[{"department":[{"_id":"ChLa"}],"quality_controlled":"1","article_processing_charge":"No","status":"public","page":"332 - 337","oa_version":"None","title":"Learning anticipation policies for robot table tennis","date_updated":"2025-07-10T11:52:30Z","publication_status":"published","abstract":[{"lang":"eng","text":"Playing table tennis is a difficult task for robots, especially due to their limitations of acceleration. A key bottleneck is the amount of time needed to reach the desired hitting position and velocity of the racket for returning the incoming ball. Here, it often does not suffice to simply extrapolate the ball's trajectory after the opponent returns it but more information is needed. Humans are able to predict the ball's trajectory based on the opponent's moves and, thus, have a considerable advantage. Hence, we propose to incorporate an anticipation system into robot table tennis players, which enables the robot to react earlier while the opponent is performing the striking movement. Based on visual observation of the opponent's racket movement, the robot can predict the aim of the opponent and adjust its movement generation accordingly. The policies for deciding how and when to react are obtained by reinforcement learning. We conduct experiments with an existing robot player to show that the learned reaction policy can significantly improve the performance of the overall system."}],"language":[{"iso":"eng"}],"user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","day":"01","publist_id":"3293","citation":{"short":"Z. Wang, C. Lampert, K. Mülling, B. Schölkopf, J. Peters, in:, IEEE, 2011, pp. 332–337.","chicago":"Wang, Zhikun, Christoph Lampert, Katharina Mülling, Bernhard Schölkopf, and Jan Peters. “Learning Anticipation Policies for Robot Table Tennis,” 332–37. IEEE, 2011. <a href=\"https://doi.org/10.1109/IROS.2011.6094892\">https://doi.org/10.1109/IROS.2011.6094892</a>.","mla":"Wang, Zhikun, et al. <i>Learning Anticipation Policies for Robot Table Tennis</i>. IEEE, 2011, pp. 332–37, doi:<a href=\"https://doi.org/10.1109/IROS.2011.6094892\">10.1109/IROS.2011.6094892</a>.","ama":"Wang Z, Lampert C, Mülling K, Schölkopf B, Peters J. Learning anticipation policies for robot table tennis. In: IEEE; 2011:332-337. doi:<a href=\"https://doi.org/10.1109/IROS.2011.6094892\">10.1109/IROS.2011.6094892</a>","apa":"Wang, Z., Lampert, C., Mülling, K., Schölkopf, B., &#38; Peters, J. (2011). Learning anticipation policies for robot table tennis (pp. 332–337). Presented at the IROS: Intelligent Robots and Systems, San Francisco, USA: IEEE. <a href=\"https://doi.org/10.1109/IROS.2011.6094892\">https://doi.org/10.1109/IROS.2011.6094892</a>","ista":"Wang Z, Lampert C, Mülling K, Schölkopf B, Peters J. 2011. Learning anticipation policies for robot table tennis. IROS: Intelligent Robots and Systems, 332–337.","ieee":"Z. Wang, C. Lampert, K. Mülling, B. Schölkopf, and J. Peters, “Learning anticipation policies for robot table tennis,” presented at the IROS: Intelligent Robots and Systems, San Francisco, USA, 2011, pp. 332–337."},"date_published":"2011-01-01T00:00:00Z","month":"01","scopus_import":"1","_id":"3337","publisher":"IEEE","conference":{"end_date":"2011-09-30","start_date":"2011-09-25","name":"IROS: Intelligent Robots and Systems","location":"San Francisco, USA"},"author":[{"full_name":"Wang, Zhikun","first_name":"Zhikun","last_name":"Wang"},{"full_name":"Lampert, Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph","orcid":"0000-0001-8622-7887","last_name":"Lampert"},{"first_name":"Katharina","last_name":"Mülling","full_name":"Mülling, Katharina"},{"last_name":"Schölkopf","first_name":"Bernhard","full_name":"Schölkopf, Bernhard"},{"first_name":"Jan","last_name":"Peters","full_name":"Peters, Jan"}],"date_created":"2018-12-11T12:02:45Z","doi":"10.1109/IROS.2011.6094892","year":"2011","type":"conference"},{"date_updated":"2025-09-30T08:48:31Z","language":[{"iso":"eng"}],"article_processing_charge":"No","publication":"IEEE Transactions on Robotics","department":[{"_id":"ChLa"}],"isi":1,"external_id":{"isi":["000291404600015"]},"status":"public","type":"journal_article","year":"2011","publist_id":"3225","issue":"3","day":"21","publisher":"IEEE","month":"05","publication_status":"published","oa_version":"None","title":"Learning