On-line Random Forests. Saffari, A., Leistner, C., Santner, J., Godec, M., & Bischof, H. In 2009 IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops, pages 1393–1400, September, 2009. doi abstract bibtex Random Forests (RFs) are frequently used in many computer vision and machine learning applications. Their popularity is mainly driven by their high computational efficiency during both training and evaluation while achieving state-of-the-art results. However, in most applications RFs are used off-line. This limits their usability for many practical problems, for instance, when training data arrives sequentially or the underlying distribution is continuously changing. In this paper, we propose a novel on-line random forest algorithm. We combine ideas from on-line bagging, extremely randomized forests and propose an on-line decision tree growing procedure. Additionally, we add a temporal weighting scheme for adaptively discarding some trees based on their out-of-bag-error in given time intervals and consequently growing of new trees. The experiments on common machine learning data sets show that our algorithm converges to the performance of the off-line RF. Additionally, we conduct experiments for visual tracking, where we demonstrate real-time state-of-the-art performance on well-known scenarios and show good performance in case of occlusions and appearance changes where we outperform trackers based on on-line boosting. Finally, we demonstrate the usability of on-line RFs on the task of interactive real-time segmentation.
@inproceedings{saffari_-line_2009,
title = {On-line {Random} {Forests}},
doi = {10.1109/ICCVW.2009.5457447},
abstract = {Random Forests (RFs) are frequently used in many computer vision and machine learning applications. Their popularity is mainly driven by their high computational efficiency during both training and evaluation while achieving state-of-the-art results. However, in most applications RFs are used off-line. This limits their usability for many practical problems, for instance, when training data arrives sequentially or the underlying distribution is continuously changing. In this paper, we propose a novel on-line random forest algorithm. We combine ideas from on-line bagging, extremely randomized forests and propose an on-line decision tree growing procedure. Additionally, we add a temporal weighting scheme for adaptively discarding some trees based on their out-of-bag-error in given time intervals and consequently growing of new trees. The experiments on common machine learning data sets show that our algorithm converges to the performance of the off-line RF. Additionally, we conduct experiments for visual tracking, where we demonstrate real-time state-of-the-art performance on well-known scenarios and show good performance in case of occlusions and appearance changes where we outperform trackers based on on-line boosting. Finally, we demonstrate the usability of on-line RFs on the task of interactive real-time segmentation.},
booktitle = {2009 {IEEE} 12th {International} {Conference} on {Computer} {Vision} {Workshops}, {ICCV} {Workshops}},
author = {Saffari, Amir and Leistner, Christian and Santner, Jakob and Godec, Martin and Bischof, Horst},
month = sep,
year = {2009},
keywords = {Application software, Bagging, Computational efficiency, Computer vision, Decision trees, Machine learning, Machine learning algorithms, Radio frequency, Training data, Usability},
pages = {1393--1400},
}
Downloads: 0
{"_id":"QLyfG39tpJwwKeFdd","bibbaseid":"saffari-leistner-santner-godec-bischof-onlinerandomforests-2009","downloads":0,"creationDate":"2018-01-22T16:01:09.729Z","title":"On-line Random Forests","author_short":["Saffari, A.","Leistner, C.","Santner, J.","Godec, M.","Bischof, H."],"year":2009,"bibtype":"inproceedings","biburl":"https://bibbase.org/zotero/mh_lenguyen","bibdata":{"bibtype":"inproceedings","type":"inproceedings","title":"On-line Random Forests","doi":"10.1109/ICCVW.2009.5457447","abstract":"Random Forests (RFs) are frequently used in many computer vision and machine learning applications. Their popularity is mainly driven by their high computational efficiency during both training and evaluation while achieving state-of-the-art results. However, in most applications RFs are used off-line. This limits their usability for many practical problems, for instance, when training data arrives sequentially or the underlying distribution is continuously changing. In