Meta-classifiers for exploiting feature dependencies in automatic target recognition

Umamahesh Srinivas, Vishal Monga, Raghu G. Raj

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

    13 Citations (Scopus)

    Abstract

    Of active interest in automatic target recognition (ATR) is the problem of combining the complementary merits of multiple classifiers. This is inspired by decades of research in the area which has seen a variety of fairly successful feature extraction techniques as well as decision engines being developed. While heuristically based fusion techniques are omnipresent, this paper explores a principled meta-classification strategy that is based on the exploitation of correlation between multiple feature extractors as well as decision engines. We present two learning algorithms respectively based on support vector machines and AdaBoost, which combine soft-outputs of state of the art individual classifiers to yield an overall improvement in recognition rates. Experimental results obtained from benchmark SAR image databases reveal that the proposed meta-classification strategies are not only asymptotically superior but also have better robustness to choice of training over state-of-the art individual classifiers.

    Original languageEnglish (US)
    Title of host publicationRadarCon'11 - In the Eye of the Storm: 2011 IEEE Radar Conference
    Pages147-151
    Number of pages5
    DOIs
    StatePublished - 2011
    Event2011 IEEE Radar Conference: In the Eye of the Storm, RadarCon'11 - Kansas City, MO, United States
    Duration: May 23 2011May 27 2011

    Other

    Other2011 IEEE Radar Conference: In the Eye of the Storm, RadarCon'11
    CountryUnited States
    CityKansas City, MO
    Period5/23/115/27/11

    Fingerprint

    Automatic target recognition
    Classifiers
    Engines
    Adaptive boosting
    Learning algorithms
    Support vector machines
    Feature extraction
    Fusion reactions

    All Science Journal Classification (ASJC) codes

    • Electrical and Electronic Engineering

    Cite this

    Srinivas, U., Monga, V., & Raj, R. G. (2011). Meta-classifiers for exploiting feature dependencies in automatic target recognition. In RadarCon'11 - In the Eye of the Storm: 2011 IEEE Radar Conference (pp. 147-151). [5960517] https://doi.org/10.1109/RADAR.2011.5960517
    Srinivas, Umamahesh ; Monga, Vishal ; Raj, Raghu G. / Meta-classifiers for exploiting feature dependencies in automatic target recognition. RadarCon'11 - In the Eye of the Storm: 2011 IEEE Radar Conference. 2011. pp. 147-151
    @inproceedings{4f400fd7d38e43c99b6ec7e67d877afa,
    title = "Meta-classifiers for exploiting feature dependencies in automatic target recognition",
    abstract = "Of active interest in automatic target recognition (ATR) is the problem of combining the complementary merits of multiple classifiers. This is inspired by decades of research in the area which has seen a variety of fairly successful feature extraction techniques as well as decision engines being developed. While heuristically based fusion techniques are omnipresent, this paper explores a principled meta-classification strategy that is based on the exploitation of correlation between multiple feature extractors as well as decision engines. We present two learning algorithms respectively based on support vector machines and AdaBoost, which combine soft-outputs of state of the art individual classifiers to yield an overall improvement in recognition rates. Experimental results obtained from benchmark SAR image databases reveal that the proposed meta-classification strategies are not only asymptotically superior but also have better robustness to choice of training over state-of-the art individual classifiers.",
    author = "Umamahesh Srinivas and Vishal Monga and Raj, {Raghu G.}",
    year = "2011",
    doi = "10.1109/RADAR.2011.5960517",
    language = "English (US)",
    isbn = "9781424489022",
    pages = "147--151",
    booktitle = "RadarCon'11 - In the Eye of the Storm: 2011 IEEE Radar Conference",

    }

    Srinivas, U, Monga, V & Raj, RG 2011, Meta-classifiers for exploiting feature dependencies in automatic target recognition. in RadarCon'11 - In the Eye of the Storm: 2011 IEEE Radar Conference., 5960517, pp. 147-151, 2011 IEEE Radar Conference: In the Eye of the Storm, RadarCon'11, Kansas City, MO, United States, 5/23/11. https://doi.org/10.1109/RADAR.2011.5960517

    Meta-classifiers for exploiting feature dependencies in automatic target recognition. / Srinivas, Umamahesh; Monga, Vishal; Raj, Raghu G.

    RadarCon'11 - In the Eye of the Storm: 2011 IEEE Radar Conference. 2011. p. 147-151 5960517.

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

    TY - GEN

    T1 - Meta-classifiers for exploiting feature dependencies in automatic target recognition

    AU - Srinivas, Umamahesh

    AU - Monga, Vishal

    AU - Raj, Raghu G.

    PY - 2011

    Y1 - 2011

    N2 - Of active interest in automatic target recognition (ATR) is the problem of combining the complementary merits of multiple classifiers. This is inspired by decades of research in the area which has seen a variety of fairly successful feature extraction techniques as well as decision engines being developed. While heuristically based fusion techniques are omnipresent, this paper explores a principled meta-classification strategy that is based on the exploitation of correlation between multiple feature extractors as well as decision engines. We present two learning algorithms respectively based on support vector machines and AdaBoost, which combine soft-outputs of state of the art individual classifiers to yield an overall improvement in recognition rates. Experimental results obtained from benchmark SAR image databases reveal that the proposed meta-classification strategies are not only asymptotically superior but also have better robustness to choice of training over state-of-the art individual classifiers.

    AB - Of active interest in automatic target recognition (ATR) is the problem of combining the complementary merits of multiple classifiers. This is inspired by decades of research in the area which has seen a variety of fairly successful feature extraction techniques as well as decision engines being developed. While heuristically based fusion techniques are omnipresent, this paper explores a principled meta-classification strategy that is based on the exploitation of correlation between multiple feature extractors as well as decision engines. We present two learning algorithms respectively based on support vector machines and AdaBoost, which combine soft-outputs of state of the art individual classifiers to yield an overall improvement in recognition rates. Experimental results obtained from benchmark SAR image databases reveal that the proposed meta-classification strategies are not only asymptotically superior but also have better robustness to choice of training over state-of-the art individual classifiers.

    UR - http://www.scopus.com/inward/record.url?scp=80052439027&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=80052439027&partnerID=8YFLogxK

    U2 - 10.1109/RADAR.2011.5960517

    DO - 10.1109/RADAR.2011.5960517

    M3 - Conference contribution

    AN - SCOPUS:80052439027

    SN - 9781424489022

    SP - 147

    EP - 151

    BT - RadarCon'11 - In the Eye of the Storm: 2011 IEEE Radar Conference

    ER -

    Srinivas U, Monga V, Raj RG. Meta-classifiers for exploiting feature dependencies in automatic target recognition. In RadarCon'11 - In the Eye of the Storm: 2011 IEEE Radar Conference. 2011. p. 147-151. 5960517 https://doi.org/10.1109/RADAR.2011.5960517