dynamic tactile sensing with robust vision based training","abstract":[{"lang":"eng","text":"Dynamic tactile sensing is a fundamental ability to recognize materials and objects. However, while humans are born with partially developed dynamic tactile sensing and quickly master this skill, today's robots remain in their infancy. The development of such a sense requires not only better sensors but the right algorithms to deal with these sensors' data as well. For example, when classifying a material based on touch, the data are noisy, high-dimensional, and contain irrelevant signals as well as essential ones. Few classification methods from machine learning can deal with such problems. In this paper, we propose an efficient approach to infer suitable lower dimensional representations of the tactile data. In order to classify materials based on only the sense of touch, these representations are autonomously discovered using visual information of the surfaces during training. However, accurately pairing vision and tactile samples in real-robot applications is a difficult problem. The proposed approach, therefore, works with weak pairings between the modalities. Experiments show that the resulting approach is very robust and yields significantly higher classification performance based on only dynamic tactile sensing."}],"quality_controlled":"1","page":"545 - 557","volume":27,"intvolume":"        27","date_published":"2011-05-21T00:00:00Z","citation":{"chicago":"Kroemer, Oliver, Christoph Lampert, and Jan Peters. “Learning Dynamic Tactile Sensing with Robust Vision Based Training.” <i>IEEE Transactions on Robotics</i>. IEEE, 2011. <a href=\"https://doi.org/10.1109/TRO.2011.2121130\">https://doi.org/10.1109/TRO.2011.2121130</a>.","mla":"Kroemer, Oliver, et al. “Learning Dynamic Tactile Sensing with Robust Vision Based Training.” <i>IEEE Transactions on Robotics</i>, vol. 27, no. 3, IEEE, 2011, pp. 545–57, doi:<a href=\"https://doi.org/10.1109/TRO.2011.2121130\">10.1109/TRO.2011.2121130</a>.","short":"O. Kroemer, C. Lampert, J. Peters, IEEE Transactions on Robotics 27 (2011) 545–557.","ama":"Kroemer O, Lampert C, Peters J. Learning dynamic tactile sensing with robust vision based training. <i>IEEE Transactions on Robotics</i>. 2011;27(3):545-557. doi:<a href=\"https://doi.org/10.1109/TRO.2011.2121130\">10.1109/TRO.2011.2121130</a>","apa":"Kroemer, O., Lampert, C., &#38; Peters, J. (2011). Learning dynamic tactile sensing with robust vision based training. <i>IEEE Transactions on Robotics</i>. IEEE. <a href=\"https://doi.org/10.1109/TRO.2011.2121130\">https://doi.org/10.1109/TRO.2011.2121130</a>","ista":"Kroemer O, Lampert C, Peters J. 2011. Learning dynamic tactile sensing with robust vision based training. IEEE Transactions on Robotics. 27(3), 545–557.","ieee":"O. Kroemer, C. Lampert, and J. Peters, “Learning dynamic tactile sensing with robust vision based training,” <i>IEEE Transactions on Robotics</i>, vol. 27, no. 3. IEEE, pp. 545–557, 2011."},"user_id":"317138e5-6ab7-11ef-aa6d-ffef3953e345","doi":"10.1109/TRO.2011.2121130","author":[{"full_name":"Kroemer, Oliver","first_name":"Oliver","last_name":"Kroemer"},{"full_name":"Lampert, Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph","last_name":"Lampert","orcid":"0000-0001-8622-7887"},{"first_name":"Jan","last_name":"Peters","full_name":"Peters, Jan"}],"date_created":"2018-12-11T12:03:01Z","_id":"3382","scopus_import":"1"},{"year":"2011","type":"journal_article","month":"08","publisher":"Elsevier","issue":"11","day":"01","publist_id":"3218","language":[{"iso":"eng"}],"date_updated":"2025-09-30T08:45:21Z","status":"public","external_id":{"isi":["000293050700010"]},"publication":"Pattern Recognition Letters","isi":1,"department":[{"_id":"ChLa"}],"article_processing_charge":"No","intvolume":"        32","acknowledgement":"The research leading to these results has received funding from the European Research Council under the European Community’s Seventh Framework Programme (FP7/2007-2013)/ERC Grant Agreement No. 228180. This work was funded in part by the EC