A Tutorial on Support Vector Machines for Pattern Recognition View Full Text


Ontology type: schema:ScholarlyArticle     


Article Info

DATE

1998-06

AUTHORS

Christopher J.C. Burges

ABSTRACT

The tutorial starts with an overview of the concepts of VC dimension and structural risk minimization. We then describe linear Support Vector Machines (SVMs) for separable and non-separable data, working through a non-trivial example in detail. We describe a mechanical analogy, and discuss when SVM solutions are unique and when they are global. We describe how support vector training can be practically implemented, and discuss in detail the kernel mapping technique which is used to construct SVM solutions which are nonlinear in the data. We show how Support Vector machines can have very large (even infinite) VC dimension by computing the VC dimension for homogeneous polynomial and Gaussian radial basis function kernels. While very high VC dimension would normally bode ill for generalization performance, and while at present there exists no theory which shows that good generalization performance is guaranteed for SVMs, there are several arguments which support the observed high accuracy of SVMs, which we review. Results of some experiments which were inspired by these arguments are also presented. We give numerous examples and proofs of most of the key theorems. There is new material, and I hope that the reader will find that even old material is cast in a fresh light. More... »

PAGES

121-167

References to SciGraph publications

Identifiers

URI

http://scigraph.springernature.com/pub.10.1023/a:1009715923555

DOI

http://dx.doi.org/10.1023/a:1009715923555

DIMENSIONS

https://app.dimensions.ai/details/publication/pub.1042048349


Indexing Status Check whether this publication has been indexed by Scopus and Web Of Science using the SN Indexing Status Tool
Incoming Citations Browse incoming citations for this publication using opencitations.net

JSON-LD is the canonical representation for SciGraph data.

TIP: You can open this SciGraph record using an external JSON-LD service: JSON-LD Playground Google SDTT

