-$3$1(6( $1' .25($1 92,&( 6($5&+ 0LNH 6FKXVWHU DQG .DLVXNH 1DNDMLPD *RRJOH ,QF 86$ &b+?mbi2`- FBbmF2'!;QQ;H2X+QK

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

RI (QJOLVK RULJLQ DV ZHOO DV 85/V QXPEHUV GDWHV QDPHV (0DLO DGGUHVVHV DEEUHYLDWLRQV SXQFWXDWLRQ DQG RWKHU VSHFLDO FDVHV ZKLFK PRVWO\ H[LVW LQ (QJOLVK DV ZHOO EXW EHFDPH HYHQ PRUH FRPSOLFDWHG ZLWK WKH PL[WXUH RI PXOWLSOH FKDUDFWHU VHWV $W *RRJOH WKH -DSDQHVH YRLFH VHDUFK V\VWHP ZDV DPRQJ WKH ILUVW GHYHORSHG DIWHU 86 (QJOLVK DQG OHG WR D VLJQLILFDQW VLPSOLILFDWLRQ DQG VSHHGXS RI WKH SURFHVV WR GHDO ZLWK QHZ ODQJXDJHV LQ JHQHUDO ± LQ ODUJH SDUWV WKH SURFHVV GHVFULEHG EHORZ EHFDPH WKH VWDQGDUG WR EXLOG DQG PDLQWDLQ DQ\ QHZ ODQJXDJH DQG ZDV ILUVW DSSOLHG WR .R UHDQ ZLWKRXW DQ\ PRGLILFDWLRQV VKRUWO\ DIWHU -DSDQHVH ZDV ODXQFKHG LQ 1RYHPEHU  -DSDQHVH DQG .RUHDQ DUH LQ WHUPV RI XVDJH FXU UHQWO\  RXU ODUJHVW LQWHUQDWLRQDO ODQJXDJHV DIWHU 86 (QJOLVK 9RLFH VHDUFK IRU 0DQGDULQ &KLQHVH DQG &DQWRQHVH DV ODXQFKHG E\ *RRJOH LQ  DQG  DUH GHVFULEHG LQ >@ DQG >@

,QGH[ 7HUPV² 6SHHFK UHFRJQLWLRQ YRLFH VHDUFK -DSDQHVH .R UHDQ

 $&2867,& '$7$ &2//(&7,21

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¶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

978-1-4673-0046-9/12/$26.00 ©2012 IEEE

5149

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±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

ICASSP 2012

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± ZH ZDQWHG WR XVH D WHFKQLTXH WKDW GHSHQGV RQO\ RQ WKH GDWD DQG SUHIHUDEO\ GRHV QRW SURGXFH DQ\ 229V :H GHYHORSHG D WHFKQLTXH :RUG3LHFH0RGHO WR OHDUQ ZRUG XQLWV IURP ODUJH DPRXQWV RI WH[W DXWRPDWLFDOO\ DQG LQ FUHPHQWDOO\ E\ UXQQLQJ D JUHHG\ DOJRULWKP DV GHVFULEHG EHORZ 7KLV JLYHV XV D XVHUVSHFLILHG QXPEHU RI ZRUG XQLWV ZH RIWHQ XVH N ZKLFK DUH FKRVHQ LQ D JUHHG\ ZD\ ZLWKRXW IRFXVLQJ RQ VHPDQWLFV WR PD[LPL]H OLNHOLKRRG RQ WKH ODQJXDJH PRGHO WUDLQLQJ GDWD ± LQFLGHQ WDOO\ WKH VDPH PHWULF WKDW ZH XVH GXULQJ GHFRGLQJ 7KH DOJRULWKP WR ILQG WKH DXWRPDWLFDOO\ OHDUQHG ZRUG LQYHQWRU\ HIILFLHQWO\ ZRUNV LQ VXPPDU\ DV IROORZV

