) ( *(!)

(494)

(495) #

(496)

(497) !" . $ "

(498) #

(499)

(500) &*" ! ""

(501) -1.-/.

(502) . " "

(503) " ". ...... learning. SVDD Chip 1. classify. learning. SVDD Chip 2. classify. learning. SVDD Chip 3. classify. ...... ...... ....... ...... .......

(504) .

(505) .

(506) &

(507)

(508) .

(509) .

(510) "

(511) .

(512) " " 3 %

(513) " "

(514) . &*" I

(515)

(516)

(517) *&!%"75

(518)

(519)

(520) 12"

(521)

(522) "

(523) 4 " --?=

(524) &*" $ G

(525) (

(526)

(527) +

(528) # #

(529) #

(530) !" . '

(531) #

(532) " #

(533)

(534)

(535)

(536) $ " " 0

(537) # +

(538)

(539)

(540) / ! Æ >

(541)

(542)

(543) '

(544) # &*"

(545) >

(546) ' I

(547)

(548) -).-/.

(549) . " "

(550) " ".

(551)

(552) &*"

(553)

(554)

(555)

(556) $ <5 &*" !

(557)

(558)

(559) ' ! # $

(560) #

(561)

(562) I

(563) # 4

(564) '

(565) >

(566) # &*"

(567) !

(568)

(569)

(570) $

(571) : "4 ;

(572) "

(573) ($. #

(574) . &*"

(575)

(576)

(577)

(578) /

(579)

(580)

(581)

(582)

(583) ! %!

(584)

(585) #

(586)

(587)

(588)

(589) &

(590)

(591) "

(592) %! +

(593)

(594)

(595)

(596) &*" *

(597)

(598)

(599) !! &*"

(600)

(601)

(602)

(603) '

(604)

(605)

(606) ! $-' /

(607)

(608) $-'

(609)

(610) "

(611)

(612) / $

(613) "

(614)

(615)

(616) . '

(617)

(618)

(619)

(620) ' %!

(621)

(622) " -2.-/.

(623) . " "

(624) " ". .

(625)

(626)

(627) %* *

(628) "

(629) *

(630) . , , !

(631)

(632) ' %! '

(633) &*" &*" &*" B I

(634)

(635) <E< 721F <1 7 <G 9E 7 12 I

(636)

(637) % J 5 8< <G"= 8 87 71 E <5 , J IJ' 9 <5 <5 ='*. .

(638)

(639)

(640) %* *

(641) "

(642) "*

(643) . , ,

(644) !

(645)

(646) ' %! $-' ' %! '

(647) 3 &*" &*" %

(648) $ &/" " $ 7 ' ' I

(649) * 5 F9 3 5 FG 5 571 7 52 5 589 - . . . . .

(650) ?

(651) "

(652)

(653) 5 1 3

(654)

(655) &*"

(656)

(657)

(658)

(659) 12"

(660)

(661)

(662)

(663) ' "

(664)

(665) &*" $

(666)

(667)

(668) 0 4 . . %! = ? * =?* . ,

(669) * - * ! * =

(670) ! !. -(.-/.

(671) . " "

(672) " ". . < 6

(673)

(674)

(675) I K <GG9 78 7 *

(676) - % 6 !??? ' % - 75<7 <9F9 8 I B H 3 - 6 !??? ,

(677) 75<7 <2G 2 K % K $ L6 !??? - 755E 87E7 9 = ' ' 3 , 6 !??? I I 7558 GG8 1 ' 3

(678) 6 !??? %! 7551 G17 E "K - 3 ' - ? + 6 - !

(679) * 755F F15 F B B 6 !??? * 75<5 <9<8 G , L 6 H H ' - 75<7 52=?<5 <5 $ $ *

(680)

(681) K - 6 - !??? ! * 3"! *

(682) 6 ' 75<< 89< << ! ' 4 H L

(683) ? K 06 - !??? ! * - 6 ' 75<7 25F <7 K D 6 - !??? ! * !

(684) 755F <9 <8 = H , - D =6 - ?

(685) ' # I I <GGG 79< <2 = H , - D =6 - , % <GGG <<G< <9 + D L + 6 - !??? ! * I *

(686) 75<5 F72 <1 H M H K % I6 - !??? ! * ? *

(687)

(688) * 75<< 7595 <E M + M *6 - !??? * !

(689) 755G 2F1 <F % L K K L * L K L6 - !??? * = * 755G 9FE< <G % % L K D K % M C K 6 !??? 75<< -3.-/.

(690) . " "

(691) " ". 8F2G 75 , ' - ' 6 - !??? ! * %! = 75<7 17 7< , L 6 - !??? ! D * I I ' 75<7 < 77 B + %6 !??? * 7559 77 78 = 6 !??? ! H * I I <GG< 2E9 72 "D * * "+ + "+ "K * = = ! ! + 6 !??? H ?

(692) * 75<< 89E 79 - L , L 6 % I *

(693) + 75<7 E71E 1<E 71 ' I B I + 6 *

(694) &4 !

(695) % *&!%"75 , *.*"559"G1 7E K 6 !??? I I 7558 <<22. -/.-/.

(696)

No results found