project CLASS, IST 027978, and the PASCAL2 network of excellence, IST 2002-506778.","volume":32,"_id":"3389","scopus_import":"1","date_created":"2018-12-11T12:03:03Z","author":[{"last_name":"Blaschko","first_name":"Matthew","full_name":"Blaschko, Matthew"},{"first_name":"Jacquelyn","last_name":"Shelton","full_name":"Shelton, Jacquelyn"},{"last_name":"Bartels","first_name":"Andreas","full_name":"Bartels, Andreas"},{"first_name":"Christoph","last_name":"Lampert","orcid":"0000-0001-8622-7887","full_name":"Lampert, Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87"},{"first_name":"Arthur","last_name":"Gretton","full_name":"Gretton, Arthur"}],"doi":"10.1016/j.patrec.2011.02.011","user_id":"317138e5-6ab7-11ef-aa6d-ffef3953e345","date_published":"2011-08-01T00:00:00Z","citation":{"short":"M. Blaschko, J. Shelton, A. Bartels, C. Lampert, A. Gretton, Pattern Recognition Letters 32 (2011) 1572–1583.","chicago":"Blaschko, Matthew, Jacquelyn Shelton, Andreas Bartels, Christoph Lampert, and Arthur Gretton. “Semi Supervised Kernel Canonical Correlation Analysis with Application to Human FMRI.” <i>Pattern Recognition Letters</i>. Elsevier, 2011. <a href=\"https://doi.org/10.1016/j.patrec.2011.02.011\">https://doi.org/10.1016/j.patrec.2011.02.011</a>.","mla":"Blaschko, Matthew, et al. “Semi Supervised Kernel Canonical Correlation Analysis with Application to Human FMRI.” <i>Pattern Recognition Letters</i>, vol. 32, no. 11, Elsevier, 2011, pp. 1572–83, doi:<a href=\"https://doi.org/10.1016/j.patrec.2011.02.011\">10.1016/j.patrec.2011.02.011</a>.","ama":"Blaschko M, Shelton J, Bartels A, Lampert C, Gretton A. Semi supervised kernel canonical correlation analysis with application to human fMRI. <i>Pattern Recognition Letters</i>. 2011;32(11):1572-1583. doi:<a href=\"https://doi.org/10.1016/j.patrec.2011.02.011\">10.1016/j.patrec.2011.02.011</a>","apa":"Blaschko, M., Shelton, J., Bartels, A., Lampert, C., &#38; Gretton, A. (2011). Semi supervised kernel canonical correlation analysis with application to human fMRI. <i>Pattern Recognition Letters</i>. Elsevier. <a href=\"https://doi.org/10.1016/j.patrec.2011.02.011\">https://doi.org/10.1016/j.patrec.2011.02.011</a>","ista":"Blaschko M, Shelton J, Bartels A, Lampert C, Gretton A. 2011. Semi supervised kernel canonical correlation analysis with application to human fMRI. Pattern Recognition Letters. 32(11), 1572–1583.","ieee":"M. Blaschko, J. Shelton, A. Bartels, C. Lampert, and A. Gretton, “Semi supervised kernel canonical correlation analysis with application to human fMRI,” <i>Pattern Recognition Letters</i>, vol. 32, no. 11. Elsevier, pp. 1572–1583, 2011."},"abstract":[{"lang":"eng","text":"Kernel canonical correlation analysis (KCCA) is a general technique for subspace learning that incorporates principal components analysis (PCA) and Fisher linear discriminant analysis (LDA) as special cases. By finding directions that maximize correlation, KCCA learns representations that are more closely tied to the underlying process that generates the data and can ignore high-variance noise directions. However, for data where acquisition in one or more modalities is expensive or otherwise limited, KCCA may suffer from small sample effects. We propose to use semi-supervised Laplacian regularization to utilize data that are present in only one modality. This approach is able to find highly correlated directions that also lie along the data manifold, resulting in a more robust estimate of correlated subspaces. Functional magnetic resonance imaging (fMRI) acquired data are naturally amenable to subspace techniques as data are well aligned. fMRI data of the human brain are a particularly interesting candidate. In this study we implemented various supervised and semi-supervised versions of KCCA on human fMRI data, with regression to single and multi-variate labels (corresponding to video content subjects viewed during the image acquisition). In each variate condition, the semi-supervised variants of KCCA performed better than the supervised variants, including a supervised variant with Laplacian regularization. We additionally analyze the weights learned by the regression in order to infer brain regions that are important to different types of visual processing."