:KLOH QRUPDO -DSDQHVH WH[W GRHV QRW KDYH VSDFHV EHWZHHQ ZRUGV WKHUH DUH FDVHV ZKHUH VRPH VSDFHV DUH XVHG LQ RWKHU ODQJXDJHV .R UHDQ IRU H[DPSOH  DQG DOVR VRPHWLPHV LQ -DSDQHVH )RU H[DPSOH VHDUFK NH\ZRUGV DUH QRUPDOO\ VHSDUDWHG E\ VSDFHV ³ऍ䌎婋婶婃 婕´ DV DUH ZRUGV ZLWK $6&,, FKDUDFWHUV (QJOLVK ZRUGV DEEUHYLD WLRQV QXPEHUV HWF ZKHQ HPEHGGHG LQ -DSDQHVH WH[W ,QLWLDOO\ ZH UDQ WKH SURGXFWLRQ V\VWHP ZLWKRXW WKH DELOLW\ WR JOXH ZRUG SLHFHV WR JHWKHU DJDLQ ± WKLV GLGQ¶W PDWWHU IRU SRSXODU ZRUGV DQG VKRUW TXHULHV DV WKRVH EHFDPH FRPSOHWH XQLWV EXW ZDV DQQR\LQJ IRU ORQJHU DQG UDUH TXHULHV DV VSDFHV DSSHDUHG ZKHUH WKH\ VKRXOGQ¶W KDYH EHHQ ,QLWLDOO\ ZH H[SHULPHQWHG ZLWK D ELJUDPEDVHG WHFKQLTXH UXQ RQ WKH K\SRWKH VHV WR GHFLGH ZKHUH WR SXW VSDFHV EXW DEDQGRQHG WKLV DIWHU ILQGLQJ WKDW LW LV PXFK PRUH HIILFLHQW DQG HOHJDQW WR GHDO ZLWK WKLV SUREOHP ULJKW GXULQJ GHFRGLQJ 7R EH DEOH WR OHDUQ ZKHUH WR SXW VSDFHV IURP WKH GDWD DQG PDNH LW SDUW RI WKH GHFRGLQJ VWUDWHJ\ ZH XVH WKH IROORZLQJ WHFKQLTXH

 ,QLWLDOL]H WKH ZRUG XQLW LQYHQWRU\ ZLWK WKH EDVLF 8QLFRGH FKDU DFWHUV .DQML +LUDJDQD .DWDNDQD IRU -DSDQHVH +DQJXO IRU .RUHDQ DQG LQFOXGLQJ DOO $6&,, HQGLQJ XS ZLWK DERXW  WRWDO IRU -DSDQHVH DQG  IRU .RUHDQ

 :KHQ VHJPHQWLQJ WKH /0 GDWD ZLWK :RUG3LHFH0RGHO DWWDFK D VSDFH PDUNHU ZH XVH DQ XQGHUVFRUH RU D WLOGH EHIRUH DQG DIWHU HDFK XQLW ZKHQ \RX VHH D VSDFH RWKHUZLVH GRQ¶W DWWDFK (DFK ZRUG XQLW FDQ KHQFH DSSHDU LQ IRXU GLIIHUHQW IRUPV ZLWK XQGHUVFRUH RQ ERWK VLGHV WKHUH ZDV D VSDFH RULJLQDOO\ RQ ERWK VLGHV LQ WKH GDWD  ZLWK D VSDFH PDUNHU RQO\ RQ WKH OHIW RU RQO\ RQ WKH ULJKW DQG ZLWKRXW VSDFH PDUNHUV DW DOO 7KH ZRUG LQ YHQWRU\ LV QRZ ODUJHU SRVVLEO\ XS WR IRXU WLPHV WKH RULJLQDO VL]H LI HYHU\ FRPELQDWLRQ H[LVWV

