ACE-Cost

Acquisition cost efficient classifier by hybrid decision tree with local SVM leaves

Liyun Li, Umut Topkara, Nasir Memon

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    The standard prediction process of SVM requires acquisition of all the feature values for every instance. In practice, however, a cost is associated with the mere act of acquisition of a feature, e.g. CPU time needed to compute the feature out of raw data, the dollar amount spent for gleaning more information, or the patient wellness sacrificed by an invasive medical test, etc. In such applications, a budget constrains the classification process from using all of the features. We present, AceCost, a novel classification method that reduces the expected test cost of SVM without compromising from the classification accuracy. Our algorithm uses a cost efficient decision tree to partition the feature space for obtaining coarse decision boundaries, and local SVM classifiers at the leaves of the tree to refine them. The resulting classifiers are also effective in scenarios where several features share overlapping acquisition procedures, hence the cost of acquiring them as a group is less than the sum of the individual acquisition costs. Our experiments on the standard UCI datasets, a network flow detection application, as well as on synthetic datasets show that, the proposed approach achieves classification accuracy of SVM while reducing the test cost by 40%-80%.

    Original languageEnglish (US)
    Title of host publicationMachine Learning and Data Mining in Pattern Recognition - 7th International Conference, MLDM 2011, Proceedings
    Pages60-74
    Number of pages15
    Volume6871 LNAI
    DOIs
    StatePublished - 2011
    Event7th International Conference on Machine Learning and Data Mining, MLDM 2011 - New York, NY, United States
    Duration: Aug 30 2011Sep 3 2011

    Publication series

    NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volume6871 LNAI
    ISSN (Print)03029743
    ISSN (Electronic)16113349

    Other

    Other7th International Conference on Machine Learning and Data Mining, MLDM 2011
    CountryUnited States
    CityNew York, NY
    Period8/30/119/3/11

    Fingerprint

    Decision trees
    Decision tree
    Classifiers
    Classifier
    Costs
    Network Flow
    CPU Time
    Feature Space
    Program processors
    Overlapping
    Acquisition
    Leaves
    Partition
    Scenarios
    Prediction
    Experiment
    Experiments

    Keywords

    • Cost Efficient Classification
    • Decision Tree
    • Postpruning
    • SVM

    ASJC Scopus subject areas

    • Computer Science(all)
    • Theoretical Computer Science

    Cite this

    Li, L., Topkara, U., & Memon, N. (2011). ACE-Cost: Acquisition cost efficient classifier by hybrid decision tree with local SVM leaves. In Machine Learning and Data Mining in Pattern Recognition - 7th International Conference, MLDM 2011, Proceedings (Vol. 6871 LNAI, pp. 60-74). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6871 LNAI). https://doi.org/10.1007/978-3-642-23199-5_5

    ACE-Cost : Acquisition cost efficient classifier by hybrid decision tree with local SVM leaves. / Li, Liyun; Topkara, Umut; Memon, Nasir.

    Machine Learning and Data Mining in Pattern Recognition - 7th International Conference, MLDM 2011, Proceedings. Vol. 6871 LNAI 2011. p. 60-74 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6871 LNAI).