}],"oa_version":"None","title":"Semi supervised kernel canonical correlation analysis with application to human fMRI","publication_status":"published","page":"1572 - 1583","quality_controlled":"1"},{"has_accepted_license":"1","file_date_updated":"2020-07-14T12:46:41Z","department":[{"_id":"ChLa"}],"status":"public","page":"69","title":"Enforcing topological constraints in random field image segmentation","oa_version":"Published Version","file":[{"file_size":26390601,"date_updated":"2020-07-14T12:46:41Z","content_type":"application/pdf","relation":"main_file","access_level":"open_access","creator":"system","checksum":"ad64c2add5fe2ad10e9d5c669f3f9526","file_id":"5495","date_created":"2018-12-12T11:53:34Z","file_name":"IST-2011-0002_IST-2011-0002.pdf"}],"date_updated":"2024-10-09T20:54:30Z","publication_status":"published","ddc":["000"],"abstract":[{"lang":"eng","text":"We introduce TopoCut: a new way to integrate knowledge about topological properties (TPs) into random field image segmentation model. Instead of including TPs as additional constraints during minimization of the energy function, we devise an efficient algorithm for modifying the unary potentials such that the resulting segmentation is guaranteed with the desired properties. Our method is more flexible in the sense that it handles more topology constraints than previous methods, which were only able to enforce pairwise or global connectivity. In particular, our method is very fast, making it for the first time possible to enforce global topological properties in practical image segmentation tasks."}],"language":[{"iso":"eng"}],"related_material":{"record":[{"relation":"later_version","status":"public","id":"3336"}]},"user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","day":"28","citation":{"ieee":"C. Chen, D. Freedman, and C. Lampert, <i>Enforcing topological constraints in random field image segmentation</i>. IST Austria, 2011.","ista":"Chen C, Freedman D, Lampert C. 2011. Enforcing topological constraints in random field image segmentation, IST Austria, 69p.","apa":"Chen, C., Freedman, D., &#38; Lampert, C. (2011). <i>Enforcing topological constraints in random field image segmentation</i>. IST Austria. <a href=\"https://doi.org/10.15479/AT:IST-2011-0002\">https://doi.org/10.15479/AT:IST-2011-0002</a>","ama":"Chen C, Freedman D, Lampert C. <i>Enforcing Topological Constraints in Random Field Image Segmentation</i>. IST Austria; 2011. doi:<a href=\"https://doi.org/10.15479/AT:IST-2011-0002\">10.15479/AT:IST-2011-0002</a>","chicago":"Chen, Chao, Daniel Freedman, and Christoph Lampert. <i>Enforcing Topological Constraints in Random Field Image Segmentation</i>. IST Austria, 2011. <a href=\"https://doi.org/10.15479/AT:IST-2011-0002\">https://doi.org/10.15479/AT:IST-2011-0002</a>.","mla":"Chen, Chao, et al. <i>Enforcing Topological Constraints in Random Field Image Segmentation</i>. IST Austria, 2011, doi:<a href=\"https://doi.org/10.15479/AT:IST-2011-0002\">10.15479/AT:IST-2011-0002</a>.","short":"C. Chen, D. Freedman, C. Lampert, Enforcing Topological Constraints in Random Field Image Segmentation, IST Austria, 2011."},"oa":1,"date_published":"2011-03-28T00:00:00Z","month":"03","_id":"5386","publisher":"IST Austria","date_created":"2018-12-12T11:39:02Z","author":[{"full_name":"Chen, Chao","id":"3E92416E-F248-11E8-B48F-1D18A9856A87","first_name":"Chao","last_name":"Chen"},{"first_name":"Daniel","last_name":"Freedman","full_name":"Freedman, Daniel"},{"full_name":"Lampert, Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph","orcid":"0000-0001-8622-7887","last_name":"Lampert"}],"doi":"10.15479/AT:IST-2011-0002","alternative_title":["IST Austria Technical Report"],"year":"2011","publication_identifier":{"issn":["2664-1690"]},"pubrep_id":"22","type":"technical_report"},{"type":"conference","year":"2010","day":"04","publist_id":"2431","conference":{"start_date":"2010-09-05","name":"ECCV: European Conference on Computer