default search action
Google
Google Scholar
Semantic Scholar
Internet Archive Scholar
CiteSeerX
ORCIDMiguel Á. Carreira-Perpiñán
Person information
Refine list
refinements active!
zoomed in on ?? of ?? records
view refined list in
export refined list as
showing all ?? records
2020 – today
- 2024
[j10]Suryabhan Singh Hada
, Miguel Á. Carreira-Perpiñán, Arman Zharmagambetov:
Sparse oblique decision trees: a tool to understand and manipulate neural net features. Data Min. Knowl. Discov. 38(5): 2863-2902 (2024)- [c102]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán, Arman Zharmagambetov:
Sparse Oblique Decision Trees: A Tool to Understand and Manipulate Neural Net Features. HI-AI@KDD 2024: 102-116 - [c101]Magzhan Gabidolla, Arman Zharmagambetov, Miguel Á. Carreira-Perpiñán:
Beyond the ROC Curve: Classification Trees Using Cost-Optimal Curves, with Application to Imbalanced Datasets. ICML 2024 - [c100]Rasul Kairgeldin, Miguel Á. Carreira-Perpiñán:
Bivariate Decision Trees: Smaller, Interpretable, More Accurate. KDD 2024: 1336-1347 - [i28]Louis Rustenholz, Maximiliano Klemen, Miguel Á. Carreira-Perpiñán, Pedro López-García:
A Machine Learning-based Approach for Solving Recurrence Relations and its use in Cost Analysis of Logic Programs. CoRR abs/2405.06972 (2024) - 2023
- [c99]Miguel Á. Carreira-Perpiñán, Suryabhan Singh Hada:
Very Fast, Approximate Counterfactual Explanations for Decision Forests. AAAI 2023: 6935-6943 - [c98]Miguel Á. Carreira-Perpiñán, Magzhan Gabidolla, Arman Zharmagambetov:
Towards Better Decision Forests: Forest Alternating Optimization. CVPR 2023: 7589-7598 - [c97]Maximiliano Klemen, Miguel Á. Carreira-Perpiñán, Pedro López-García:
Solving Recurrence Relations using Machine Learning, with Application to Cost Analysis. ICLP 2023: 155-168 - [i27]Miguel Á. Carreira-Perpiñán, Suryabhan Singh Hada:
Very fast, approximate counterfactual explanations for decision forests. CoRR abs/2303.02883 (2023) - [i26]Miguel Á. Carreira-Perpiñán, Suryabhan Singh Hada:
Inverse classification with logistic and softmax classifiers: efficient optimization. CoRR abs/2309.08945 (2023) - 2022
- [c96]Arman Serikuly Zharmagambetov, Miguel Á. Carreira-Perpiñán:
Learning Interpretable, Tree-Based Projection Mappings for Nonlinear Embeddings. AISTATS 2022: 9550-9570 - [c95]Magzhan Gabidolla, Miguel Á. Carreira-Perpiñán:
Pushing the Envelope of Gradient Boosting Forests via Globally-Optimized Oblique Trees. CVPR 2022: 285-294 - [c94]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán:
Interpretable Image Classification Using Sparse Oblique Decision Trees. ICASSP 2022: 2759-2763 - [c93]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
Exploring the Effect of ℓ0/ℓ2 Regularization in Neural Network Pruning using the LC Toolkit. ICASSP 2022: 3373-3377 - [c92]Magzhan Gabidolla, Arman Zharmagambetov, Miguel Á. Carreira-Perpiñán:
Improved Multiclass AdaBoost Using Sparse Oblique Decision Trees. IJCNN 2022: 1-8 - [c91]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán:
Sparse Oblique Decision Trees: A Tool to Interpret Natural Language Processing Datasets. IJCNN 2022: 1-8 - [c90]Magzhan Gabidolla, Miguel Á. Carreira-Perpiñán:
Optimal Interpretable Clustering Using Oblique Decision Trees. KDD 2022: 400-410 - [c89]Arman Zharmagambetov, Miguel Á. Carreira-Perpiñán:
