{
    "paper_id": "O07-1006",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T08:07:55.152449Z"
    },
    "title": "",
    "authors": [],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "",
    "pdf_parse": {
        "paper_id": "O07-1006",
        "_pdf_hash": "",
        "abstract": [],
        "body_text": [
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "\u7136\u800c\uff0c\uf974\u5728\u8072\u5b78\u6a21\u578b\u548c\u8a9e\u8a00\u6a21\u578b\u7684\u8a13\uf996\u4e0a\uff0c\u5247\u9700\u8981\u8a08\u7b97\u5168\u9762\u98a8\u96aa(Overall Risk)\uff0c\u4e26 \u4e14\u6700\u5c0f\u5316\u6b64\u5168\u9762\u98a8\u96aa [1]\uff1a (3) dO O) R all \u5176\u4e2d W \u70ba\u8a9e\u97f3\u7279\u5fb5\u5411\uf97e\u5e8f\uf99c O \u5c0d\u61c9\u4e4b\u6b63\u78ba\u8f49\u8b6f\u8a5e\u5e8f\uf99c\uff0c \u70ba \u7684\u4e8b\u524d\u6a5f\uf961(Prior Probability)\uff1b\u5168\u9762\u98a8\u96aa \u662f\u5728\u8a9e\uf906\u7a7a\u9593(\u8a9e\u97f3\u7279\u5fb5\u5411\uf97e\u5e8f\uf99c\u7a7a\u9593)\u4e0a\u4f5c\u7a4d\u5206\uff0c\u70ba\u6240\u6709\u8a13\uf996 \u8a9e\uf906(\u8a9e\u97f3\u7279\u5fb5\u5411\uf97e\u5e8f\uf99c)\u7684\u671f\u671b\u689d\u4ef6\u98a8\u96aa(Expected Conditional Risk)\u3002\u7531\u65bc\u8a13\uf996\u8a9e\uf9be\u6709 \u9650\uff0c\u6545\u5168\u9762\u98a8\u96aa\u53ef\u7c21\u5316\u70ba ) (O P O all R Z \u500b\u8a13\uf996\u8a9e\uf906\u7684\u689d\u4ef6\u98a8\u96aa\u7e3d\u548c\uff1a \u2211 \u2211 = Z \u2211 = \u2208 \u2032 = \u2032 = z W z z z all P O W P W l R R 1 1 ) ) | ( ) ( ( W Z ) ( z P z z z O P O W ) ( ) | (4) \u2032 W , z O ( | O W \u2032 , = {\u03bb } \u0393 \u03bb \u53ca\u8a9e\u8a00\u6a21\u578b \u0393 \u6240\u6c7a\u5b9a\uff0c\uf9a8\u03b8 \uf974\u4e8b\u5f8c\u6a5f\uf961\u5206\u5e03 \u7531\u8072\u5b78\u6a21\u578b \uff0c\u6240\u4ee5\u4e8b\u5f8c \u6a5f\uf961\u6211\u5011\u5c07\u4e4b\u8868\u793a\u70ba ) ; | ( \u03b8 z O W P \u2032 \u2211 \u2211 \u2211 = \u2208 \u2032 = \u2032 \u2032 = = Z z W z z z Z z z z z all O P O W P W W l O P O W R R 1 1 ) ( ) ; | ( ) , ( ) ( ) | ( W \u03b8 ) ( z O P z \uff0c\u5247\u5168\u9762\u98a8\u96aa\u53ef\u6539\u5beb\u6210\uff1a (5) \uf974\u5047\u8a2d \u5c0d\u6240\u6709 O \u5747\u6709\u4e00\u81f4(Uniform)\u7684\u6a5f\uf961\uff0c\u4e14\u6b64\u9805\u8207\u6a21\u578b\uf96b\uf969 \u03bb \u53ca \u0393 \u7121\u95dc\uff0c\u5247\u53ef \u5c07\u6b64\u9805\uf96d\uf976\uff1a \u2211 \u2211 = \u2208 \u2032 \u2032 \u2032 = Z z W z z all O W P W W l R 1 ) ; | ( ) , ( W \u03b8 (6) \u5728\u4f30\u6e2c\u8072\u5b78\u6a21\u578b\u548c\u8a9e\u8a00\u6a21\u578b\u6642\uff0c\u5e0c\u671b\u4f30\u6e2c\u4e4b\u6a21\u578b\u03b8 \u80fd\u5c07\u5168\u9762\u98a8\u96aa\ufa09\u81f3\u6700\u4f4e\uff1a \u2211 \u2211 \u2032 \u2032 \u2032 = Z z z O W P W W l ) ; | ( ) , ( min ar\u011d \u03b8 \u03b8 \u03b8 = \u2208 z W 1 W",
                        "eq_num": "("
                    }
                ],
                "section": "",
                "sec_num": null
            },
            {
                "text": "\u5668 \u5c0d \u6240 \u6709 \u8a13 \uf996 \u8a9e \uf906 ( \u8a9e \u97f3 \u7279 \u5fb5 \u5411 \uf97e \u5e8f \uf99c ) \u7684 \u53ef \u80fd \u8fa8 \uf9fc \u51fa \u5019 \u9078 \u8a5e \u5e8f \uf99c ( )\u7684\u671f\u671b\u97f3\u7d20\u6b63\u78ba\uf961(\u4e5f\u5c31\u662f\u6700\u5c0f\u5316\u8a9e\u97f3\u8fa8\uf9fc\u5668\u5c0d\u6240\u6709\u8a13\uf996 \u8a9e\uf906\u53ef\u80fd\u8fa8\uf9fc\u51fa\u5019\u9078\u8a5e\u5e8f\uf99c \u7684\u671f\u671b\u932f\u8aa4\uf961) \uff0c\u6700\u5c0f\u5316\u97f3\u7d20\u932f\u8aa4\u7684\u76ee\u6a19\u51fd\uf969\u53ef\u8868\u793a\u5982\u4e0b\uff1a ) , ( z i W W A i W W i \u2208 W ) , ( z i W W l } z O { L , , , 3 2 1 W W W z = i W \u2211 \u2211 = Z \u2211 \u2211 = \u2208 = \u2208 = z z i W z i i z Z z z i W z i W W A O p W P W O p W W A O W p z i z i 1 1 ) , ( ) ( ) ( ) | ( ) , ( ) | ( ) W W \u03bb MPE F (\u03bb (8) \u5176\u4e2d \u53ef\u7528\u8a9e\u97f3\u8fa8\uf9fc\u5668\u7522\u751f\u7684\u8a5e\u5716 \uf92d\u8fd1\u4f3c[11]\uff0c\u56e0\u6b64\u76ee\u6a19\u51fd\uf969\u53ef\u9032\u4e00\u6b65\u8868\u793a \u6210\uff1a ) ( z O p lattice z , W \u2211 \u2211 \u2211 \u2208 \u2208 z z i W W k k z i i W W A W P W O p W P lattice z i lattice z k ) , ( ) ( ) | ( ) ( ) , , W W \u03bb i k z O z \u2248 ) (\u03bb z W O p | ( \u03bb MPE F (9) \u5176\u4e2dW \u8207W \u5206\u5225\u8868\u793a\u8a5e\u5716 \u4e0a\u4efb\uf978\u689d\u5019\u9078\u8a5e\u5e8f\uf99c(\u5047\u8a2d \u5c0d\u61c9\u7684\u6b63\u78ba\u8a5e\u5e8f\uf99cW \u4ea6\u5305\u542b\u5728\u8a5e\u5716\uf9e8)\u3002 lattice z , W \u70ba\uf9ba\u5c0d\u76ee\u6a19\u51fd\uf969 ) \u03bb ( MPE F \u9032\ufa08\u6700\u4f73\u5316\uff0cPovey \u7b49\u4eba\u63d0\u51fa\u6700\u5c0f\u5316\u97f3\u7d20\u932f\u8aa4\u7684\u5f31\u6027 (Weak-sense)\u8f14\u52a9\u51fd\uf969 ) , ( \u03bb \u03bb H \u70ba[12]\uff1a MPE ( ) ) | ( log ) | ( log ) , ( 1 , q O p q O p F z Z z q z MPE lattice z \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u2211 \u2211 = \u2208 = \u23a5 \u23a6 \u23a4 \u23a2 \u23a3 \u23a1 \u2202 \u2202 = W ( ) H MPE (10) \u5176\u4e2d \u03bb \u03bb \u03bb \u03bb = \u2202 \u2202 ) | ( log q O p F z MPE ( ) q z z avg \u7684\u503c\u53ef\u70ba\u6b63\u6216\u8ca0\uff0c\u53d6\u6c7a\u65bc\u8a5e\u5716\u4e0a\u901a\u904e\u6b64\u97f3\u7d20\u7684\u5019\u9078\u8a5e\u5e8f\uf99c\u7684\u671f\u671b\u6b63 \u78ba\uf961 c \u662f\u5426\u5927\u65bc\u8a5e\u5716\u4e0a\u6240\u6709\u5019\u9078\u8a5e\u5e8f\uf99c\u7684\u671f\u671b\u6b63\u78ba\uf961 c \u3002\u4e5f\u5c31\u662f\uff1a ( ) ( ) ( ) z avg z z q z MPE c q c q O p F \u2212 = \u2202 \u2202 = \u03b3 \u03bb \u03bb \u03bb \u03bb ) | ( log (11) \u5176\u4e2d\uff1a \u2211 \u2211 \u2208 \u2208 \u2208 = k lattice z i W k k z W q W k k z z q W P W O p W P W O p , ) ( ) | ( ) ( ) | ( : W W \u03bb \u03bb \u03b3 (12) lattice z k , \u70ba\u8a5e\u5716\u4e0a\u901a\u904e\u97f3\u7d20\u6bb5\uf918 q \u7684\u5019\u9078\u8a5e\u5e8f\uf99c\u7684\u4e8b\u5f8c\u6a5f\uf961\u548c\uff0c\u800c ( ) \u2211 \u2211 W W A W P W O p ) , ( ) ( ) | ( \u2208 \u2208 \u2208 \u2208 = k lattice z k i lattice z i W q W k k z W q W z i i i z z W P W O p q c : , , , ) ( ) | ( W W \u03bb \u03bb (13) \u70ba\u8a5e\u5716\u4e0a\u901a\u904e\u6b64\u97f3\u7d20\u6bb5\uf918\u7684\u5019\u9078\u8a5e\u5e8f\uf99c\u7684\u671f\u671b\u6b63\u78ba\uf961\uff0c\u800c \u2211 \u2211 \u2208 \u2208 = lattice z k lattice z i W k k z W z i i i z z avg W P W O p W W A W P W O p c , , ) ( ) | ( ) , ( ) ( ) | ( W W \u03bb \u03bb (14) \u70ba\u8a5e\u5716\u4e0a\u6240\u6709\u5019\u9078\u8a5e\u5e8f\uf99c\u7684\u671f\u671b\u6b63\u78ba\uf961\u3002 \u3001 z q \u03b3 ( ) q c z \u8207 \u7684\u7d71\u8a08\uf97e\u53ef\u5728\u8a5e\u5716\u4e0a\u4f7f\u7528\u6ce2 \u6c0f\u91cd\u4f30\u6f14\u7b97\u6cd5\uf92d\u6c42\u5f97[12]\u3002 z avg c ( ) q O p | log \u53e6 \u4e00 \u65b9 \u9762 \uff0c \u91dd \u5c0d \u5c0d \uf969 \u6a5f \uf961 \u51fd \uf969 \uff0c \u5fc5 \u9700 \u900f \u904e \u4e00 \u500b \u5f37 \u6027 \u8f14 \u52a9 \u51fd \uf969 z \u03bb ) , q Q ML , , ( z \u03bb \u03bb \uf92d\u4f30\u6e2c\u65b0\u7684\u6a21\u578b\uf96b\uf969\u503c\uff0c\u56e0\u6b64\u5f31\u6027\u8f14\u52a9\u51fd\uf969 ) , ( \u03bb \u03bb MPE H \u53ef\u8868\u793a\u6210\uff1a ( ) ) , , , ( q z ML \u03bb \u03bb ( ) ) | ( log ) , ( 1 , Q q O p F H Z z q z MPE MPE lattice z \u03bb \u03bb \u03bb \u03bb \u03bb \u03bb \u2211 \u2211 = \u2208 = \u23a5 \u23a6 \u23a4 \u23a2 \u23a3 \u23a1 \u2202 \u2202 = \u2032 W (15) \uf974\u4ee5 \uf92d\u8868\u793a MPE z q \u03b3 \u03bb \u03bb \u03bb = \u2202 ) | ( q O p z MPE \u2202 log F ) , , , ( q z Q \u53ef\u8868\u793a\u5982\u4e0b: ML \u03bb \u03bb \uff0c\u4e14 \u2211\u2211 = \u03a3 = q q s t m qm qm z z q ML t o N t q z Q ) , ); ( ( log ) ( ) , , , ( \u03bc \u03b3 \u03bb \u03bb e ) z t , ; ( qm qm N (16) \u5176\u4e2d \u70ba O \u7684\u7b2c \u500b\u8a9e\u97f3\u7279\u5fb5\u5411\uf97e\uff1b (t o z ) \u22c5 \u03bc \u03a3 \u662f\u97f3\u7d20\u6bb5\uf918 q \u7684\u7b2c \u500b\u9ad8\u65af\u5206\u5e03\uff0c m qm \u03bc \u8207 \u03a3 \u5206\u5225\u662f\u5b83\u7684\u5e73\u5747\u503c\u5411\uf97e\u8207\u5171\u8b8a\uf962\u77e9\u9663\u3002\u56e0\u6b64\u5f31\u6027\u8f14\u52a9\u51fd\uf969 ) MPE H , ( \u03bb \u03bb qm \u53ef\u9032\u4e00 \u6b65\u8868\u793a\u6210\uff1a ) , ); ( ( log ) ( ) , ( , qm qm z z qm MPE z q m s t q z MPE t o N t H q lattice z \u03a3 = \u2032 \u2211 \u2211 \u2211 \u2211 = \u2208 \u03bc \u03b3 \u03b3 \u03bb \u03bb W q s q e ) (t z qm \u03b3 ( ) t o z q m e q t = (17) \u5176\u4e2d \u8207 \u5206\u5225\u70ba\u97f3\u7d20\u6bb5\uf918 q \u7684\u958b\u59cb\u8207\u7d50\u675f\u6642\u9593\uff0c \u70ba\u8a9e\u97f3\u7279\u5fb5\u5411\uf97e \u5728\u97f3\u7d20\u6bb5 \uf918 \u4e0a \u7684 \u9ad8 \u65af \u5206 \u5e03 \u7684 \u4f54 \u6709 \u6a5f \uf961 \u3002 \uf974 \u628a \u5e73 \uf904 \u51fd \uf969 ) , ( \u03bb \u03bb SM H \u52a0 \u5165 \u5f31 \u6027 \u8f14 \u52a9 \u51fd \uf969 ) , ( \u03bb \u03bb MPE H \u2032 \uff0c\u5247 ) , ( \u03bb \u03bb MPE H \u2032 \u53ef\u9032\u4e00\u6b65\u8868\u793a\u6210[12]\uff1a [ ] ) ( ) ( ) ( |) log(| 2 ) , ), ( ( log ) ( ) , ( 1 1 , , \u2212 \u2212 = = \u2208 \u03a3 \u03a3 + \u2212 \u03a3 \u2212 + \u03a3 \u2212 \u03a3 = \u2032 \u2032 \u2211 \u2211 \u2211 \u2211",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "\u03bc \u03bc \u03bc \u03bc \u03bc \u03b3 \u03b3 \u03bb \u03bb W (18) \u800c\u5e73\uf904\u51fd\uf969 ) , ( \u03bb \u03bb SM H \u8868\u793a\u70ba\uff1a [ ] ) ( ) ( ) ( |) log(| 1 1 2 \u2212 \u2212 \u03a3 \u03a3 + \u2212 \u03a3 \u2212 +",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "D \u03bc \u03b3 \u03b3 \u03c3 \u2212 + \u2212 = MPE z q \u03b3 MPE z q \u03b3 \u2211 = \u2212 = s t z z q z qm den qmd lattice q t o t O 2 2 ) ( ) , 0 max( ) ( ) ( \u03b3 \u03b3 \u03b8 qmd D 2 2 2 2 ) ( )} ( ) ( { } { )} ( ) ( {",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "q q MPE t O 1 , ) , 0 max( ) ( ) ( W \u03b3 \u03b3 \u03b8 Z e q MPE",
                        "eq_num": "(21)"
                    }
                ],
                "section": "",
                "sec_num": null
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "(26) z t o ) ( ) z MPE t o ) ( ) z t o )",
                        "eq_num": "(22)(23)(24)(25)"
                    }
                ],
                "section": "",
                "sec_num": null
            },
            {
                "text": "-1 \u8207 1 \u4e4b\u9593)\uff0c \u5716 1 \u70ba \u8a08 \u7b97 \u539f \u59cb \u97f3 \u7d20 \u6b63 \u78ba \uf961 \u7684 \u4e00 \u500b \u7bc4 \uf9b5 \u3002 \u800c \u5247 \u53ef \u4ee5 \u7531 \u524d \u5411 \u5f8c \u5411 \u6f14 \u7b97 \u6cd5 (Forward-Backward Algorithm)\u6c42\u5f97[12]\u3002 ) (q A u q MPE z \u03b3 q b\u7684\u97f3\u7d20\u6b63\u78ba\uf961\u70ba\u53d6\u6700\u5927=1 a b c \u6b63\u78ba\u8f49\u8b6f \u97f3\u7d20\u5e8f\uf99c b time \u8fa8\uf9fc\u4e4b\u97f3\u7d20 -1+3/3=0 -1+3/3=0 -1+2*(3/3)=1 a b c b\u7684\u97f3\u7d20\u6b63\u78ba\uf961\u70ba\u53d6\u6700\u5927=1 \u6b63\u78ba\u8f49\u8b6f \u97f3\u7d20\u5e8f\uf99c b time \u8fa8\uf9fc\u4e4b\u97f3\u7d20 -1+3/3=0 -1+3/3=0 -1+2*(3/3)=1 \u5716 1",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "a \u6b63\u78ba\u8f49\u8b6f\u97f3\u7d20 c b a ( ) ( ) ( ) \u23a9 \u23a8 \u23a7 \u2260 = = , , ) , ( \u03c1 , t q if t q if t u q \u23ad \u23ac \u23ab < 1 e < 0 s \u2212 1 \u03c1 \u03b4 u u (30) \u5176\u4e2d q \u70ba\u8a5e\u5716\u4e2d\u67d0\u4e00\u97f3\u7d20\u6bb5\uf918\uff0c q \u548c q \u5206\u5225\u70ba\u97f3\u7d20\u6bb5\uf918 q \u7684\u958b\u59cb\u6642\u9593\u53ca\u7d50\u675f\u6642\u9593\uff0c ) (t \u70ba\u6b63\u78ba\u97f3\u7d20\u6bb5\uf918 u \u5728\u6642\u9593 t \u6642\u7684\u97f3\u7d20\u6a19\u8a18(Phone Label)\uff0c u \u03c1 \u70ba\u522a\u9664\u932f\u8aa4\u7684\u61f2\u7f70\u6b0a\u91cd (Deletion Penalty Weight)\uff0c\u7528\uf92d\u61f2\u7f70\u67d0\uf967\u5b8c\u5168\u6b63\u78ba\u97f3\u7d20\u6bb5\uf918 q \u7684\u6b63\u78ba\uf961\uff0c\u56e0\u6b64\u67d0\u4e00\u97f3\u7d20 \u6bb5\uf918\u5728\u67d0\u500b\u6642\u9593\u9ede t \u7684\u6b63\u78ba\uf961\u503c\u57df\u7bc4\u570d \u4ecb\u65bc \u70ba \u03c1 \u2212 \u5230 1",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "\u2211 \u2211 = \u2208 = Z z W i z i i z lattice z i W cc TimeFrameA O p W P W O p 1 , ) ( ) ( ) ( ) | ( W \u03bb \u2211 \u2211 = \u2208 = Z z i W z i MTFA lattice z i W cc TimeFrameA O W p F 1 , ) ( ) | ( ) ( W \u03bb (32) \u53e6\u5916\uff0c\u70ba\uf9ba\uf901\u5145\u5206\u5730\u61f2\u7f70\u522a\u9664\u932f\u8aa4\u4e14\u4f7f\u5176\u503c\u57df\u8207\u539f\u59cb\u97f3\u7d20\u6b63\u78ba\uf961\u540c\u70ba\u4ecb\u65bc-1 \u5230 1 \u4e4b\u9593\uff0c \u672c\uf941\u6587\u4f7f\u7528\uf9ba S \u578b\u51fd\uf969(Sigmoid",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "E I t o t O t E I t 1 1 , , ) ) ( ( ) ( ) , 0 max( ) ( ) ( ) ) ( ( ) , 0 max( ) ( W W \u03c1 \u03b3 \u03b3 \u03b8 \u03c1 \u03b3 \u03b3 \u03b3 (38) \u03b8 = \u2208 = z q s t lattice z q 1 , W \u03c1 \u70ba\u4e8b\u5148\u5b9a\u7fa9\u7684\u9580\u6abb\u503c(Threshold)\uff0c\u5176\u5024\u4ecb\u65bc 0 \u5230 1 \u4e4b\u9593\uff0c ) ) ( ( \u03c1 > t E I z \u23a9 \u23a8 \u23a7 \u2264 > = > \u03c1 \u03c1 \u03c1 ) ( , 0 ) ( , 1 ) ) ( ( t E",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "Pattern Classification",