Vision","end_date":"2010-09-11","location":"Heraklion, Crete, Greece"},"publisher":"Springer","month":"11","file":[{"file_name":"2010_ECCV_Nowozin.pdf","date_created":"2020-05-19T16:27:34Z","file_id":"7871","checksum":"3716e10e161f7c714fd17ec193a223c3","creator":"dernst","access_level":"open_access","relation":"main_file","file_size":4087332,"date_updated":"2020-07-14T12:46:16Z","content_type":"application/pdf"}],"ddc":["000"],"date_updated":"2021-01-12T07:52:14Z","language":[{"iso":"eng"}],"article_processing_charge":"No","department":[{"_id":"ChLa"}],"has_accepted_license":"1","status":"public","volume":6316,"alternative_title":["LNCS"],"intvolume":"      6316","date_published":"2010-11-04T00:00:00Z","oa":1,"citation":{"apa":"Nowozin, S., Gehler, P., &#38; Lampert, C. (2010). On parameter learning in CRF-based approaches to object class image segmentation (Vol. 6316, pp. 98–111). Presented at the ECCV: European Conference on Computer Vision, Heraklion, Crete, Greece: Springer. <a href=\"https://doi.org/10.1007/978-3-642-15567-3_8\">https://doi.org/10.1007/978-3-642-15567-3_8</a>","ama":"Nowozin S, Gehler P, Lampert C. On parameter learning in CRF-based approaches to object class image segmentation. In: Vol 6316. Springer; 2010:98-111. doi:<a href=\"https://doi.org/10.1007/978-3-642-15567-3_8\">10.1007/978-3-642-15567-3_8</a>","mla":"Nowozin, Sebastian, et al. <i>On Parameter Learning in CRF-Based Approaches to Object Class Image Segmentation</i>. Vol. 6316, Springer, 2010, pp. 98–111, doi:<a href=\"https://doi.org/10.1007/978-3-642-15567-3_8\">10.1007/978-3-642-15567-3_8</a>.","short":"S. Nowozin, P. Gehler, C. Lampert, in:, Springer, 2010, pp. 98–111.","chicago":"Nowozin, Sebastian, Peter Gehler, and Christoph Lampert. “On Parameter Learning in CRF-Based Approaches to Object Class Image Segmentation,” 6316:98–111. Springer, 2010. <a href=\"https://doi.org/10.1007/978-3-642-15567-3_8\">https://doi.org/10.1007/978-3-642-15567-3_8</a>.","ieee":"S. Nowozin, P. Gehler, and C. Lampert, “On parameter learning in CRF-based approaches to object class image segmentation,” presented at the ECCV: European Conference on Computer Vision, Heraklion, Crete, Greece, 2010, vol. 6316, pp. 98–111.","ista":"Nowozin S, Gehler P, Lampert C. 2010. On parameter learning in CRF-based approaches to object class image segmentation. ECCV: European Conference on Computer Vision, LNCS, vol. 6316, 98–111."},"user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","doi":"10.1007/978-3-642-15567-3_8","date_created":"2018-12-11T12:05:12Z","author":[{"first_name":"Sebastian","last_name":"Nowozin","full_name":"Nowozin, Sebastian"},{"full_name":"Gehler, Peter","last_name":"Gehler","first_name":"Peter"},{"full_name":"Lampert, Christoph","id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","first_name":"Christoph","orcid":"0000-0001-8622-7887","last_name":"Lampert"}],"_id":"3793","scopus_import":1,"publication_status":"published","oa_version":"Submitted Version","title":"On parameter learning in CRF-based approaches to object class image segmentation","abstract":[{"text":"Recent progress in per-pixel object class labeling of natural images can be attributed to the use of multiple types of image features and sound statistical learning approaches. Within the latter, Conditional Random Fields (CRF) are prominently used for their ability to represent interactions between random variables. Despite their popularity in computer vision, parameter learning for CRFs has remained difficult, popular approaches being cross-validation and piecewise training.\r\nIn this work, we propose a simple yet expressive tree-structured CRF based on a recent hierarchical image segmentation method. Our model combines and weights multiple image features within a hierarchical representation and allows simple and efficient globally-optimal learning of ≈ 105 parameters. The tractability of our model allows us to pose and answer some of the open questions regarding parameter learning applying to CRF-based approaches. The key findings for learning CRF models are, from the obvious to the surprising, i) multiple image features always help, ii) the limiting dimension with respect to current models is the amount of training data, iii) piecewise training is competitive, iv) current methods for max-margin training fail for models with many parameters.