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Li, L, Topkara, U & Memon, N 2011, ACE-Cost: Acquisition cost efficient classifier by hybrid decision tree with local SVM leaves. in Machine Learning and Data Mining in Pattern Recognition - 7th International Conference, MLDM 2011, Proceedings. vol. 6871 LNAI, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 6871 LNAI, pp. 60-74, 7th International Conference on Machine Learning and Data Mining, MLDM 2011, New York, NY, United States, 8/30/11. https://doi.org/10.1007/978-3-642-23199-5_5
    Li L, Topkara U, Memon N. ACE-Cost: Acquisition cost efficient classifier by hybrid decision tree with local SVM leaves. In Machine Learning and Data Mining in Pattern Recognition - 7th International Conference, MLDM 2011, Proceedings. Vol. 6871 LNAI. 2011. p. 60-74. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-642-23199-5_5
    Li, Liyun ; Topkara, Umut ; Memon, Nasir. / ACE-Cost : Acquisition cost efficient classifier by hybrid decision tree with local SVM leaves. Machine Learning and Data Mining in Pattern Recognition - 7th International Conference, MLDM 2011, Proceedings. Vol. 6871 LNAI 2011. pp. 60-74 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
    @inproceedings{d673687c83d944ad8388ce51093ef7e6,
    title = "ACE-Cost: Acquisition cost efficient classifier by hybrid decision tree with local SVM leaves",
    abstract = "The standard prediction process of SVM requires acquisition of all the feature values for every instance. In practice, however, a cost is associated with the mere act of acquisition of a feature, e.g. CPU time needed to compute the feature out of raw data, the dollar amount spent for gleaning more information, or the patient wellness sacrificed by an invasive medical test, etc. In such applications, a budget constrains the classification process from using all of the features. We present, AceCost, a novel classification method that reduces the expected test cost of SVM without compromising from the classification accuracy. Our algorithm uses a cost efficient decision tree to partition the feature space for obtaining coarse decision boundaries, and local SVM classifiers at the leaves of the tree to refine them. The resulting classifiers are also effective in scenarios where several features share overlapping acquisition procedures, hence the cost of acquiring them as a group is less than the sum of the individual acquisition costs. Our experiments on the standard UCI datasets, a network flow detection application, as well as on synthetic datasets show that, the proposed approach achieves classification accuracy of SVM while reducing the test cost by 40{\%}-80{\%}.",
    keywords = "Cost Efficient Classification, Decision Tree, Postpruning, SVM",
    author = "Liyun Li and Umut Topkara and Nasir Memon",
    year = "2011",
    doi = "10.1007/978-3-642-23199-5_5",
    language = "English (US)",
    isbn = "9783642231988",
    volume = "6871 LNAI",
    series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
    pages = "60--74",
    booktitle = "Machine Learning and Data Mining in Pattern Recognition - 7th International Conference, MLDM 2011, Proceedings",

    }

    TY - GEN

    T1 - ACE-Cost

    T2 - Acquisition cost efficient classifier by hybrid decision tree with local SVM leaves

    AU - Li, Liyun

    AU - Topkara, Umut

    AU - Memon, Nasir

    PY - 2011

    Y1 - 2011

    N2 - The standard prediction process of SVM requires acquisition of all the feature values for every instance. In practice, however, a cost is associated with the mere act of acquisition of a feature, e.g. CPU time needed to compute the feature out of raw data, the dollar amount spent for gleaning more information, or the patient wellness sacrificed by an invasive medical test, etc. In such applications, a budget constrains the classification process from using all of the features. We present, AceCost, a novel classification method that reduces the expected test cost of SVM without compromising from the classification accuracy. Our algorithm uses a cost efficient decision tree to partition the feature space for obtaining coarse decision boundaries, and local SVM classifiers at the leaves of the tree to refine them. The resulting classifiers are also effective in scenarios where several features share overlapping acquisition procedures, hence the cost of acquiring them as a group is less than the sum of the individual acquisition costs. Our experiments on the standard UCI datasets, a network flow detection application, as well as on synthetic datasets show that, the proposed approach achieves classification accuracy of SVM while reducing the test cost by 40%-80%.

    AB - The standard prediction process of SVM requires acquisition of all the feature values for every instance. In practice, however, a cost is associated with the mere act of acquisition of a feature, e.g. CPU time needed to compute the feature out of raw data, the dollar amount spent for gleaning more information, or the patient wellness sacrificed by an invasive medical test, etc. In such applications, a budget constrains the classification process from using all of the features. We present, AceCost, a novel classification method that reduces the expected test cost of SVM without compromising from the classification accuracy. Our algorithm uses a cost efficient decision tree to partition the feature space for obtaining coarse decision boundaries, and local SVM classifiers at the leaves of the tree to refine them. The resulting classifiers are also effective in scenarios where several features share overlapping acquisition procedures, hence the cost of acquiring them as a group is less than the sum of the individual acquisition costs. Our experiments on the standard UCI datasets, a network flow detection application, as well as on synthetic datasets show that, the proposed approach achieves classification accuracy of SVM while reducing the test cost by 40%-80%.

    KW - Cost Efficient Classification

    KW - Decision Tree

    KW - Postpruning

    KW - SVM

    UR - http://www.scopus.com/inward/record.url?scp=80052325722&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=80052325722&partnerID=8YFLogxK

    U2 - 10.1007/978-3-642-23199-5_5

    DO - 10.1007/978-3-642-23199-5_5

    M3 - Conference contribution

    SN - 9783642231988

    VL - 6871 LNAI

    T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

    SP - 60

    EP - 74

    BT - Machine Learning and Data Mining in Pattern Recognition - 7th International Conference, MLDM 2011, Proceedings

    ER -