Active zero-shot learning: a novel approach to extreme multi-labeled classification View Full Text


Ontology type: schema:ScholarlyArticle      Open Access: True


Article Info

DATE

2017-02-11

AUTHORS

Sihong Xie, Philip S. Yu

ABSTRACT

Big data bring a huge volume of data in a great speed and in many formats with extremely many labels and concepts to be modeled and predicted, such as in text and image tagging, online advertisement placement, recommendation systems, NLP. This emerging issue of big data is termed “extreme multi-labeled classification” (XMLC) and is challenging due to the time, space and sample complexity in predictive model training and testing. We first define general XMLC and then categorize and review recent methods based on two specific forms of XMLC. We propose a novel method called active zero-shot learning to reduce the above complexities. Since the performance of the unseen class prediction largely depends on the seen classes that have labeled data, we challenge the critical and yet often overlooked assumption that the labeled data can only be passively acquired. We propose a new learning paradigm aiming at accurate predictions of a large number of unseen labels using labeled data from only an intelligently selected small set of seed classes with the help of external knowledge. We further demonstrate that the proposed strategy has desirable probabilistic properties to facilitate unseen classes prediction. Experiments on 4 datasets demonstrate that the proposed algorithm is superior to a wide spectrum of baselines. Based on our findings, we point out several critical and promising future directions in XMLC. More... »

PAGES

151-160

References to SciGraph publications

Identifiers

URI

http://scigraph.springernature.com/pub.10.1007/s41060-017-0042-5

DOI

http://dx.doi.org/10.1007/s41060-017-0042-5

DIMENSIONS

https://app.dimensions.ai/details/publication/pub.1083766144


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/2008/08", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Information and Computing Sciences", 
        "type": "DefinedTerm"
      }, 
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/0801", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Artificial Intelligence and Image Processing", 
        "type": "DefinedTerm"
      }
    ], 
    "author": [
      {
        "affiliation": {
          "alternateName": "Department of Computer Science and Engineering, Lehigh University, Bethlehem, PA, USA", 
          "id": "http://www.grid.ac/institutes/grid.259029.5", 
          "name": [
            "Department of Computer Science and Engineering, Lehigh University, Bethlehem, PA, USA"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Xie", 
        "givenName": "Sihong", 
        "id": "sg:person.014402567553.41", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.014402567553.41"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "Department of Computer Science, University of Illinois at Chicago, Chicago, IL, USA", 
          "id": "http://www.grid.ac/institutes/grid.185648.6", 
          "name": [
            "Department of Computer Science, University of Illinois at Chicago, Chicago, IL, USA"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Yu", 
        "givenName": "Philip S.", 
        "id": "sg:person.011016356115.95", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.011016356115.95"
        ], 
        "type": "Person"
      }
    ], 
    "citation": [
      {
        "id": "sg:pub.10.1007/978-3-642-15555-0_6", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1017308440", 
          "https://doi.org/10.1007/978-3-642-15555-0_6"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-319-10605-2_38", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1044946925", 
          "https://doi.org/10.1007/978-3-319-10605-2_38"
        ], 
        "type": "CreativeWork"
      }
    ], 
    "datePublished": "2017-02-11", 
    "datePublishedReg": "2017-02-11", 