[
  {
    "@context": "https://springernature.github.io/scigraph/jsonld/sgcontext.json", 
    "about": [
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/0806", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Information Systems", 
        "type": "DefinedTerm"
      }, 
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/08", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Information and Computing Sciences", 
        "type": "DefinedTerm"
      }
    ], 
    "author": [
      {
        "affiliation": {
          "name": [
            "Bell Laboratories, Lucent Technologies, USA."
          ], 
          "type": "Organization"
        }, 
        "familyName": "Burges", 
        "givenName": "Christopher J.C.", 
        "id": "sg:person.015522010763.41", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.015522010763.41"
        ], 
        "type": "Person"
      }
    ], 
    "citation": [
      {
        "id": "https://doi.org/10.1162/neco.1992.4.1.1", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1007421220"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/3-540-61510-5_45", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1009128808", 
          "https://doi.org/10.1007/3-540-61510-5_45"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/3-540-61510-5_12", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1015690232", 
          "https://doi.org/10.1007/3-540-61510-5_12"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1162/089976698300017269", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1016907700"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1162/089976698300017467", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1019671707"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1016/0024-3795(80)90172-x", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1023079072"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/bf00994018", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1025150743", 
          "https://doi.org/10.1007/bf00994018"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1090/s0025-5718-1977-0428694-0", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1039637161"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/pl00013831", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1047784398", 
          "https://doi.org/10.1007/pl00013831"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1109/78.650102", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1061229999"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1137/0801008", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1062854121"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1287/ijoc.6.1.32", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1064707482"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1109/cvpr.1997.609310", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1093529535"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1109/nnsp.1997.622433", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1093652497"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1109/nnsp.1997.622408", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1095767720"
        ], 
        "type": "CreativeWork"
      }
    ], 
    "datePublished": "1998-06", 
    "datePublishedReg": "1998-06-01", 
    "description": "The tutorial starts with an overview of the concepts of VC dimension and structural risk minimization. We then describe linear Support Vector Machines (SVMs) for separable and non-separable data, working through a non-trivial example in detail. We describe a mechanical analogy, and discuss when SVM solutions are unique and when they are global. We describe how support vector training can be practically implemented, and discuss in detail the kernel mapping technique which is used to construct SVM solutions which are nonlinear in the data. We show how Support Vector machines can have very large (even infinite) VC dimension by computing the VC dimension for homogeneous polynomial and Gaussian radial basis function kernels. While very high VC dimension would normally bode ill for generalization performance, and while at present there exists no theory which shows that good generalization performance is guaranteed for SVMs, there are several arguments which support the observed high accuracy of SVMs, which we review. Results of some experiments which were inspired by these arguments are also presented. We give numerous examples and proofs of most of the key theorems. There is new material, and I hope that the reader will find that even old material is cast in a fresh light.", 
    "genre": "research_article", 
    "id": "sg:pub.10.1023/a:1009715923555", 
    "inLanguage": [
      "en"
    ], 
    "isAccessibleForFree": false, 
    "isPartOf": [
      {
        "id": "sg:journal.1041853", 
        "issn": [
          "1384-5810", 
          "1573-756X"
        ], 
        "name": "Data Mining and Knowledge Discovery", 
        "type": "Periodical"
      }, 
      {
        "issueNumber": "2", 
        "type": "PublicationIssue"
      }, 
      {
        "type": "PublicationVolume", 
        "volumeNumber": "2"
      }
    ], 
    "name": "A Tutorial on Support Vector Machines for Pattern Recognition", 
    "pagination": "121-167", 
    "productId": [
      {
        "name": "readcube_id", 
        "type": "PropertyValue", 
        "value": [
          "c84c46cc5ee29f6f90b0e4f5b4e2bf684f8cdd944b947a9108dbe171a6ae2ebc"
        ]
      }, 
      {
        "name": "doi", 
        "type": "PropertyValue", 
        "value": [
          "10.1023/a:1009715923555"
        ]
      }, 
      {
        "name": "dimensions_id", 
        "type": "PropertyValue", 
        "value": [
          "pub.1042048349"
        ]
      }
    ], 
    "sameAs": [
      "https://doi.org/10.1023/a:1009715923555", 
      "https://app.dimensions.ai/details/publication/pub.1042048349"
    ], 
    "sdDataset": "articles", 
    "sdDatePublished": "2019-04-11T00:21", 
    "sdLicense": "https://scigraph.springernature.com/explorer/license/", 
    "sdPublisher": {
      "name": "Springer Nature - SN SciGraph project", 
      "type": "Organization"
    }, 
    "sdSource": "s3://com-uberresearch-data-dimensions-target-20181106-alternative/cleanup/v134/2549eaecd7973599484d7c17b260dba0a4ecb94b/merge/v9/a6c9fde33151104705d4d7ff012ea9563521a3ce/jats-lookup/v90/0000000001_0000000264/records_8695_00000537.jsonl", 
    "type": "ScholarlyArticle", 
    "url": "http://link.springer.com/10.1023%2FA%3A1009715923555"
  }
]
 

Download the RDF metadata as:  json-ld nt turtle xml License info

HOW TO GET THIS DATA PROGRAMMATICALLY:

JSON-LD is a popular format for linked data which is fully compatible with JSON.

curl -H 'Accept: application/ld+json' 'https://scigraph.springernature.com/pub.10.1023/a:1009715923555'

N-Triples is a line-based linked data format ideal for batch operations.

curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/pub.10.1023/a:1009715923555'

Turtle is a human-readable linked data format.

curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/pub.10.1023/a:1009715923555'

RDF/XML is a standard XML format for linked data.

curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/pub.10.1023/a:1009715923555'


 

This table displays all metadata directly associated to this object as RDF triples.