Semi-Supervised Learning with Decision Trees: Graph Laplacian Tree Alternating Optimization. NeurIPS 2022 - 2021
- [c88]Miguel Á. Carreira-Perpiñán, Suryabhan Singh Hada:
Counterfactual Explanations for Oblique Decision Trees: Exact, Efficient Algorithms. AAAI 2021: 6903-6911 - [c87]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
LC: A Flexible, Extensible Open-Source Toolkit for Model Compression. CIKM 2021: 4504-4514 - [c86]Yerlan Idelbayev, Pavlo Molchanov, Maying Shen, Hongxu Yin, Miguel Á. Carreira-Perpiñán, José M. Álvarez:
Optimal Quantization Using Scaled Codebook. CVPR 2021: 12095-12104 - [c85]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
Neural Network Compression via Additive Combination of Reshaped, Low-Rank Matrices. DCC 2021: 243-252 - [c84]Arman Zharmagambetov, Magzhan Gabidolla, Miguel Á. Carreira-Perpiñán:
Softmax Tree: An Accurate, Fast Classifier When the Number of Classes Is Large. EMNLP (1) 2021: 10730-10745 - [c83]Arman Zharmagambetov, Miguel Á. Carreira-Perpiñán:
Learning a Tree of Neural Nets. ICASSP 2021: 3140-3144 - [c82]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
Optimal Selection of Matrix Shape and Decomposition Scheme for Neural Network Compression. ICASSP 2021: 3250-3254 - [c81]Arman Zharmagambetov, Miguel Á. Carreira-Perpiñán:
A Simple, Effective Way To Improve Neural Net Classification: Ensembling Unit Activations With A Sparse Oblique Decision Tree. ICIP 2021: 369-373 - [c80]Arman Zharmagambetov, Magzhan Gabidolla, Miguel Á. Carreira-Perpiñán:
Improved Multiclass Adaboost For Image Classification: The Role Of Tree Optimization. ICIP 2021: 424-428 - [c79]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
Beyond Flops In Low-Rank Compression Of Neural Networks: Optimizing Device-Specific Inference Runtime. ICIP 2021: 2843-2847 - [c78]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán, Arman Zharmagambetov:
Understanding And Manipulating Neural Net Features Using Sparse Oblique Classification Trees. ICIP 2021: 3707-3711 - [c77]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán:
Sampling The "Inverse Set" of a Neuron. ICIP 2021: 3712-3716 - [c76]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
An Empirical Comparison of Quantization, Pruning and Low-rank Neural Network Compression using the LC Toolkit. IJCNN 2021: 1-8 - [c75]Arman Zharmagambetov, Magzhan Gabidolla, Miguel Á. Carreira-Perpiñán:
Improved Boosted Regression Forests Through Non-Greedy Tree Optimization. IJCNN 2021: 1-8 - [c74]Arman Zharmagambetov, Suryabhan Singh Hada, Magzhan Gabidolla, Miguel Á. Carreira-Perpiñán:
Non-Greedy Algorithms for Decision Tree Optimization: An Experimental Comparison. IJCNN 2021: 1-8 - [c73]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
More General and Effective Model Compression via an Additive Combination of Compressions. ECML/PKDD (3) 2021: 233-248 - [c72]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán:
Exploring Counterfactual Explanations for Classification and Regression Trees. PKDD/ECML Workshops (1) 2021: 489-504 - [c71]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán:
Style Transfer by Rigid Alignment in Neural Net Feature Space. WACV 2021: 2575-2584 - [i25]Miguel Á. Carreira-Perpiñán, Suryabhan Singh Hada:
Counterfactual Explanations for Oblique Decision Trees: Exact, Efficient Algorithms. CoRR abs/2103.01096 (2021) - [i24]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán, Arman Zharmagambetov:
Sparse Oblique Decision Trees: A Tool to Understand and Manipulate Neural Net Features. CoRR abs/2104.02922 (2021) - [i23]Miguel Á. Carreira-Perpiñán, Yerlan Idelbayev:
Model compression as constrained optimization, with application to neural nets. Part V: combining compressions. CoRR abs/2107.04380 (2021) - 2020
- [j9]Daniel A. Winkler, Miguel Á. Carreira-Perpiñán, Alberto E. Cerpa:
OPTICS: OPTimizing Irrigation Control at Scale. ACM Trans. Sens. Networks 16(3): 22:1-22:38 (2020) - [c70]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
Low-Rank Compression of Neural Nets: Learning the Rank of Each Layer. CVPR 2020: 8046-8056 - [c69]Elad Eban, Yair Movshovitz-Attias, Hao Wu, Mark Sandler, Andrew Poon, Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
Structured Multi-Hashing for Model Compression. CVPR 2020: 11900-11909 - [c68]Miguel Á. Carreira-Perpiñán, Arman Zharmagambetov:
Ensembles of Bagged TAO Trees Consistently Improve over Random Forests, AdaBoost and Gradient Boosting. FODS 2020: 35-46 - [c67]Arman Zharmagambetov, Miguel Á. Carreira-Perpiñán:
Smaller, more accurate regression forests using tree alternating optimization. ICML 2020: 11398-11408 - [i22]Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
A flexible, extensible software framework for model compression based on the LC algorithm. CoRR abs/2005.07786 (2020)
2010 – 2019
- 2019
- [j8]Daniel A. Winkler, Robert Wang, François Blanchette, Miguel Á. Carreira-Perpiñán, Alberto E. Cerpa:
DICTUM: Distributed Irrigation aCtuation with Turf hUmidity Modeling. ACM Trans. Sens. Networks 15(4): 41:1-41:33 (2019) - [c66]Miguel Á. Carreira-Perpiñán, Mehdi Alizadeh:
Parmac: Distributed Optimisation Of Nested Functions, With Application To Learning Binary Autoencoders. SysML 2019 - [i21]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán:
Style Transfer by Rigid Alignment in Neural Net Feature Space. CoRR abs/1909.13690 (2019) - [i20]Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán:
Sampling the "Inverse Set" of a Neuron: An Approach to Understanding Neural Nets. CoRR abs/1910.04857 (2019) - [i19]Arman Zharmagambetov, Suryabhan Singh Hada, Miguel Á. Carreira-Perpiñán:
An Experimental Comparison of Old and New Decision Tree Algorithms. CoRR abs/1911.03054 (2019) - [i18]Elad Eban, Yair Movshovitz-Attias, Hao Wu, Mark Sandler, Andrew Poon, Yerlan Idelbayev, Miguel Á. Carreira-Perpiñán:
Structured Multi-Hashing for Model Compression. CoRR abs/1911.11177 (2019) - 2018
- [c65]Miguel Á. Carreira-Perpiñán, Yerlan Idelbayev:
"Learning-Compression" Algorithms for Neural Net Pruning. CVPR 2018: 8532-8541 - [c64]Daniel A. Winkler, Miguel Á. Carreira-Perpiñán, Alberto E. Cerpa:
Plug-and-play irrigation control at scale. IPSN 2018: 1-12 - [c63]Miguel Á. Carreira-Perpiñán, Pooya Tavallali:
Alternating optimization of decision trees, with application to learning sparse oblique trees. NeurIPS 2018: 1219-1229 - 2017
- [c62]Ramin Raziperchikolaei, Miguel Á. Carreira-Perpiñán:
