Supervised LogEuclidean Metric Learning for Symmetric Positive Definite Matrices

Metric learning has been shown to be highly effective to improve the performance of nearest neighbor classification. In this paper, we address the problem of metric learning for Symmetric Positive Definite (SPD) matrices such as covariance matrices, which arise in many real-world applications. Naively using standard Mahalanobis metric learning methods under the Euclidean geometry for SPD matrices is not appropriate, because the difference of SPD matrices can be a non-SPD matrix and thus the obtained solution can be uninterpretable. To cope with this problem, we propose to use a properly parameterized LogEuclidean distance and optimize the metric with respect to kernel-target alignment, which is a supervised criterion for kernel learning. Then the resulting non-trivial optimization problem is solved by utilizing the Riemannian geometry. Finally, we experimentally demonstrate the usefulness of our LogEuclidean metric learning algorithm on real-world classification tasks for EEG signals and texture patches.

• Publication year

  2015

• Venue

  arXiv.org

• Publication date

  2015-02-11

• Fields of study

  Mathematics, Computer Science

• Identifiers
  arXiv 1502.03505
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Textures A Photographic Album For Artists And Designers

  2016cited by this paper
• From Manifold to Manifold: Geometry-Aware Dimensionality Reduction for SPD Matrices

  2014cited by this paper
• A Survey on Metric Learning for Feature Vectors and Structured Data

  2013cited by this paper
• Robust Structural Metric Learning

  2013cited by this paper
• Classification of covariance matrices using a Riemannian-based kernel for BCI applications

  2013influential reference
• A review of kernels on covariance matrices for BCI applications

  2013influential reference
• Manopt, a matlab toolbox for optimization on manifolds

  2013cited by this paper
• Kernel Methods on the Riemannian Manifold of Symmetric Positive Definite Matrices

  2013cited by this paper
• Neighborhood repulsed metric learning for kinship verification

  2012cited by this paper
• Geodesic Convexity and Covariance Estimation

  2012cited by this paper
• Multiclass Feature Selection With Kernel Gram-Matrix-Based Criteria

  2012cited by this paper
• Hamming Distance Metric Learning

  2012cited by this paper
• Algorithms for Learning Kernels Based on Centered Alignment

  2012cited by this paper
• Metric Learning for Large Scale Image Classification: Generalizing to New Classes at Near-Zero Cost

  2012cited by this paper
• A new metric on the manifold of kernel matrices with application to matrix geometric means

  2012cited by this paper
• ProDis-ContSHC: learning protein dissimilarity measures and hierarchical context coherently for protein-protein comparison in protein database retrieval

  2012cited by this paper
• A survey and comparison of contemporary algorithms for computing the matrix geometric mean

  2012cited by this paper
• Non-linear Metric Learning

  2012cited by this paper
• Regularizing Common Spatial Patterns to Improve BCI Designs: Unified Theory and New Algorithms

  2011cited by this paper
• Stochastic Gradient Descent on Riemannian Manifolds

  2011cited by this paper
• Discrete regression methods on the cone of positive-definite matrices

  2011cited by this paper
• Efficient similarity search for covariance matrices via the Jensen-Bregman LogDet Divergence

  2011cited by this paper
• Ground metric learning

  2011cited by this paper
• Learning a Distance Metric from a Network

  2011cited by this paper
• Wavelet kernel learning

  2011cited by this paper
• Learning Good Edit Similarities with Generalization Guarantees

  2011cited by this paper
• Positive definite matrices and the S-divergence

  2011cited by this paper
• Riemannian Geometry Applied to BCI Classification

  2010cited by this paper
• Non-Euclidean statistics for covariance matrices, with applications to diffusion tensor imaging

  2009cited by this paper
• Gabor Filters as Feature Images for Covariance Matrix on Texture Classification Problem

  2009cited by this paper
• Optimization algorithms on matrix manifolds

  2009cited by this paper
• Supervised Dictionary Learning

  2008cited by this paper
• Pedestrian Detection via Classification on Riemannian Manifolds

  2008cited by this paper
• Computing the Fréchet Derivative of the Matrix Exponential, with an Application to Condition Number Estimation