    "description": "Big data bring a huge volume of data in a great speed and in many formats with extremely many labels and concepts to be modeled and predicted, such as in text and image tagging, online advertisement placement, recommendation systems, NLP. This emerging issue of big data is termed \u201cextreme multi-labeled classification\u201d (XMLC) and is challenging due to the time, space and sample complexity in predictive model training and testing. We first define general XMLC and then categorize and review recent methods based on two specific forms of XMLC. We propose a novel method called active zero-shot learning to reduce the above complexities. Since the performance of the unseen class prediction largely depends on the seen classes that have labeled data, we challenge the critical and yet often overlooked assumption that the labeled data can only be passively acquired. We propose a new learning paradigm aiming at accurate predictions of a large number of unseen labels using labeled data from only an intelligently selected small set of seed classes with the help of external knowledge. We further demonstrate that the proposed strategy has desirable probabilistic properties to facilitate unseen classes prediction. Experiments on 4 datasets demonstrate that the proposed algorithm is superior to a wide spectrum of baselines. Based on our findings, we point out several critical and promising future directions in XMLC.", 
    "genre": "article", 
    "id": "sg:pub.10.1007/s41060-017-0042-5", 
    "isAccessibleForFree": true, 
    "isFundedItemOf": [
      {
        "id": "sg:grant.4314437", 
        "type": "MonetaryGrant"
      }
    ], 
    "isPartOf": [
      {
        "id": "sg:journal.1156617", 
        "issn": [
          "2364-415X", 
          "2364-4168"
        ], 
        "name": "International Journal of Data Science and Analytics", 
        "publisher": "Springer Nature", 
        "type": "Periodical"
      }, 
      {
        "issueNumber": "3", 
        "type": "PublicationIssue"
      }, 
      {
        "type": "PublicationVolume", 
        "volumeNumber": "3"
      }
    ], 
    "keywords": [
      "multi-labeled classification", 
      "zero-shot learning", 
      "big data", 
      "class prediction", 
      "predictive model training", 
      "new learning paradigm", 
      "image tagging", 
      "unseen labels", 
      "recommendation system", 
      "huge volume", 
      "model training", 
      "above complexities", 
      "advertisement placement", 
      "learning paradigm", 
      "promising future directions", 
      "small set", 
      "XMLC", 
      "external knowledge", 
      "recent methods", 
      "novel approach", 
      "sample complexity", 
      "learning", 
      "greater speed", 
      "complexity", 
      "novel method", 
      "classification", 
      "probabilistic properties", 
      "labels", 
      "accurate prediction", 
      "large number", 
      "NLP", 
      "dataset", 
      "algorithm", 
      "format", 
      "tagging", 
      "future directions", 
      "prediction", 
      "data", 
      "paradigm", 
      "set", 
      "text", 
      "method", 
      "performance", 
      "system", 
      "class", 
      "issues", 
      "concept", 
      "help", 
      "training", 
      "speed", 
      "space", 
      "wide spectrum", 
      "knowledge", 
      "experiments", 
      "number", 
      "time", 
      "strategies", 
      "seed class", 
      "testing", 
      "assumption", 
      "placement", 
      "specific form", 
      "direction", 
      "form", 
      "volume", 
      "baseline", 
      "properties", 
      "spectra", 
      "findings", 
      "approach"
    ], 
    "name": "Active zero-shot learning: a novel approach to extreme multi-labeled classification", 
    "pagination": "151-160", 
    "productId": [
      {
        "name": "dimensions_id", 
        "type": "PropertyValue", 
        "value": [
          "pub.1083766144"
        ]
      }, 
      {
        "name": "doi", 
        "type": "PropertyValue", 
        "value": [
          "10.1007/s41060-017-0042-5"
        ]
      }
    ], 
    "sameAs": [
      "https://doi.org/10.1007/s41060-017-0042-5", 
      "https://app.dimensions.ai/details/publication/pub.1083766144"
    ], 
    "sdDataset": "articles", 
    "sdDatePublished": "2022-08-04T17:05", 
    "sdLicense": "https://scigraph.springernature.com/explorer/license/", 
    "sdPublisher": {
      "name": "Springer Nature - SN SciGraph project", 
      "type": "Organization"
    }, 
    "sdSource": "s3://com-springernature-scigraph/baseset/20220804/entities/gbq_results/article/article_721.jsonl", 
    "type": "ScholarlyArticle", 
    "url": "https://doi.org/10.1007/s41060-017-0042-5"
  }
]
 