Learning circulant support vector machines for fast image search. ICIP 2017: 385-389 - [c61]Ramin Raziperchikolaei, Miguel Á. Carreira-Perpiñán:
Learning supervised binary hashing: Optimization vs diversity. ICIP 2017: 3695-3699 - [c60]Max Vladymyrov, Miguel Á. Carreira-Perpiñán:
Fast, accurate spectral clustering using locally linear landmarks. IJCNN 2017: 3870-3879 - [i17]Miguel Á. Carreira-Perpiñán:
Model compression as constrained optimization, with application to neural nets. Part I: general framework. CoRR abs/1707.01209 (2017) - [i16]Miguel Á. Carreira-Perpiñán, Yerlan Idelbayev:
Model compression as constrained optimization, with application to neural nets. Part II: quantization. CoRR abs/1707.04319 (2017) - 2016
- [c59]Ramin Raziperchikolaei, Miguel Á. Carreira-Perpiñán:
Learning Independent, Diverse Binary Hash Functions: Pruning and Locality. ICDM 2016: 1173-1178 - [c58]Max Vladymyrov, Miguel Á. Carreira-Perpiñán:
The Variational Nystrom method for large-scale spectral problems. ICML 2016: 211-220 - [c57]Daniel A. Winkler, Robert Wang, François Blanchette, Miguel Á. Carreira-Perpiñán, Alberto E. Cerpa:
MAGIC: Model-Based Actuation for Ground Irrigation Control. IPSN 2016: 9:1-9:12 - [c56]Ramin Raziperchikolaei, Miguel Á. Carreira-Perpiñán:
Optimizing affinity-based binary hashing using auxiliary coordinates. NIPS 2016: 640-648 - [c55]Miguel Á. Carreira-Perpiñán, Ramin Raziperchikolaei:
An ensemble diversity approach to supervised binary hashing. NIPS 2016: 757-765 - [i15]Miguel Á. Carreira-Perpiñán, Ramin Raziperchikolaei:
An ensemble diversity approach to supervised binary hashing. CoRR abs/1602.01557 (2016) - [i14]Miguel Á. Carreira-Perpiñán, Mehdi Alizadeh:
ParMAC: distributed optimisation of nested functions, with application to learning binary autoencoders. CoRR abs/1605.09114 (2016) - 2015
- [c54]Miguel Á. Carreira-Perpiñán, Ramin Raziperchikolaei:
Hashing with binary autoencoders. CVPR 2015: 557-566 - [c53]Miguel Á. Carreira-Perpiñán, Max Vladymyrov:
A fast, universal algorithm to learn parametric nonlinear embeddings. NIPS 2015: 253-261 - [c52]Daniel A. Winkler, Robert Wang, François Blanchette, Miguel Á. Carreira-Perpiñán, Alberto E. Cerpa:
Poster: MICO: Model-Based Irrigation Control Optimization. SenSys 2015: 409-410 - [i13]Miguel Á. Carreira-Perpiñán, Ramin Raziperchikolaei:
Hashing with binary autoencoders. CoRR abs/1501.00756 (2015) - [i12]Ramin Raziperchikolaei, Miguel Á. Carreira-Perpiñán:
Learning hashing with affinity-based loss functions using auxiliary coordinates. CoRR abs/1501.05352 (2015) - [i11]Miguel Á. Carreira-Perpiñán:
A review of mean-shift algorithms for clustering. CoRR abs/1503.00687 (2015) - 2014
- [j7]Varick L. Erickson, Miguel Á. Carreira-Perpiñán, Alberto Cerpa:
Occupancy Modeling and Prediction for Building Energy Management. ACM Trans. Sens. Networks 10(3): 42:1-42:28 (2014) - [c51]Miguel Á. Carreira-Perpiñán, Weiran Wang:
LASS: A Simple Assignment Model with Laplacian Smoothing. AAAI 2014: 1715-1721 - [c50]Weiran Wang, Miguel Á. Carreira-Perpiñán:
The Role of Dimensionality Reduction in Classification. AAAI 2014: 2128-2134 - [c49]Miguel Á. Carreira-Perpiñán, Weiran Wang:
Distributed optimization of deeply nested systems. AISTATS 2014: 10-19 - [c48]Max Vladymyrov, Miguel Á. Carreira-Perpiñán:
Linear-time training of nonlinear low-dimensional embeddings. AISTATS 2014: 968-977 - [i10]Miguel Á. Carreira-Perpiñán, Weiran Wang:
LASS: a simple assignment model with Laplacian smoothing. CoRR abs/1405.5960 (2014) - [i9]Weiran Wang, Miguel Á. Carreira-Perpiñán:
The role of dimensionality reduction in linear classification. CoRR abs/1405.6444 (2014) - [i8]Weiran Wang, Miguel Á. Carreira-Perpiñán:
The Laplacian K-modes algorithm for clustering. CoRR abs/1406.3895 (2014) - [i7]Miguel Á. Carreira-Perpiñán:
An ADMM algorithm for solving a proximal bound-constrained quadratic program. CoRR abs/1412.8493 (2014) - 2013
- [j6]Ankur Kamthe, Miguel Á. Carreira-Perpiñán, Alberto Cerpa:
Improving wireless link simulation using multilevel markov models. ACM Trans. Sens. Networks 10(1): 17:1-17:28 (2013) - [c47]Max Vladymyrov, Miguel Á. Carreira-Perpiñán:
Entropic Affinities: Properties and Efficient Numerical Computation. ICML (3) 2013: 477-485 - [c46]Ankur Kamthe, Miguel Á. Carreira-Perpiñán, Alberto Cerpa:
Quick construction of data-driven models of the short-term behavior of wireless links. INFOCOM 2013: 160-164 - [c45]Max Vladymyrov, Miguel Á. Carreira-Perpiñán:
Locally Linear Landmarks for Large-Scale Manifold Learning. ECML/PKDD (3) 2013: 256-271 - [i6]Miguel Á. Carreira-Perpiñán, Weiran Wang:
The K-modes algorithm for clustering. CoRR abs/1304.6478 (2013) - [i5]Weiran Wang, Miguel Á. Carreira-Perpiñán:
Projection onto the probability simplex: An efficient algorithm with a simple proof, and an application. CoRR abs/1309.1541 (2013) - 2012
- [c44]Mohsen Farhadloo, Miguel Á. Carreira-Perpiñán:
Learning and adaptation of a tongue shape modelwith missing data. ICASSP 2012: 3981-3984 - [c43]Mohsen Farhadloo, Miguel Á. Carreira-Perpiñán:
Regularising an adaptation algorithm for tongue shape models. ICASSP 2012: 4481-4484 - [c42]Max Vladymyrov, Miguel Á. Carreira-Perpiñán:
Fast Training of Nonlinear Embedding Algorithms. ICML 2012 - [c41]Weiran Wang, Miguel Á. Carreira-Perpiñán:
Nonlinear low-dimensional regression using auxiliary coordinates. AISTATS 2012: 1295-1304 - [i4]Max Vladymyrov, Miguel Á. Carreira-Perpiñán:
Partial-Hessian Strategies for Fast Learning of Nonlinear Embeddings. CoRR abs/1206.4646 (2012) - [i3]Miguel Á. Carreira-Perpiñán, Weiran Wang:
Distributed optimization of deeply nested systems. CoRR abs/1212.5921 (2012) - 2011
- [c40]Miguel Á. Carreira-Perpiñán, Zhengdong Lu:
Manifold Learning and Missing Data Recovery through Unsupervised Regression. ICDM 2011: 1014-1019 - [c39]Ankur Kamthe, Miguel Á. Carreira-Perpiñán, Alberto Cerpa:
Adaptation of a Mixture of Multivariate Bernoulli Distributions. IJCAI 2011: 1336-1341 - [c38]Varick L. Erickson, Miguel Á. Carreira-Perpiñán, Alberto Cerpa:
OBSERVE: Occupancy-based system for efficient reduction of HVAC energy. IPSN 2011: 258-269 - [c37]Weiran Wang, Miguel Á. Carreira-Perpiñán, Zhengdong Lu:
A Denoising View of Matrix Completion. NIPS 2011: 334-342 - [c36]Ankur Kamthe, Varick L. Erickson, Miguel Á. Carreira-Perpiñán, Alberto Cerpa:
Enabling building energy auditing using adapted occupancy models. BuildSys@SenSys 2011: 31-36 - [i2]Miguel Á. Carreira-Perpiñán, Geoffrey J. Goodhill:
Generalised elastic nets. CoRR abs/1108.2840 (2011) - [i1]Miguel Á. Carreira-Perpiñán:
Reconstruction of sequential data with density models. CoRR abs/1109.3248 (2011) - 2010
- [c35]Weiran Wang, Miguel Á. Carreira-Perpiñán:
Manifold blurring mean shift algorithms for manifold denoising. CVPR 2010: 1759-1766 - [c34]Miguel Á. Carreira-Perpiñán, Zhengdong Lu:
Parametric dimensionality reduction by unsupervised regression. CVPR 2010: 1895-1902 - [c33]Chao Qin, Miguel Á. Carreira-Perpiñán:
Reconstructing the full tongue contour from EMA/X-ray microbeam. ICASSP 2010: 4190-4193 - [c32]Ling Xie, Miguel Á. Carreira-Perpiñán, Shawn D. Newsam:
Semi-supervised regression with temporal image sequences. ICIP 2010: 2637-2640 - [c31]Miguel Á. Carreira-Perpiñán:
The Elastic Embedding Algorithm for Dimensionality Reduction. ICML 2010: 167-174 - [c30]Chao Qin, Miguel Á. Carreira-Perpiñán:
Estimating missing data sequences in x-ray microbeam recordings. INTERSPEECH 2010: 1592-1595 - [c29]Chao Qin, Miguel Á. Carreira-Perpiñán, Mohsen Farhadloo:
Adaptation of a tongue shape model by local feature transformations. INTERSPEECH 2010: 1596-1599 - [c28]Chao Qin, Miguel Á. Carreira-Perpiñán:
Articulatory inversion of american English /turnr/ by conditional density modes. INTERSPEECH 2010: 1998-2001
2000 – 2009
- 2009
- [c27]Dominic W. Massaro, Miguel Á. Carreira-Perpiñán, David J. Merrill:
Optimizing Visual Feature Perception for an Automatic Wearable Speech Supplement in Face-to-Face Communication and Classroom Situations. HICSS 2009: 1-10 - [c26]Chao Qin, Miguel Á. Carreira-Perpiñán:
Adaptation of a predictive model of tongue shapes. INTERSPEECH 2009: 772-775 - [c25]Ankur Kamthe, Miguel Á. Carreira-Perpiñán, Alberto Cerpa:
M&M: multi-level Markov model for wireless link simulations. SenSys 2009: 57-70 - [c24]Ankur Kamthe, Miguel Á. Carreira-Perpiñán, Alberto Cerpa:
Wireless link simulations using multi-level Markov models. SenSys 2009: 391-392 - 2008
- [c23]Miguel Á. Carreira-Perpiñán:
Generalised blurring mean-shift algorithms for nonparametric clustering. CVPR 2008 - [c22]Miguel Á. Carreira-Perpiñán, Zhengdong Lu:
Dimensionality reduction by unsupervised regression. CVPR 2008 - [c21]Zhengdong Lu, Miguel Á. Carreira-Perpiñán:
Constrained spectral clustering through affinity propagation. CVPR 2008 - [c20]Umut Ozertem, Deniz Erdogmus, Miguel Á. Carreira-Perpiñán:
Density geodesics for similarity clustering. ICASSP 2008: 1977-1980 - [c19]Chao Qin, Miguel Á. Carreira-Perpiñán:
Trajectory inverse kinematics by nonlinear, nongaussian tracking. ICASSP 2008: 2057-2060 - [c18]Dominic W. Massaro, Miguel Á. Carreira-Perpiñán, David J. Merrill, Cass Sterling, Stephanie Bigler, Elise Piazza, Marcus Perlman:
IGlasses: an automatic wearable speech supplementin face-to-face communication and classroom situations. ICMI 2008: 197-198 - [c17]Chao Qin, Miguel Á. Carreira-Perpiñán:
Trajectory inverse kinematics by conditional density modes. ICRA 2008: 1979-1986 - [c16]Chao Qin, Miguel Á. Carreira-Perpiñán, Korin Richmond, Alan Wrench, Steve Renals:
Predicting tongue shapes from a few landmark locations. INTERSPEECH 2008: 2306-2309 - 2007
- [j5]Miguel Á. Carreira-Perpiñán:
Gaussian Mean-Shift Is an EM Algorithm. IEEE Trans. Pattern Anal. Mach. Intell. 29(5): 767-776 (2007)