109 TRIPLES      21 PREDICATES      42 URIs      19 LITERALS      7 BLANK NODES

Subject Predicate Object
1 sg:pub.10.1023/a:1009715923555 schema:about anzsrc-for:08
2 anzsrc-for:0806
3 schema:author N7ed9c5e099204736a4ab6db37a16ffd9
4 schema:citation sg:pub.10.1007/3-540-61510-5_12
5 sg:pub.10.1007/3-540-61510-5_45
6 sg:pub.10.1007/bf00994018
7 sg:pub.10.1007/pl00013831
8 https://doi.org/10.1016/0024-3795(80)90172-x
9 https://doi.org/10.1090/s0025-5718-1977-0428694-0
10 https://doi.org/10.1109/78.650102
11 https://doi.org/10.1109/cvpr.1997.609310
12 https://doi.org/10.1109/nnsp.1997.622408
13 https://doi.org/10.1109/nnsp.1997.622433
14 https://doi.org/10.1137/0801008
15 https://doi.org/10.1162/089976698300017269
16 https://doi.org/10.1162/089976698300017467
17 https://doi.org/10.1162/neco.1992.4.1.1
18 https://doi.org/10.1287/ijoc.6.1.32
19 schema:datePublished 1998-06
20 schema:datePublishedReg 1998-06-01
21 schema:description The tutorial starts with an overview of the concepts of VC dimension and structural risk minimization. We then describe linear Support Vector Machines (SVMs) for separable and non-separable data, working through a non-trivial example in detail. We describe a mechanical analogy, and discuss when SVM solutions are unique and when they are global. We describe how support vector training can be practically implemented, and discuss in detail the kernel mapping technique which is used to construct SVM solutions which are nonlinear in the data. We show how Support Vector machines can have very large (even infinite) VC dimension by computing the VC dimension for homogeneous polynomial and Gaussian radial basis function kernels. While very high VC dimension would normally bode ill for generalization performance, and while at present there exists no theory which shows that good generalization performance is guaranteed for SVMs, there are several arguments which support the observed high accuracy of SVMs, which we review. Results of some experiments which were inspired by these arguments are also presented. We give numerous examples and proofs of most of the key theorems. There is new material, and I hope that the reader will find that even old material is cast in a fresh light.
22 schema:genre research_article
23 schema:inLanguage en
24 schema:isAccessibleForFree false
25 schema:isPartOf N78577427ffa34b6db098b8ef97c5b722
26 N8b9141eb50ed46a282de280f3af02d64
27 sg:journal.1041853
28 schema:name A Tutorial on Support Vector Machines for Pattern Recognition
29 schema:pagination 121-167
30 schema:productId N604d7eca1af24ecaaf9e2262b41bfcd4
31 Nbe6fa01f59d94e7eaa9bf274a811a74e
32 Nf46488ba6a3844bbb324ce6bee4643ef
33 schema:sameAs https://app.dimensions.ai/details/publication/pub.1042048349
34 https://doi.org/10.1023/a:1009715923555
35 schema:sdDatePublished 2019-04-11T00:21
36 schema:sdLicense https://scigraph.springernature.com/explorer/license/
37 schema:sdPublisher Na89170367d644b9abe1af3acacba9dd5
38 schema:url http://link.springer.com/10.1023%2FA%3A1009715923555
39 sgo:license sg:explorer/license/
40 sgo:sdDataset articles
41 rdf:type schema:ScholarlyArticle
42 N604d7eca1af24ecaaf9e2262b41bfcd4 schema:name readcube_id
43 schema:value c84c46cc5ee29f6f90b0e4f5b4e2bf684f8cdd944b947a9108dbe171a6ae2ebc
44 rdf:type schema:PropertyValue
45 N78577427ffa34b6db098b8ef97c5b722 schema:issueNumber 2
46 rdf:type schema:PublicationIssue
47 N7ed9c5e099204736a4ab6db37a16ffd9 rdf:first sg:person.015522010763.41
48 rdf:rest rdf:nil
49 N8b9141eb50ed46a282de280f3af02d64 schema:volumeNumber 2
50 rdf:type schema:PublicationVolume