 %XLOG D ODQJXDJH PRGHO RQ WKH WUDLQLQJ GDWD XVLQJ WKH LQYHQ WRU\ IURP   *HQHUDWH D QHZ ZRUG XQLW E\ FRPELQLQJ WZR XQLWV RXW RI WKH FXUUHQW ZRUG LQYHQWRU\ WR LQFUHPHQW WKH ZRUG XQLW LQYHQWRU\ E\ RQH &KRRVH WKH QHZ ZRUG XQLW RXW RI DOO SRVVLEOH RQHV WKDW LQFUHDVHV WKH OLNHOLKRRG RQ WKH WUDLQLQJ GDWD WKH PRVW ZKHQ DGGHG WR WKH PRGHO  *RWR  XQWLO D SUHGHILQHG OLPLW RI ZRUG XQLWV LV UHDFKHG RU WKH OLNHOLKRRG LQFUHDVH IDOOV EHORZ D FHUWDLQ WKUHVKROG 7UDLQLQJ RI WKH VHJPHQWHU LV D FRPSXWDWLRQDOO\ H[SHQVLYH SURFH GXUH LI GRQH EUXWHIRUFH DV IRU HDFK LWHUDWLRQ DGGLWLRQ RI D QHZ ZRUG SLHFH XQLW E\ FRPELQLQJ WZR H[LVWLQJ RQHV  DV DOO SRVVLEOH SDLU FRP ELQDWLRQV QHHG WR EH WHVWHG DQG D QHZ ODQJXDJH PRGHO QHHGV WR EH EXLOW 7KLV ZRXOG EH RI FRPSXWDWLRQDO FRPSOH[LW\ O(K 2 ) SHU LWHU DWLRQ ZLWK K EHLQJ WKH QXPEHU RI FXUUHQW ZRUG XQLWV 7KH WUDLQLQJ DOJRULWKP FDQ EH VSHG XS VLJQLILFDQWO\ E\ D WHVWLQJ RQO\ WKH SDLUV WKDW DFWXDOO\ H[LVW LQ WKH WUDLQLQJ GDWD E WHVWLQJ RQO\ SDLUV ZLWK D VLJQLI LFDQW FKDQFH RI EHLQJ WKH EHVW IRU H[DPSOH KLJK SULRUV  F FRPELQ LQJ VHYHUDO FOXVWHULQJ VWHSV LQWR D VLQJOH LWHUDWLRQ ZKLFK LV SRVVLEOH IRU JURXSV RI SDLUV WKDW GRQ¶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± LVR ODWLQJ WKH EDVLF FKDUDFWHUV DQG FRPELQLQJ WKHP DV JRLQJ GRZQ WKH WUHH ZLOO JLYH D GHWHUPLQLVWLF VHJPHQWDWLRQ LQ OLQHDU WLPH ZLWK UHVSHFW WR WKH OHQJWK RI WKH VHQWHQFH 6LQFH WKH ZRUG LQYHQWRU\ LV OLPLWHG WR N XQLWV WKDW FDQ PRGHO HYHU\ SRVVLEOH FKDUDFWHU VHTXHQFH DQG GRHV QRW SURGXFH DQ\ 229V LW VLPSOLILHV VRPHZKDW WKH JHQHUDWLRQ RI WKH GLFWLRQDU\ ± LQ RXU FDVH URXJKO\ RQO\  RI WKH ZRUGV KDYH PRUH WKDQ RQH SURQXQFLDWLRQ $V GHVFULEHG LQ PRUH GHWDLO EHORZ ZH IRXQG WKDW DGGLQJ WRR PDQ\ SURQXQFLDWLRQV KXUWV SHUIRUPDQFH OLNHO\ EHFDXVH RI WKH LQFUHDVHG QXPEHU RI VLPLODU K\SRWKHVHV GXULQJ WKH DOLJQPHQW IRU WUDLQLQJ DQG GHFRGLQJ

5150

 7KH RULJLQDO ODQJXDJH PRGHO GDWD LV XVHG ³DV ZULWWHQ´ PHDQ LQJ VRPH GDWD ZLWK VRPH ZLWKRXW VSDFHV

 %XLOG /0 DQG GLFWLRQDU\ EDVHG RQ WKLV QHZ LQYHQWRU\  'XULQJ GHFRGLQJ WKH EHVW SDWK DFFRUGLQJ WR WKH PRGHO ZLOO EH FKRVHQ ZKLFK SUHVHUYHV ZKHUH WR SXW VSDFHV DQG ZKHUH QRW 7KH DWWDFKHG VSDFH PDUNHUV REYLRXVO\ KDYH WR EH ILOWHUHG RXW IURP WKH GHFRGLQJ RXWSXW IRU FRUUHFW GLVSOD\ $VVXPLQJ WKH FRPPRQ VFHQDULR ZKHUH WKH GHFRGHU SXWV DXWRPDWLFDOO\ D VSDFH EHWZHHQ HDFK XQLW LQ WKH RXWSXW VWULQJ WKLV SURFHGXUH WKHQ LV D 5HPRYH DOO VSDFHV ´ ´ ! ´´ E 5HSODFH GRXEOH VSDFH PDUNHU E\ VSDFH ´BB´ ! ´ ´ F 5HPRYH UHPDLQLQJ VLQJOH VSDFH PDUNHUV ´B´ ! ´´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´ ऍ䌎 ⒌⋤ᔇ姝௺ⱺ´ RULJLQDO WH[W  ´ ऍ䌎 ⒌⋤ᔇ 姝௺ⱺ´ DIWHU VHJPHQWDWLRQ  ´ 埍ऍ䌎埍 埍⒌⋤ᔇ 姝௺ⱺ埍´ DIWHU DGGLWLRQ RI XQGHU VFRUHV XVHG IRU /0 WUDLQLQJ GLFWLRQDU\ HWF  ´ 埍ऍ䌎埍 埍⒌⋤ᔇ 姝௺ⱺ埍´ GHFRGHU UHVXOW  ´ ऍ䌎 ⒌⋤ᔇ姝௺ⱺ´ GLVSOD\HG RXWSXW 7KH GDWDGULYHQ ODQJXDJHLQGHSHQGHQW WHFKQLTXHV GHVFULEHG DERYH ZLWK D PLQLPXP RI UXOHEDVHG SURFHVVLQJ DUH VXFFHVVIXOO\ XVHG IRU VRPH RI RXU $VLDQ ODQJXDJHV -DSDQHVH .RUHDQ 0DQGDULQ DFURVV PXOWLSOH DSSOLFDWLRQV 9RLFH 6HDUFK GLFWDWLRQ DV ZHOO DV
7HVW 6HW UDQGN UDQGN UDQGN WRSN WRSN WRSN