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.10.1007/s41060-017-0042-5'

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

curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/pub.10.1007/s41060-017-0042-5'

Turtle is a human-readable linked data format.

curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/pub.10.1007/s41060-017-0042-5'

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

curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/pub.10.1007/s41060-017-0042-5'


 

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

147 TRIPLES      21 PREDICATES      96 URIs      86 LITERALS      6 BLANK NODES

Subject Predicate Object
1 sg:pub.10.1007/s41060-017-0042-5 schema:about anzsrc-for:08
2 anzsrc-for:0801
3 schema:author N20db0c1c13994a7a9f6fa8b81ae13c5c
4 schema:citation sg:pub.10.1007/978-3-319-10605-2_38
5 sg:pub.10.1007/978-3-642-15555-0_6
6 schema:datePublished 2017-02-11
7 schema:datePublishedReg 2017-02-11
8 schema:description Big data bring a huge volume of data in a great speed and in many formats with extremely many labels and concepts to be modeled and predicted, such as in text and image tagging, online advertisement placement, recommendation systems, NLP. This emerging issue of big data is termed “extreme multi-labeled classification” (XMLC) and is challenging due to the time, space and sample complexity in predictive model training and testing. We first define general XMLC and then categorize and review recent methods based on two specific forms of XMLC. We propose a novel method called active zero-shot learning to reduce the above complexities. Since the performance of the unseen class prediction largely depends on the seen classes that have labeled data, we challenge the critical and yet often overlooked assumption that the labeled data can only be passively acquired. We propose a new learning paradigm aiming at accurate predictions of a large number of unseen labels using labeled data from only an intelligently selected small set of seed classes with the help of external knowledge. We further demonstrate that the proposed strategy has desirable probabilistic properties to facilitate unseen classes prediction. Experiments on 4 datasets demonstrate that the proposed algorithm is superior to a wide spectrum of baselines. Based on our findings, we point out several critical and promising future directions in XMLC.
9 schema:genre article
10 schema:isAccessibleForFree true
11 schema:isPartOf N0f2dd98cd6ae481bb1b9895f142c3ae0
12 N5679d0d8035a404493a4c14479bbc66d
13 sg:journal.1156617
14 schema:keywords NLP
15 XMLC
16 above complexities
17 accurate prediction
18 advertisement placement
19 algorithm
20 approach
21 assumption
22 baseline
23 big data
24 class
25 class prediction
26 classification
27 complexity
28 concept
29 data
30 dataset
31 direction
32 experiments
33 external knowledge
34 findings
35 form
36 format
37 future directions
38 greater speed
39 help
40 huge volume
41 image tagging
42 issues
43 knowledge
44 labels
45 large number
46 learning
47 learning paradigm
48 method
49 model training
50 multi-labeled classification
51 new learning paradigm
52 novel approach
53 novel method
54 number
55 paradigm
56 performance
57 placement
58 prediction
59 predictive model training
60 probabilistic properties
61 promising future directions
62 properties
63 recent methods
64 recommendation system
65 sample complexity
66 seed class
67 set
68 small set
69 space
70 specific form
71 spectra
72 speed
73 strategies
74 system
75 tagging
76 testing
77 text
78 time
79 training
80 unseen labels
81 volume
82 wide spectrum
83 zero-shot learning
84 schema:name Active zero-shot learning: a novel approach to extreme multi-labeled classification
85 schema:pagination 151-160
86 schema:productId N38035089bd364dbb9f3cc6e1c0eb3f65
87 Nd54dcafe58e742c59d0f35897fb4e8ff
88 schema:sameAs https://app.dimensions.ai/details/publication/pub.1083766144
89 https://doi.org/10.1007/s41060-017-0042-5
90 schema:sdDatePublished 2022-08-04T17:05
91 schema:sdLicense https://scigraph.springernature.com/explorer/license/
92 schema:sdPublisher N81a9c2e483fc4871ac25ed27dde1ad4e
93 schema:url https://doi.org/10.1007/s41060-017-0042-5
94 sgo:license sg:explorer/license/
95 sgo:sdDataset articles
96 rdf:type schema:ScholarlyArticle
97 N0f2dd98cd6ae481bb1b9895f142c3ae0 schema:volumeNumber 3
98 rdf:type schema:PublicationVolume
99 N12ea9c7a99834fcf805cc437a7098b48 rdf:first sg:person.011016356115.95
100 rdf:rest rdf:nil
101 N20db0c1c13994a7a9f6fa8b81ae13c5c rdf:first sg:person.014402567553.41
102 rdf:rest N12ea9c7a99834fcf805cc437a7098b48
103 N38035089bd364dbb9f3cc6e1c0eb3f65 schema:name dimensions_id
104 schema:value pub.1083766144
105 rdf:type schema:PropertyValue
106 N5679d0d8035a404493a4c14479bbc66d schema:issueNumber 3
107 rdf:type schema:PublicationIssue
108 N81a9c2e483fc4871ac25ed27dde1ad4e schema:name Springer Nature - SN SciGraph project
109 rdf:type schema:Organization
110 Nd54dcafe58e742c59d0f35897fb4e8ff schema:name doi
111 schema:value 10.1007/s41060-017-0042-5
112 rdf:type schema:PropertyValue
113 anzsrc-for:08 schema:inDefinedTermSet anzsrc-for:
114 schema:name Information and Computing Sciences
115 rdf:type schema:DefinedTerm
116 anzsrc-for:0801 schema:inDefinedTermSet anzsrc-for:
117 schema:name Artificial Intelligence and Image Processing
118 rdf:type schema:DefinedTerm
119 sg:grant.4314437 http://pending.schema.org/fundedItem sg:pub.10.1007/s41060-017-0042-5
120 rdf:type schema:MonetaryGrant
121 sg:journal.1156617 schema:issn 2364-415X
122 2364-4168
123 schema:name International Journal of Data Science and Analytics
124 schema:publisher Springer Nature
125 rdf:type schema:Periodical
126 sg:person.011016356115.95 schema:affiliation grid-institutes:grid.185648.6
127 schema:familyName Yu
128 schema:givenName Philip S.
129 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.011016356115.95
130 rdf:type schema:Person
131 sg:person.014402567553.41 schema:affiliation grid-institutes:grid.259029.5
132 schema:familyName Xie
133 schema:givenName Sihong
134 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.014402567553.41
135 rdf:type schema:Person
136 sg:pub.10.1007/978-3-319-10605-2_38 schema:sameAs https://app.dimensions.ai/details/publication/pub.1044946925
137 https://doi.org/10.1007/978-3-319-10605-2_38
138 rdf:type schema:CreativeWork
139 sg:pub.10.1007/978-3-642-15555-0_6 schema:sameAs https://app.dimensions.ai/details/publication/pub.1017308440
140 https://doi.org/10.1007/978-3-642-15555-0_6
141 rdf:type schema:CreativeWork
142 grid-institutes:grid.185648.6 schema:alternateName Department of Computer Science, University of Illinois at Chicago, Chicago, IL, USA
143 schema:name Department of Computer Science, University of Illinois at Chicago, Chicago, IL, USA
144 rdf:type schema:Organization
145 grid-institutes:grid.259029.5 schema:alternateName Department of Computer Science and Engineering, Lehigh University, Bethlehem, PA, USA
146 schema:name Department of Computer Science and Engineering, Lehigh University, Bethlehem, PA, USA
147 rdf:type schema:Organization
 




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


...