\r\n","lang":"eng"}],"quality_controlled":"1","file_date_updated":"2020-07-14T12:46:16Z","page":"98 - 111"},{"publist_id":"2433","day":"10","OA_type":"closed access","conference":{"location":"Heraklion, Crete, Greece","start_date":"2010-09-05","name":"ECCV: European Conference on Computer Vision","end_date":"2010-09-11"},"publisher":"Springer","month":"11","type":"conference","year":"2010","article_processing_charge":"No","publication":"11th European Conference on Computer Vision","department":[{"_id":"ChLa"}],"status":"public","date_updated":"2025-05-20T06:43:33Z","language":[{"iso":"eng"}],"date_published":"2010-11-10T00:00:00Z","citation":{"ama":"Lampert C, Krömer O. Weakly-paired maximum covariance analysis for multimodal dimensionality reduction and transfer learning. In: <i>11th European Conference on Computer Vision</i>. Vol 6312. Springer; 2010:566-579. doi:<a href=\"https://doi.org/10.1007/978-3-642-15552-9_41\">10.1007/978-3-642-15552-9_41</a>","mla":"Lampert, Christoph, and Oliver Krömer. “Weakly-Paired Maximum Covariance Analysis for Multimodal Dimensionality Reduction and Transfer Learning.” <i>11th European Conference on Computer Vision</i>, vol. 6312, Springer, 2010, pp. 566–79, doi:<a href=\"https://doi.org/10.1007/978-3-642-15552-9_41\">10.1007/978-3-642-15552-9_41</a>.","chicago":"Lampert, Christoph, and Oliver Krömer. “Weakly-Paired Maximum Covariance Analysis for Multimodal Dimensionality Reduction and Transfer Learning.” In <i>11th European Conference on Computer Vision</i>, 6312:566–79. Springer, 2010. <a href=\"https://doi.org/10.1007/978-3-642-15552-9_41\">https://doi.org/10.1007/978-3-642-15552-9_41</a>.","short":"C. Lampert, O. Krömer, in:, 11th European Conference on Computer Vision, Springer, 2010, pp. 566–579.","apa":"Lampert, C., &#38; Krömer, O. (2010). Weakly-paired maximum covariance analysis for multimodal dimensionality reduction and transfer learning. In <i>11th European Conference on Computer Vision</i> (Vol. 6312, pp. 566–579). Heraklion, Crete, Greece: Springer. <a href=\"https://doi.org/10.1007/978-3-642-15552-9_41\">https://doi.org/10.1007/978-3-642-15552-9_41</a>","ista":"Lampert C, Krömer O. 2010. Weakly-paired maximum covariance analysis for multimodal dimensionality reduction and transfer learning. 11th European Conference on Computer Vision. ECCV: European Conference on Computer Vision, LNCS, vol. 6312, 566–579.","ieee":"C. Lampert and O. Krömer, “Weakly-paired maximum covariance analysis for multimodal dimensionality reduction and transfer learning,” in <i>11th European Conference on Computer Vision</i>, Heraklion, Crete, Greece, 2010, vol. 6312, pp. 566–579."},"user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","doi":"10.1007/978-3-642-15552-9_41","author":[{"id":"40C20FD2-F248-11E8-B48F-1D18A9856A87","full_name":"Lampert, Christoph","last_name":"Lampert","orcid":"0000-0001-8622-7887","first_name":"Christoph"},{"full_name":"Krömer, Oliver","last_name":"Krömer","first_name":"Oliver"}],"date_created":"2018-12-11T12:05:12Z","_id":"3794","scopus_import":"1","volume":6312,"publication_identifier":{"eisbn":["9783642155529"],"eissn":["1611-3349"]},"alternative_title":["LNCS"],"intvolume":"      6312","quality_controlled":"1","page":"566 - 579","publication_status":"published","oa_version":"None","title":"Weakly-paired maximum covariance analysis for multimodal dimensionality reduction and transfer learning","abstract":[{"lang":"eng","text":"We study the problem of multimodal dimensionality reduction assuming that data samples can be missing at training time, and not all data modalities may be present at application time. Maximum covariance analysis, as a generalization of PCA, has many desirable properties, but its application to practical problems is limited by its need for perfectly paired data. We overcome this limitation by a latent variable approach that allows working with weakly paired data and is still able to efficiently process large datasets using standard numerical routines. The resulting weakly paired maximum covariance analysis often finds better representations than alternative methods, as we show in two exemplary tasks: texture discrimination and transfer learning."}]}]