                "authors": [
                    {
                        "first": "R",
                        "middle": [
                            "O"
                        ],
                        "last": "Duda",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [
                            "E"
                        ],
                        "last": "Hart",
                        "suffix": ""
                    },
                    {
                        "first": "D",
                        "middle": [
                            "G"
                        ],
                        "last": "Stork",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "R. O. Duda, P. E. Hart and D. G. Stork, Pattern Classification, Second Edition. New York: John & Wiley, 2000.",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "A Maximum Likelihood Approach to Continuous Speech Recognition",
                "authors": [
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Lalit",
                        "suffix": ""
                    },
                    {
                        "first": "F",
                        "middle": [],
                        "last": "Bahl",
                        "suffix": ""
                    },
                    {
                        "first": "Robert",
                        "middle": [
                            "L"
                        ],
                        "last": "Jelinek",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Mercer",
                        "suffix": ""
                    }
                ],
                "year": 1983,
                "venue": "IEEE Trans. Pattern Analysis and Machine Intelligence",
                "volume": "",
                "issue": "5",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Lalit R. Bahl, F. Jelinek and Robert L. Mercer, A Maximum Likelihood Approach to Continuous Speech Recognition, IEEE Trans. Pattern Analysis and Machine Intelligence, vol. PAMI-5, no.2, March 1983.",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "A Post-processing System to Yield Reduced Word Error Rates: Recognizer Output Voting Error Reduction (ROVER)",
                "authors": [
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Fiscus",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "Proc. ASRU",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "J. Fiscus, A Post-processing System to Yield Reduced Word Error Rates: Recognizer Output Voting Error Reduction (ROVER), in Proc. ASRU 1997.",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "Minimum Bayes-Risk Automatic Speech Recognition",
                "authors": [
                    {
                        "first": "V",
                        "middle": [],
                        "last": "Goel",
                        "suffix": ""
                    },
                    {
                        "first": "W",
                        "middle": [],
                        "last": "Byrne",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "Computer Speech and Language",
                "volume": "14",
                "issue": "",
                "pages": "115--135",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "V. Goel and W. Byrne, Minimum Bayes-Risk Automatic Speech Recognition, Computer Speech and Language, Vol. 14, pp.115-135, 2000.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "Explicit Word Error Minimization Using Word Hypothesis Posterior Probability",
                "authors": [
                    {
                        "first": "F",
                        "middle": [],
                        "last": "Wessel",
                        "suffix": ""
                    },
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Schluter",
                        "suffix": ""
                    },
                    {
                        "first": "K",
                        "middle": [],
                        "last": "Macherey",