1XQLWV N N N N N N

1ZRUGV      

3HUSOH[LW\      

&RVW6HQW      

3URQXQFLDWLRQV DGGHG      

$FFXUDF\      

7DEOH  1XPEHU RI ZRUGV ZRUG SHUSOH[LW\ DQG DYHUDJH FRVWVHQWHQFH IRU VHYHUDO VL]HV RI VHJPHQWHU LQYHQWRU\

7DEOH  1XPEHU RI SURQXQFLDWLRQV DGGHG IRU QDPHV DQG UHVXOWLQJ VHQWHQFH DFFXUDF\ RQ D -DSDQHVH WHVW VHW

 /$1*8$*( 02'(/,1*

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

0DNLQJ WKH DVVXPSWLRQ WKDW SHRSOH VD\ ZKDW WKH\ W\SH ZH XVHG ORFDO ZHE TXHULHV IURP DQRQ\PL]HG ORJV DV RXU /0 GDWD 2QFH WKH ZRUG LQYHQWRU\ LV GHFLGHG LW LV VWUDLJKWIRUZDUG WR HVWLPDWH /0V LQ RXU FDVH XVXDOO\ HQWURS\SUXQHG  WR JUDPV ZLWK .DW] EDFNRII DIWHU UHPRYLQJ XQZDQWHG V\PEROV HWF DV PXFK DV SRVVLEOH VLPLODU WR ZKDW LV GHVFULEHG LQ >@ )RU -DSDQHVH DQG .RUHDQ ZH PDGH QR GLVWLQFWLRQ EHWZHHQ QDWLYH DQG (QJOLVK ORDQ ZRUGV DERXW  RI WKH GLFWLRQDU\ IRU -DSDQHVH DQG  IRU .RUHDQ DUH $6&,, ZRUGV PRVW RI WKHP (QJOLVK ±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³$.%´ ³:,1'6´ ³.LV0\)W´ RU PL[HG FKDUDFWHU ZRUGV OLNH ³娙婅࣒ 4´ WKUHH FKDUDFWHU VHWV PL[HG LQ D VLQJOH ZRUG RU ³ਂέFRP´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³*RRJOH -DSDQHVH ,QSXW´

5151

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± LQ WKLV FDVH LW LV UHDVRQDEOH WR DVVXPH WKDW WKH FRUUHFW ZHE SDJH ZDV IRXQG JLYHQ WKH VHDUFK HQJLQH  :HE 6FRUH# WKHUHIRUH VLPSO\ WHVWV LI WKH WRS K\SRWKHVLV 85/ DQG WKH WRS UHIHUHQFH 85/ DUH HTXDO 7KLV PHDVXUH KDG WZR DGYDQWDJHV RYHU UHJXODU ZRUG HUURU UDWH  $V RXU ZRUG GHILQLWLRQ LV GHILQHG E\ WKH