this paper, we propose a novel on-line random forest algorithm. We combine ideas from on-line bagging, extremely randomized forests and propose an on-line decision tree growing procedure. Additionally, we add a temporal weighting scheme for adaptively discarding some trees based on their out-of-bag-error in given time intervals and consequently growing of new trees. The experiments on common machine learning data sets show that our algorithm converges to the performance of the off-line RF. Additionally, we conduct experiments for visual tracking, where we demonstrate real-time state-of-the-art performance on well-known scenarios and show good performance in case of occlusions and appearance changes where we outperform trackers based on on-line boosting. Finally, we demonstrate the usability of on-line RFs on the task of interactive real-time segmentation.","booktitle":"2009 IEEE 12th International Conference on Computer Vision Workshops, ICCV Workshops","author":[{"propositions":[],"lastnames":["Saffari"],"firstnames":["Amir"],"suffixes":[]},{"propositions":[],"lastnames":["Leistner"],"firstnames":["Christian"],"suffixes":[]},{"propositions":[],"lastnames":["Santner"],"firstnames":["Jakob"],"suffixes":[]},{"propositions":[],"lastnames":["Godec"],"firstnames":["Martin"],"suffixes":[]},{"propositions":[],"lastnames":["Bischof"],"firstnames":["Horst"],"suffixes":[]}],"month":"September","year":"2009","keywords":"Application software, Bagging, Computational efficiency, Computer vision, Decision trees, Machine learning, Machine learning algorithms, Radio frequency, Training data, Usability","pages":"1393–1400","bibtex":"@inproceedings{saffari_-line_2009,\n\ttitle = {On-line {Random} {Forests}},\n\tdoi = {10.1109/ICCVW.2009.5457447},\n\tabstract = {Random Forests (RFs) are frequently used in many computer vision and machine learning applications. Their popularity is mainly driven by their high computational efficiency during both training and evaluation while achieving state-of-the-art results. However, in most applications RFs are used off-line. This limits their usability for many practical problems, for instance, when training data arrives sequentially or the underlying distribution is continuously changing. In this paper, we propose a novel on-line random forest algorithm. We combine ideas from on-line bagging, extremely randomized forests and propose an on-line decision tree growing procedure. Additionally, we add a temporal weighting scheme for adaptively discarding some trees based on their out-of-bag-error in given time intervals and consequently growing of new trees. The experiments on common machine learning data sets show that our algorithm converges to the performance of the off-line RF. Additionally, we conduct experiments for visual tracking, where we demonstrate real-time state-of-the-art performance on well-known scenarios and show good performance in case of occlusions and appearance changes where we outperform trackers based on on-line boosting. Finally, we demonstrate the usability of on-line RFs on the task of interactive real-time segmentation.},\n\tbooktitle = {2009 {IEEE} 12th {International} {Conference} on {Computer} {Vision} {Workshops}, {ICCV} {Workshops}},\n\tauthor = {Saffari, Amir and Leistner, Christian and Santner, Jakob and Godec, Martin and Bischof, Horst},\n\tmonth = sep,\n\tyear = {2009},\n\tkeywords = {Application software, Bagging, Computational efficiency, Computer vision, Decision trees, Machine learning, Machine learning algorithms, Radio frequency, Training data, Usability},\n\tpages = {1393--1400},\n}\n\n\n\n","author_short":["Saffari, A.","Leistner, C.","Santner, J.","Godec, M.","Bischof, H."],"key":"saffari_-line_2009","id":"saffari_-line_2009","bibbaseid":"saffari-leistner-santner-godec-bischof-onlinerandomforests-2009","role":"author","urls":{},"keyword":["Application software","Bagging","Computational efficiency","Computer vision","Decision trees","Machine learning","Machine learning algorithms","Radio frequency","Training data","Usability"],"metadata":{"authorlinks":{}},"downloads":0,"html":""},"search_terms":["line","random","forests","saffari","leistner","santner","godec","bischof"],"keywords":["application software","bagging","computational efficiency","computer vision","decision trees","machine learning","machine learning algorithms","radio frequency","training data","usability"],"authorIDs":[],"dataSources":["9cexBw6hrwgyZphZZ","iwKepCrWBps7ojhDx"]}