                        "suffix": ""
                    },
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Ney",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Proc. ICASSP",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "F. Wessel, R. Schluter, K. Macherey and H. Ney, Explicit Word Error Minimization Using Word Hypothesis Posterior Probability, in Proc. ICASSP 2001.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Finding Consensus in Speech Recognition: Word Error Minimization and Other Applications of Confusion Networks",
                "authors": [
                    {
                        "first": "L",
                        "middle": [],
                        "last": "Mangu",
                        "suffix": ""
                    },
                    {
                        "first": "E",
                        "middle": [],
                        "last": "Brill",
                        "suffix": ""
                    },
                    {
                        "first": "A",
                        "middle": [],
                        "last": "Stolcke",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "Computer Speech and Language",
                "volume": "14",
                "issue": "",
                "pages": "373--400",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "L. Mangu, E. Brill and A. Stolcke, Finding Consensus in Speech Recognition: Word Error Minimization and Other Applications of Confusion Networks, Computer Speech and Language, Vol. 14, pp.373-400, 2000.",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "Hidden Markov Models, Maximum Mutual Information Estimation, and the Speech Recognition Problem",
                "authors": [
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Normandin",
                        "suffix": ""
                    }
                ],
                "year": 1991,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Y. Normandin, Hidden Markov Models, Maximum Mutual Information Estimation, and the Speech Recognition Problem, Ph.D Dissertation, McGill University, Montreal, 1991.",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "Overall Risk Criterion Estimation of Hidden Markov Model Parameters",
                "authors": [
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Kaiser",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Horvat",
                        "suffix": ""
                    },
                    {
                        "first": "Z",
                        "middle": [],
                        "last": "Kacic",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "Speech Communication",
                "volume": "38",
                "issue": "",
                "pages": "383--398",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "J. Kaiser, B. Horvat and Z. Kacic, Overall Risk Criterion Estimation of Hidden Markov Model Parameters, Speech Communication, Vol. 38, pp.383-398, 2002.",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "Lattice Segmentation and Minimum Bayes Risk Discriminative Training",
                "authors": [
                    {
                        "first": "V",
                        "middle": [],
                        "last": "Doumpiotis",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Tsakalidis",
                        "suffix": ""
                    },
                    {
                        "first": "W",
                        "middle": [],
                        "last": "Byrne",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Proc. Eurospeech",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "V. Doumpiotis, S. Tsakalidis and W. Byrne, Lattice Segmentation and Minimum Bayes Risk Discriminative Training, in Proc. Eurospeech 2004.",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "Minimum Phone Error and I-smoothing for Improved Discriminative Training",
                "authors": [
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Povey",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [
                            "C"
                        ],
