Disturbance Grassmann Kernels for Subspace-Based Learning

  title={Disturbance Grassmann Kernels for Subspace-Based Learning},
  author={Junyuan Hong and Huanhuan Chen and Feng Lin},
  journal={Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery \& Data Mining},
  • Junyuan Hong, Huanhuan Chen, Feng Lin
  • Published 10 February 2018
  • Computer Science
  • Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining
In this paper, we focus on subspace-based learning problems, where data elements are linear subspaces instead of vectors. To handle this kind of data, Grassmann kernels were proposed to measure the space structure and used with classifiers, e.g., Support Vector Machines (SVMs). However, the existing discriminative algorithms mostly ignore the instability of subspaces, which would cause the classifiers to be misled by disturbed instances. Thus we propose considering all potential disturbances of… 

Figures and Tables from this paper

Variant Grassmann Manifolds
This article introduces infinitely many variant Grassmann manifolds (VGM) subject to a known distribution, then represents each action video as different Grassmann points leading to augmented representations, so the manifold distribution can be adaptively determined, balancing discrimination and representation.
Projection Metric Learning of Updated-Subspaces for Radar Target Classification
This study proposes a radar based UAV classification framework that integrates signal detection by a subspace-updating method with projection metric learning on a Grassmann manifold to solve the classification problem of small radar targets.
Semantically-Aware Statistical Metrics via Weighting Kernels
This paper investigates the collisions that can arise with distance metrics, and proposes a solution based on kernels for augmenting distance metrics and summary statistics, thus avoiding collisions and highlighting semantically-relevant phenomena.
Multicenter Computer-Aided Diagnosis for Lymph Nodes Using Unsupervised Domain-Adaptation Networks Based on Cross-Domain Confounding Representations
This paper extracts domain-invariant features based on a cross- domain confounding representation and proposes a cycle-consistency learning framework to encourage the network to preserve class-conditioning information through cross-domain image translations.
Learning Model-Based Privacy Protection under Budget Constraints
A novel Learning-to-Protect algorithm that automatically learns a model-based protector from a set of nonprivate learning tasks that can achieve a superior utility with a given privacy constraint and generalize well to new private datasets distributed differently as compared to the hand-designed competitors.
Human Activity Recognition based on Dynamic Spatio-Temporal Relations
In this paper, an activity is divided into several successive actions represented by spatio temporal patterns, and the evolution of these actions are captured by a sequential model, which will enable long videos of human activity to be better recognized.
Learning Dynamic Spatio-Temporal Relations for Human Activity Recognition
A method for human activity recognition is developed that will enable long videos of human activity to be better recognized and a hierarchical decomposition of the human body is introduced to obtain a discriminative representation for a single action.
Recent Trends in Food Intake Monitoring using Wearable Sensors
This research work delineates the automatic food intake monitoring paradigm and presents a survey of research studies, which provides brief comparison of these studies along with shortcomings based upon experimentation results conducted under these studies.


Extended Grassmann Kernels for Subspace-Based Learning
The relationship between Grassmann kernels and probabilistic similarity measures is analyzed and it is shown that the KL distance in the limit yields the Projection kernel on the Grassmann manifold, whereas the Bhattacharyya kernel becomes trivial in thelimit and is suboptimal for subspace-based problems.
Grassmann discriminant analysis: a unifying view on subspace-based learning
This paper proposes a discriminant learning framework for problems in which data consist of linear subspaces instead of vectors, and treats each sub-space as a point in the Grassmann space, and performs feature extraction and classification in the same space.
Dictionary Learning and Sparse Coding on Grassmann Manifolds: An Extrinsic Solution
To handle non-linearity in data, this paper proposes to embed Grassmann manifolds into the space of symmetric matrices by an isometric mapping, which enables it to devise a closed-form solution for updating a Grassmann dictionary, atom by atom.
Projection Metric Learning on Grassmann Manifold with Application to Video based Face Recognition
This work proposes a novel method to learn the Projection Metric directly on Grassmann manifold rather than in Hilbert space, which can be regarded as performing a geometry-aware dimensionality reduction from the original Grassmann manifolds to a lower-dimensional, more discriminative Grassman manifold where more favorable classification can be achieved.
Learning over Sets using Kernel Principal Angles
A new positive definite kernel f(A,B) defined over pairs of matrices A,B is derived based on the concept of principal angles between two linear subspaces and it is shown that the principal angles can be recovered using only inner-products between pairs of column vectors of the input matrices thereby allowing the original column vectors to be mapped onto arbitrarily high-dimensional feature spaces.
Low Rank Representation on Grassmann Manifolds
The novelty of this paper is to generalize LRR on Euclidean space into the LRR model on Grassmann manifold, a new method that has many applications in computer vision tasks.
Graph embedding discriminant analysis on Grassmannian manifolds for improved image set matching
It is shown that by introducing within-class and between-class similarity graphs to characterise intra-class compactness and inter-class separability, the geometrical structure of data can be exploited.
Statistical Computations on Grassmann and Stiefel Manifolds for Image and Video-Based Recognition
This paper discusses how commonly used parametric models for videos and image sets can be described using the unified framework of Grassmann and Stiefel manifolds, and derives statistical modeling of inter and intraclass variations that respect the geometry of the space.
Product Grassmann Manifold Representation and Its LRR Models
This paper proposes a novel representation, namely Product Grassmann Manifold (PGM), to represent complex high dimensional data with multiple varying factors and obtains superior accuracy compared with the clustering methods on manifolds or conventional Euclidean spaces.
Grassmann Registration Manifolds for Face Recognition
A novel method combining image perturbation and the geometry of manifolds is presented, which forms a tangent space from a set of perturbed images and observes that the tangentspace admits a vector space structure.