7LPH EHIRUH ODXQFK EHIRUH ODXQFK EHIRUH ODXQFK ODXQFK DIWHU ODXQFK DIWHU ODXQFK DIWHU ODXQFK

7HVW 6HW LQLWLDO UHDG LQWHUQDO UHDO LQWHUQDO UHDO LQWHUQDO UHDO LQWHUQDO UHDO SURG $ UHDO SURG % UHDO

:6&#       

7UDLQLQJ N UHDG N UHDG N UHDG N UHDG 0 SURG 0 SURG 0 SURG

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³ड़ኛᆤ´ ³娭婕娨婕 娫娈´ ³姚娂姕娂姘妵´ ³QLQWHQGR´ DUH DOO YDOLG ZD\V WR VHDUFK IRU WKH -DSDQHVH JDPH FRPSDQ\ DOO DUH SURQRXQFHG WKH VDPH DQG KDYH WKH VDPH PHDQLQJ ± WKH UHVXOW 85/ VKRXOG EH WKH VDPH IRU DQ\ RI WKHVH :HE6FRUH WDNHV FDUH RI WKHVH DPELJXLWLHV WKH UHIHUHQFH ZLOO FRQWDLQ RQO\ RQH RI WKH DERYH DQG KHQFH JLYHV D EHWWHU HVWLPDWH IRU WKH WUXH HUURU UDWH WKDQ IRU H[DPSOH VHQWHQFH HUURU UDWH

WHFKQLTXHV GHVFULEHG LQ WKLV SDSHU PD\ VHHP REYLRXV WR XVH QRZ EXW ZHUHQ¶W DW WKH WLPH RI GHYHORSPHQW 2YHU WLPH ZH IRXQG WKDW ZKLOH IRU PRVW VSRNHQ TXHULHV DQG WH[W WKH WHFKQLTXHV GHVFULEHG DERYH ZRUN ZHOO DQG RQ DYHUDJH TXDOLW\ RI WKH V\VWHP LPSURYHV ZLWK HYHU\ XSGDWH WKHUH DUH VWLOO VRPH SURE OHPDWLF FRUQHU FDVHV ZKLFK FDQ RQO\ EH DGGUHVVHG E\ DGGLQJ PRUH FRPSOH[LW\ WR WKH V\VWHP )RU H[DPSOH UDUH ORQJHU QXPEHUV DUH RIWHQ QRW UHFRJQL]HG FRUUHFWO\ EHFDXVH WKH FRPELQHG SURQXQFLDWLRQV RI WKHLU VHJPHQWDWLRQ ZLOO QRW PDWFK WKH DFWXDO SURQXQFLDWLRQ H[DP SOH LQ (QJOLVK ³´ VDLG DV ³RQH PLOOLRQ WZR KXQGUHG WKLUW\ IRXU WKRXVDQG ILYH KXQGUHG DQG VL[W\ VHYHQ´ ± WKH SURQXQFLDWLRQV RI D SRVVLEOH VHJPHQWDWLRQ RI ³ ´ ZLOO OLNHO\ QRW FRYHU WKH RULJLQDO SURQXQFLDWLRQ  ,Q JHQHUDO VLPLODU LV WUXH IRU DQ\ FKDUDFWHU VHTXHQFHV ZKRVH SURQXQFLDWLRQ GHSHQGV RQ D ZLGHU FRQWH[W WKLQN SULFHV ³´ ! ³RQH KXQGUHG DQG IRXU GROODUV´  GDWHV HWF ±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¶V -DSDQHVH DQG .RUHDQ YRLFH VHDUFK V\VWHPV :KLOH GHYHORSLQJ WKHVH V\VWHPV ZH IRXQG WKDW WKH VLPSOLILHG SURFHVV GHVFULEHG DERYH EDVHG RQ PRGHOLQJ LQ WKH ZULWWHQ GRPDLQ ZLWK D FORVHG GLFWLRQDU\ EXW LQILQLWH YRFDEXODU\ EDVHG RQ WKH :RUG3LHFH0RGHO VHJPHQWHU UHVXOWHG LQ UHODWLYHO\ ORZFRPSOH[LW\ V\VWHPV WR PDLQWDLQ DQG XSGDWH 8VLQJ XQVXSHUYLVHG GDWD IRU LQLWLDO WUDLQLQJ DQG VXEVHTXHQW XSGDWHV RI WKH DFRXVWLF PRGHOV KDV FXW GRZQ RXU GHYHORSPHQW WLPH IRU QHZ ODQJXDJHV VLJQLILFDQWO\ 0DQ\ RI WKH