                        "last": "Woodland",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "Proc. ICASSP",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "D. Povey and P. C. Woodland, Minimum Phone Error and I-smoothing for Improved Discriminative Training, in Proc. ICASSP 2002.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "A Word Graph Algorithm for Large Vocabulary Continuous Speech Recognition",
                "authors": [
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Ortmanns",
                        "suffix": ""
                    },
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Ney",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Aubert",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "Computer Speech and Language",
                "volume": "11",
                "issue": "",
                "pages": "11--72",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "S. Ortmanns, H. Ney and X Aubert, A Word Graph Algorithm for Large Vocabulary Continuous Speech Recognition, Computer Speech and Language, Vol. 11, pp.11-72, 1997.",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Discriminative Training for Large Vocabulary Speech Recognition. Ph.D Dissertation, Peterhouse",
                "authors": [
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Povey",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "D. Povey, Discriminative Training for Large Vocabulary Speech Recognition. Ph.D Dissertation, Peterhouse, University of Cambridge, July 2004.",
                "links": null
            },
            "BIBREF12": {
                "ref_id": "b12",
                "title": "Investigating Data Selection for Minimum Phone Error Training of Acoustic Models",
                "authors": [
                    {
                        "first": "Shih-Hung",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "Fang-Hui",
                        "middle": [],
                        "last": "Chu",
                        "suffix": ""
                    },
                    {
                        "first": "Shih-Hsiang",
                        "middle": [],
                        "last": "Lin",
                        "suffix": ""
                    },
                    {
                        "first": "Berlin",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2007,
                "venue": "Proc. ICME",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Shih-Hung Liu, Fang-Hui Chu, Shih-Hsiang Lin and Berlin Chen, Investigating Data Selection for Minimum Phone Error Training of Acoustic Models, in Proc. ICME 2007.",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "Minimum Exact Word Error Training",
                "authors": [
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Heigold",
                        "suffix": ""
                    }
                ],
                "year": null,
                "venue": "Proc. ASRU 2005",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "G. Heigold et al, Minimum Exact Word Error Training, in Proc. ASRU 2005",
                "links": null
            },
            "BIBREF14": {
                "ref_id": "b14",
                "title": "Advances in Large Margin Classifiers",
                "authors": [
                    {
                        "first": "A",
                        "middle": [
                            "J"
                        ],
                        "last": "Smola",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [],
                        "last": "Bartlett",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Scholkopf",
                        "suffix": ""
                    },
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Schuurmans",
                        "suffix": ""
                    }
                ],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "A. J. Smola, P. Bartlett, B. Scholkopf and D. Schuurmans, Advances in Large Margin Classifiers, The MIT Press.",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "The Nature of Statistical Learning Theory",
                "authors": [
                    {
                        "first": "V",
                        "middle": [],
                        "last": "Vapnik",
                        "suffix": ""
                    }
                ],