51 Na89170367d644b9abe1af3acacba9dd5 schema:name Springer Nature - SN SciGraph project
52 rdf:type schema:Organization
53 Nb3d9af65b3644d0a8293da669bcd8ad1 schema:name Bell Laboratories, Lucent Technologies, USA.
54 rdf:type schema:Organization
55 Nbe6fa01f59d94e7eaa9bf274a811a74e schema:name doi
56 schema:value 10.1023/a:1009715923555
57 rdf:type schema:PropertyValue
58 Nf46488ba6a3844bbb324ce6bee4643ef schema:name dimensions_id
59 schema:value pub.1042048349
60 rdf:type schema:PropertyValue
61 anzsrc-for:08 schema:inDefinedTermSet anzsrc-for:
62 schema:name Information and Computing Sciences
63 rdf:type schema:DefinedTerm
64 anzsrc-for:0806 schema:inDefinedTermSet anzsrc-for:
65 schema:name Information Systems
66 rdf:type schema:DefinedTerm
67 sg:journal.1041853 schema:issn 1384-5810
68 1573-756X
69 schema:name Data Mining and Knowledge Discovery
70 rdf:type schema:Periodical
71 sg:person.015522010763.41 schema:affiliation Nb3d9af65b3644d0a8293da669bcd8ad1
72 schema:familyName Burges
73 schema:givenName Christopher J.C.
74 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.015522010763.41
75 rdf:type schema:Person
76 sg:pub.10.1007/3-540-61510-5_12 schema:sameAs https://app.dimensions.ai/details/publication/pub.1015690232
77 https://doi.org/10.1007/3-540-61510-5_12
78 rdf:type schema:CreativeWork
79 sg:pub.10.1007/3-540-61510-5_45 schema:sameAs https://app.dimensions.ai/details/publication/pub.1009128808
80 https://doi.org/10.1007/3-540-61510-5_45
81 rdf:type schema:CreativeWork
82 sg:pub.10.1007/bf00994018 schema:sameAs https://app.dimensions.ai/details/publication/pub.1025150743
83 https://doi.org/10.1007/bf00994018
84 rdf:type schema:CreativeWork
85 sg:pub.10.1007/pl00013831 schema:sameAs https://app.dimensions.ai/details/publication/pub.1047784398
86 https://doi.org/10.1007/pl00013831
87 rdf:type schema:CreativeWork
88 https://doi.org/10.1016/0024-3795(80)90172-x schema:sameAs https://app.dimensions.ai/details/publication/pub.1023079072
89 rdf:type schema:CreativeWork
90 https://doi.org/10.1090/s0025-5718-1977-0428694-0 schema:sameAs https://app.dimensions.ai/details/publication/pub.1039637161
91 rdf:type schema:CreativeWork
92 https://doi.org/10.1109/78.650102 schema:sameAs https://app.dimensions.ai/details/publication/pub.1061229999
93 rdf:type schema:CreativeWork
94 https://doi.org/10.1109/cvpr.1997.609310 schema:sameAs https://app.dimensions.ai/details/publication/pub.1093529535
95 rdf:type schema:CreativeWork
96 https://doi.org/10.1109/nnsp.1997.622408 schema:sameAs https://app.dimensions.ai/details/publication/pub.1095767720
97 rdf:type schema:CreativeWork
98 https://doi.org/10.1109/nnsp.1997.622433 schema:sameAs https://app.dimensions.ai/details/publication/pub.1093652497
99 rdf:type schema:CreativeWork
100 https://doi.org/10.1137/0801008 schema:sameAs https://app.dimensions.ai/details/publication/pub.1062854121
101 rdf:type schema:CreativeWork
102 https://doi.org/10.1162/089976698300017269 schema:sameAs https://app.dimensions.ai/details/publication/pub.1016907700
103 rdf:type schema:CreativeWork
104 https://doi.org/10.1162/089976698300017467 schema:sameAs https://app.dimensions.ai/details/publication/pub.1019671707
105 rdf:type schema:CreativeWork
106 https://doi.org/10.1162/neco.1992.4.1.1 schema:sameAs https://app.dimensions.ai/details/publication/pub.1007421220
107 rdf:type schema:CreativeWork
108 https://doi.org/10.1287/ijoc.6.1.32 schema:sameAs https://app.dimensions.ai/details/publication/pub.1064707482
109 rdf:type schema:CreativeWork
 




Preview window. Press ESC to close (or click here)


...