5152

>@ -RKDQ 6FKDONZ\N 'RXJ %HHIHUPDQ )UDQFRLVH %HDXID\V %LOO %\UQH &LSULDQ &KHOED 0LNH &RKHQ 0DU\DP *DUUHWW DQG %ULDQ 6WURSH ³*RRJOH 6HDUFK E\ 9RLFH $ &DVH 6WXG\´ LQ $GYDQFHV LQ 6SHHFK 5HFRJQLWLRQ 0RELOH (QYLURQPHQWV &DOO &HQWHUV DQG &OLQLFV $P\ 1HXVWHLQ (G 6SULQJHU9HUODJ  >@ -LXORQJ 6KDQ *HQTLQJ :X =KLKRQJ +X ;LOLX 7DQJ 0DUWLQ -DQVFKH DQG 3HGUR 0RUHQR ³6HDUFK E\ 9RLFH LQ 0DQGDULQ &KLQHVH´ LQ 3URFHHGLQJV RI ,QWHUVSHHFK  >@ @ 7KDG +XJKHV .DLVXNH 1DNDMLPD /LQQH +D $WXO 9DVX 3HGUR 0RUHQR DQG 0LNH /H%HDX ³%XLOGLQJ WUDQVFULEHG VSHHFK FRU SRUD TXLFNO\ DQG FKHDSO\ IRU PDQ\ ODQJXDJHV´ LQ 3URFHHGLQJV RI ,QWHUVSHHFK  >@ &LSULDQ &KHOED -RKDQ 6FKDONZ\N 7KRUVWHQ %UDQWV 9LGD +D %RXORV +DUE :LOO 1HYHLWW &DUROLQD 3DUDGD DQG 3HQJ ;X ³4XHU\ ODQJXDJH PRGHO IRU YRLFH VHDUFK´ LQ 3URFHHGLQJV RI WKH  ,((( :RUNVKRS RQ 6SRNHQ /DQJXDJH 7HFKQRORJ\  >@ 0HFDE ³KWWSPHFDEVRXUFHIRUJHQHW´  >@ ³KWWSWRROVJRRJOHFRPGOSDJHMDSDQHVHLQSXWHXODKWPO´  >@ 0DUN *DOHV ³6HPLWLHG FRYDULDQFH PDWULFHV IRU +LGGHQ 0DUNRY 0RGHOV´ LQ ,((( 7UDQVDFWLRQV RQ 6SHHFK DQG $XGLR 3URFHVVLQJ  >@ 'DQLHO 3RYH\ 'LPLWUL .DQHYVN\ %ULDQ .LQJVEXU\ %KX YDQD 5DPDEKDGUDQ *HRUJH 6DRQ DQG .DUWKLN 9LVZHVZDULDK ³%RRVWHG 00, IRU PRGHO DQG IHDWXUH VSDFH GLVFULPLQDWLYH WUDLQLQJ´ LQ ,&$663  >@ 2SHQ)67 ³KWWSZZZRSHQIVWRUJ´ 

japanese and korean voice search - Research at Google

iPhone and Android phones in US English [1]. Soon after it was ..... recognition most important groups of pronunciations: a) the top 10k words as occurring in the ...

141KB Sizes 12 Downloads 411 Views

Recommend Documents

Google Search by Voice - Research at Google
May 2, 2011 - 1.5. 6.2. 64. 1.8. 4.6. 256. 3.0. 4.6. CompressedArray. 8. 2.3. 5.0. 64. 5.6. 3.2. 256 16.4. 3.1 .... app phones (Android, iPhone) do high quality.

Google Search by Voice - Research at Google
Feb 3, 2012 - 02/03/2012 Ciprian Chelba et al., Voice Search Language Modeling – p. 1 ..... app phones (Android, iPhone) do high quality speech capture.

Google Search by Voice - Research at Google
Kim et al., “Recent advances in broadcast news transcription,” in IEEE. Workshop on Automatic ... M-phones (including back-off) in an N-best list .... Technology.

Voice Search for Development - Research at Google
26-30 September 2010, Makuhari, Chiba, Japan. INTERSPEECH ... phone calls are famously inexpensive, but this is not true in most developing countries.).

google's cross-dialect arabic voice search - Research at Google
our DataHound Android application [5]. This application displays prompts based on common ... pruning [10]. All the systems described in this paper make use of ...

Search by Voice in Mandarin Chinese - Research at Google
client application running on an Android mobile telephone with an intermittent ... 26-30 September 2010, Makuhari, Chiba, Japan .... lar Mandarin phone.