                "year": 1995,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "V. Vapnik, The Nature of Statistical Learning Theory, Springer-Verlag, New York, 1995.",
                "links": null
            },
            "BIBREF16": {
                "ref_id": "b16",
                "title": "Large Margin HMMs for Speech Recognition",
                "authors": [
                    {
                        "first": "Xinwei",
                        "middle": [],
                        "last": "Li",
                        "suffix": ""
                    },
                    {
                        "first": "Hui",
                        "middle": [],
                        "last": "Jiang",
                        "suffix": ""
                    },
                    {
                        "first": "Chaojun",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "Proc. ICASSP",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Xinwei Li, Hui Jiang and Chaojun Liu, Large Margin HMMs for Speech Recognition, in Proc. ICASSP 2005.",
                "links": null
            },
            "BIBREF17": {
                "ref_id": "b17",
                "title": "Soft Margin Estimation of Hidden Markov Model Parameters",
                "authors": [
                    {
                        "first": "Jinyu",
                        "middle": [],
                        "last": "Li",
                        "suffix": ""
                    },
                    {
                        "first": "Ming",
                        "middle": [],
                        "last": "Yuan",
                        "suffix": ""
                    },
                    {
                        "first": "Chin-Hui",
                        "middle": [],
                        "last": "Lee",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "Proc. ICSLP",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Jinyu Li, Ming Yuan and Chin-Hui Lee, Soft Margin Estimation of Hidden Markov Model Parameters, in Proc. ICSLP 2006.",
                "links": null
            },
            "BIBREF18": {
                "ref_id": "b18",
                "title": "Lightly Supervised and Data-Driven Approaches to Mandarin Broadcast News Transcription",
                "authors": [
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [
                            "W"
                        ],
                        "last": "Kuo",
                        "suffix": ""
                    },
                    {
                        "first": "W",
                        "middle": [
                            "H"
                        ],
                        "last": "Tsai",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Proc. ICASSP",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "B. Chen, J. W. Kuo and W. H. Tsai, Lightly Supervised and Data-Driven Approaches to Mandarin Broadcast News Transcription, in Proc. ICASSP 2004.",
                "links": null
            },
            "BIBREF19": {
                "ref_id": "b19",
                "title": "Investigation of Silicon-Auditory Models and Generalizaion of Linar Discriminant Analysis for Improved Speech Recognition",
                "authors": [
                    {
                        "first": "N",
                        "middle": [],
                        "last": "Kumar",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "N. Kumar, Investigation of Silicon-Auditory Models and Generalizaion of Linar Discriminant Analysis for Improved Speech Recognition, Ph.D. Thesis, John Hopkins University, Baltimore, 1997.",
                "links": null
            },
            "BIBREF20": {
                "ref_id": "b20",
                "title": "Maximum Likelihood Modeling with Gaussian Distributions",
                "authors": [
                    {
                        "first": "R",
                        "middle": [
                            "A"
                        ],
                        "last": "Gopinath",
                        "suffix": ""
                    }
                ],
                "year": 1998,