  2008cited by this paper
• Geometric Means in a Novel Vector Space Structure on Symmetric Positive-Definite Matrices

  2007cited by this paper
• Seperability of four-class motor imagery data using independent components analysis

  2006cited by this paper
• Region Covariance: A Fast Descriptor for Detection and Classification

  2006cited by this paper
• A Riemannian Framework for Tensor Computing

  2005cited by this paper
• A Differential Geometric Approach to the Geometric Mean of Symmetric Positive-Definite Matrices

  2005cited by this paper
• Principal Geodesic Analysis on Symmetric Spaces: Statistics of Diffusion Tensors

  2004cited by this paper
• Learning with Distance Substitution Kernels

  2004cited by this paper
• Motor imagery and direct brain-computer communication

  2001cited by this paper
• On Kernel-Target Alignment

  2001cited by this paper
• Nearest neighbor pattern classification

  1967cited by this paper
• Magnetic Resonance in Medicine 56:411–421 (2006) Log-Euclidean Metrics for Fast and Simple Calculus on Diffusion Tensors

  year unknowninfluential reference

• Automatic Classification of Sleep Stages from EEG Signals Using Riemannian Metrics and Transformer Networks

  2024influential citation
• Unsupervised manifold learning with polynomial mapping on symmetric positive definite matrices

  2022cites this paper
• Clinical BCI Challenge-WCCI2020: RIGOLETTO - RIemannian GeOmetry LEarning, applicaTion To cOnnectivity

  2021cites this paper
• Riemannian Geometry on Connectivity for Clinical BCI

  2021cites this paper
• Rough Set & Riemannian Covariance Matrix Theory for Mining the Multidimensionality of Artificial Consciousness

  2020cites this paper
• Geodesically-convex optimization for averaging partially observed covariance matrices

  2020cites this paper
• EEG Classification for MI-BCI using CSP with Averaging Covariance Matrices: An Experimental Study

  2019influential citation
• A survey on Laplacian eigenmaps based manifold learning methods

  2019cites this paper
• Learning Neural Bag-of-Matrix-Summarization with Riemannian Network

  2019cites this paper
• Towards Adaptive Classification using Riemannian Geometry approaches in Brain-Computer Interfaces

  2019cites this paper
• Second-Order Networks in PyTorch

  2019cites this paper
• Riemannian batch normalization for SPD neural networks

  2019cites this paper
• Riemannian Classification for SSVEP-Based BCI : Offline versus Online Implementations

  2018cites this paper
• Sparse Representations and Feature Learning for Image Set Classification and Correspondence Estimation

  2018cites this paper
• Riemannian geometry for EEG-based brain-computer interfaces; a primer and a review

  2017cites this paper
• Geometry-aware principal component analysis for symmetric positive definite matrices

  2017influential citation
• Robust Averaging of Covariances for EEG Recordings Classification in Motor Imagery Brain-Computer Interfaces

  2017cites this paper
• Riemannian Approaches in Brain-Computer Interfaces: A Review

  2017cites this paper
• Stochastic Dykstra Algorithms for Distance Metric Learning with Covariance Descriptors

  2017influential citation
• Dimensionality Reduction on SPD Manifolds: The Emergence of Geometry-Aware Methods

  2016influential citation
• Geometry-Aware Similarity Learning on SPD Manifolds for Visual Recognition

  2016cites this paper
• Discriminative Log-Euclidean Feature Learning for Sparse Representation-Based Recognition of Faces from Videos

  2016cites this paper
• Stochastic Dykstra Algorithms for Metric Learning on Positive Semi-Definite Cone

  2016cites this paper
• Stochastic Dykstra Algorithms for Metric Learning with Positive Definite Covariance Descriptors

  2016cites this paper
• Data augmentation in Riemannian space for Brain-Computer Interfaces

  2015cites this paper
• Averaging covariance matrices for EEG signal classification based on the CSP: An empirical study

  2015cites this paper
• Brain–Computer Interfaces Handbook Technological and Theoretical Advances

  year unknowncites this paper
• Analyse automatique de l’état de sommeil sur données EEG par utilisation de Transformers et de matrices de covariance

  year unknowncites this paper