Google Search by Voice: A case study - Research at Google
of most value to end-users, and supplying a steady flow of data for training systems. Given the .... for directory assistance that we built on top of GMM. ..... mance of the language model on unseen query data (10K) when using Katz ..... themes, soci

Google Search by Voice
Mar 2, 2012 - Epoch t+1. SSTable. Feature-. Weights: Epoch t. SSTable. Utterances. SSTableService. Rerank-Mappers. Identity-Mappers. Reducers. Cache.

Query-Free News Search - Research at Google
Keywords. Web information retrieval, query-free search ..... algorithm would be able to achieve 100% relative recall. ..... Domain-specific keyphrase extraction. In.

Automata Evaluation and Text Search Protocols ... - Research at Google
Jun 3, 2010 - out in the ideal world; of course, in the ideal world the adversary can do almost ... †Dept. of Computer Science and Applied Mathematics, Weizmann Institute and IDC, Israel. ... Perhaps some trusted certification authorities might one

User Preference and Search Engine Latency - Research at Google
branding (yellow vs. blue coloring) and latency (fast vs. slow). The fast latency was fixed ... the second 6 blocks, assignments of query set to engine were swapped. .... tion of one or more independent (predictor variables) plus a random error, e ..

Brand Attitudes and Search Engine Queries - Research at Google
Sep 30, 2016 - more searches that included the word “android” were submitted by ... interest in a brand (Lehmann et al., 2008; Srinivasan et al., 2010; ..... queries for a period of 14 days received an email or phone reminder to stay logged.

QUERY LANGUAGE MODELING FOR VOICE ... - Research at Google
ABSTRACT ... data (10k queries) when using Katz smoothing is shown in Table 1. ..... well be the case that the increase in PPL for the BIG model is in fact.

ICMI'12 grand challenge: haptic voice recognition - Research at Google
Oct 26, 2012 - Voice Recognition (HVR) [10], a novel multimodal text en- try method for ... that on desktop and laptop computers with full-sized key- board [4].

recurrent neural networks for voice activity ... - Research at Google
28th International Conference on Machine Learning. (ICML), 2011. [7] R. Gemello, F. Mana, and R. De Mori, “Non-linear es- timation of voice activity to improve ...

Scalable all-pairs similarity search in metric ... - Research at Google
Aug 14, 2013 - call each Wi = 〈Ii, Oi〉 a workset of D. Ii, Oi are the inner set and outer set of Wi ..... Figure 4 illustrates the inefficiency by showing a 4-way partitioned dataset ...... In WSDM Conference, pages 203–212, 2013. [2] D. A. Arb

Query Suggestions for Mobile Search ... - Research at Google
Apr 10, 2008 - suggestions in order to provide UI guidelines for mobile text prediction ... If the user mis-entered a query, the application would display an error ..... Hart, S.G., Staveland, L.E. Development of NASA-TLX Results of empirical and ...

Incremental Clicks Impact Of Mobile Search ... - Research at Google
[2]. This paper continues this line of research by focusing exclusively on the .... Figure 2: Boxplot of Incremental Ad Clicks by ... ad-effectiveness-using-geo.html.

On the Difficulty of Nearest Neighbor Search - Research at Google
plexity to find the nearest neighbor (with a high prob- ability)? These questions .... σ is usually very small for high dimensional data, e.g., much smaller than 0.1).

Evaluating Web Search Using Task Completion ... - Research at Google
for two search algorithms which we call search algorithm. A and search algorithm B. .... if we change a search algorithm in a way that leads users to take less time that ..... SIGIR conference on Research and development in information retrieval ...

Topical Clustering of Search Results - Research at Google
Feb 12, 2012 - that the last theme is easily identifiable even though the last three ..... It goes without saying that we have to add the cost of annotating the short ...

TechWare: Mobile Media Search Resources - AT&T Labs Research
weka): A collection of machine learn- ing algorithms in Java for data mining tasks. It contains data preprocessing, classification, regression, clustering, association rules, and visualization. □ LIBSVM (www.csie.ntu.edu. tw/~cjlin/libsvm): A libra

Predicting Bounce Rates in Sponsored Search ... - Research at Google
Among the best known metrics for these pur- poses is click ... ments in response to search queries on an internet search engine. ... ing aggregate data from the webserver hosting the adver- tiser's site. ..... The top ten scoring terms per source ...