                "venue": "Proc. of ICASSP",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "R. A. Gopinath, Maximum Likelihood Modeling with Gaussian Distributions, in Proc. of ICASSP 1998.",
                "links": null
            },
            "BIBREF21": {
                "ref_id": "b21",
                "title": "LDC: Linguistic Data Consortium",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "LDC: Linguistic Data Consortium, http://www.ldc.upenn.edu",
                "links": null
            },
            "BIBREF22": {
                "ref_id": "b22",
                "title": "Estimation of Probabilities from Sparse Data for Other Language Component of a Speech Recognizer",
                "authors": [
                    {
                        "first": "S",
                        "middle": [
                            "M"
                        ],
                        "last": "Katz",
                        "suffix": ""
                    }
                ],
                "year": 1987,
                "venue": "IEEE Trans. Acoustics, Speech and Signal Processing",
                "volume": "35",
                "issue": "3",
                "pages": "400--401",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "S. M. Katz, Estimation of Probabilities from Sparse Data for Other Language Component of a Speech Recognizer, IEEE Trans. Acoustics, Speech and Signal Processing, Vol. 35, No.3, pp. 400-401, 1987.",
                "links": null
            },
            "BIBREF23": {
                "ref_id": "b23",
                "title": "SRI language Modeling Toolkit",
                "authors": [
                    {
                        "first": "A",
                        "middle": [],
                        "last": "Stolcke",
                        "suffix": ""
                    }
                ],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "A. Stolcke, SRI language Modeling Toolkit, version 1.3.3, http://www.speech.sri.com/projects/srilm/",
                "links": null
            },
            "BIBREF24": {
                "ref_id": "b24",
                "title": "MATBN: A Mandarin Chinese Broadcast News Corpus",
                "authors": [
                    {
                        "first": "H",
                        "middle": [
                            "M"
                        ],
                        "last": "Wang",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "J.-W",
                        "middle": [],
                        "last": "Kuo",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [
                            "S"
                        ],
                        "last": "Cheng",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "International Journal of Computational Linguistics and Chinese Language Processing",
                "volume": "10",
                "issue": "2",
                "pages": "219--236",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "H. M. Wang, B. Chen, J.-W. Kuo, and S.S. Cheng. MATBN: A Mandarin Chinese Broadcast News Corpus, International Journal of Computational Linguistics and Chinese Language Processing, Vol. 10, No. 2, pp. 219-236, 2005.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF0": {
                "text": "\u5176\u4e2d\u7d71\u8a08\u503c\u8cc7\u8a0a\u53ef\u5206\u70ba\uf978\uf9d0\uff0c\u4ea6\u5373 num(numerator)\u8207 den(denominator)\uf978\uf9d0\uff0cnum \u4ee3\u8868 \u70ba\u6b63\u6642\u7684\u7d71\u8a08\u503c\u8cc7\u8a0a\uff0c\u800c den \u5247\u4ee3\u8868 \u70ba\u8ca0\u6642\u7684\u7d71\u8a08\u503c\u8cc7\u8a0a\uff0c\u8a73\u7d30\u7d71\u8a08\u8cc7\u8a0a\u53ef \u5206\u5225\u8868\u793a\u5982\u4e0b\uff1a",
                "num": null,
                "type_str": "figure",
                "uris": null
            },
            "FIGREF1": {
                "text": "Function)\uf92d\u6b63\u898f\u5316\u6642\u9593\u97f3\u6846\u97f3\u7d20\u6b63\u78ba\uf961\u51fd\uf969(\u5f0f(29))\u7684 \u5206\u5b50\u9805\uff0c\u7a31\u4e4b\u70ba S \u578b\u6642\u9593\u97f3\u6846\u97f3\u7d20\u6b63\u78ba\uf961\u51fd\uf969(Sigmoid Time Frame Phone Accuracy, \u8a18 \u4f5c STFA):",
                "num": null,
                "type_str": "figure",
                "uris": null
            }
        }
    }
}