@article{karasikov2014dimensionality,
  title={Dimensionality reduction for multi-class learning problems reduced to multiple binary problems},
  author={Karasikov, M. E. and Maximov, Y. V.},
  journal={Journal of Machine Learning and Data Analysis},
  volume={1},
  number={9},
  pages={1273--1290},
  year={2014},
  url={http://jmlda.org/papers/doc/2014/no9/Karasikov2014Multiclass.pdf},
  abstract={Modern machine learning problems, such as image classification, video recognition, text retrieval or engineering diagnostics, leads to the analysis of multi-class learning methods for high-dimensional datasets which can not be solved without data pre-processing. Principal Component Analysis and its randomized versions are some of the most widespread dimensionality reduction methods. We analyze the classification performance of various approaches to multi-class classification (One-vs-One, One-vs-All, Error-Correcting Output Codes) in combination with the dimensionality reduction based on Random Gaussian Projections. Computational efficiency of the Random Projections distinguishes it from other dimensionality reduction methods. With that, low-distortion property of this mapping allows to reduce dimensionality thrice and more with imperceptible quality losses. This leads to an effective and computationally cheap approach for solving multi-class problems in high-dimensional space. Basic theoretical foundations of the approach as well as its computational complexity analysis are discussed. Numerical stability and quality of the method proposed is supported by empirical evaluation of the approach. We provide a number of experiments for different machine learning methods over various real datasets from the open-source machine learning repositories. Experiments show applicability of Random Projections for cheap selection of the most suitable classifier, its parameters optimization and multi-class classification approach selection.},
  keywords={multi-class classification; One-vs-All; One-vs-One; Error-Correcting Output Codes; Johnson-Lindenstrauss lemma; dimensionality reduction; Random Projections.},
  note={(In Russian)},
}

{"_id":"HihDhcA8pXEw5dq5i","bibbaseid":"karasikov-maximov-dimensionalityreductionformulticlasslearningproblemsreducedtomultiplebinaryproblems-2014","downloads":0,"creationDate":"2017-03-12T08:22:54.108Z","title":"Dimensionality reduction for multi-class learning problems reduced to multiple binary problems","author_short":["Karasikov, M. E.","Maximov, Y. V."],"year":2014,"bibtype":"article","biburl":"http://karasikov.com/publications/publications.bib","bibdata":{"bibtype":"article","type":"article","title":"Dimensionality reduction for multi-class learning problems reduced to multiple binary problems","author":[{"propositions":[],"lastnames":["Karasikov"],"firstnames":["M.","E."],"suffixes":[]},{"propositions":[],"lastnames":["Maximov"],"firstnames":["Y.","V."],"suffixes":[]}],"journal":"Journal of Machine Learning and Data Analysis","volume":"1","number":"9","pages":"1273--1290","year":"2014","url":"http://jmlda.org/papers/doc/2014/no9/Karasikov2014Multiclass.pdf","abstract":"Modern machine learning problems, such as image classification, video recognition, text retrieval or engineering diagnostics, leads to the analysis of multi-class learning methods for high-dimensional datasets which can not be solved without data pre-processing. Principal Component Analysis and its randomized versions are some of the most widespread dimensionality reduction methods. We analyze the classification performance of various approaches to multi-class classification (One-vs-One, One-vs-All, Error-Correcting Output Codes) in combination with the dimensionality reduction based on Random Gaussian Projections. Computational efficiency of the Random Projections distinguishes it from other dimensionality reduction methods. With that, low-distortion property of this mapping allows to reduce dimensionality thrice and more with imperceptible quality losses. This leads to an effective and computationally cheap approach for solving multi-class problems in high-dimensional space. Basic theoretical foundations of the approach as well as its computational complexity analysis are discussed. Numerical stability and quality of the method proposed is supported by empirical evaluation of the approach. We provide a number of experiments for different machine learning methods over various real datasets from the open-source machine learning repositories. Experiments show applicability of Random Projections for cheap selection of the most suitable classifier, its parameters optimization and multi-class classification approach selection.","keywords":"multi-class classification; One-vs-All; One-vs-One; Error-Correcting Output Codes; Johnson-Lindenstrauss lemma; dimensionality reduction; Random Projections.","note":"(In Russian)","bibtex":"@article{karasikov2014dimensionality,\n title={Dimensionality reduction for multi-class learning problems reduced to multiple binary problems},\n author={Karasikov, M. E. and Maximov, Y. V.},\n journal={Journal of Machine Learning and Data Analysis},\n volume={1},\n number={9},\n pages={1273--1290},\n year={2014},\n url={http://jmlda.org/papers/doc/2014/no9/Karasikov2014Multiclass.pdf},\n abstract={Modern machine learning problems, such as image classification, video recognition, text retrieval or engineering diagnostics, leads to the analysis of multi-class learning methods for high-dimensional datasets which can not be solved without data pre-processing. Principal Component Analysis and its randomized versions are some of the most widespread dimensionality reduction methods. We analyze the classification performance of various approaches to multi-class classification (One-vs-One, One-vs-All, Error-Correcting Output Codes) in combination with the dimensionality reduction based on Random Gaussian Projections. Computational efficiency of the Random Projections distinguishes it from other dimensionality reduction methods. With that, low-distortion property of this mapping allows to reduce dimensionality thrice and more with imperceptible quality losses. This leads to an effective and computationally cheap approach for solving multi-class problems in high-dimensional space. Basic theoretical foundations of the approach as well as its computational complexity analysis are discussed. Numerical stability and quality of the method proposed is supported by empirical evaluation of the approach. We provide a number of experiments for different machine learning methods over various real datasets from the open-source machine learning repositories. Experiments show applicability of Random Projections for cheap selection of the most suitable classifier, its parameters optimization and multi-class classification approach selection.},\n keywords={multi-class classification; One-vs-All; One-vs-One; Error-Correcting Output Codes; Johnson-Lindenstrauss lemma; dimensionality reduction; Random Projections.},\n note={(In Russian)},\n}\n","author_short":["Karasikov, M. E.","Maximov, Y. V."],"key":"karasikov2014dimensionality","id":"karasikov2014dimensionality","bibbaseid":"karasikov-maximov-dimensionalityreductionformulticlasslearningproblemsreducedtomultiplebinaryproblems-2014","role":"author","urls":{"Paper":"http://jmlda.org/papers/doc/2014/no9/Karasikov2014Multiclass.pdf"},"keyword":["multi-class classification; One-vs-All; One-vs-One; Error-Correcting Output Codes; Johnson-Lindenstrauss lemma; dimensionality reduction; Random Projections."],"downloads":0,"html":""},"search_terms":["dimensionality","reduction","multi","class","learning","problems","reduced","multiple","binary","problems","karasikov","maximov"],"keywords":["multi-class classification; one-vs-all; one-vs-one; error-correcting output codes; johnson-lindenstrauss lemma; dimensionality reduction; random projections."],"authorIDs":[],"dataSources":["axreh2qBaCxGyngLn"]}