Search Results (7)

At present, the method of using rule set partitioning technology to assist in constructing multiple decision trees for packet classification has been widely recognized. Rule set partitioning demonstrates a unique symmetry-breaking mechanism, systematically transforming the initial overlapping rule space into a more structured and balanced configuration. By separating overlapping rules in the initial stage, this method significantly reduces rule replication within trees, thereby improving the algorithm’s classification performance. The asymmetric characteristics of this partitioning process are particularly noteworthy: through the strategic disruption of the initial rule set’s symmetric distribution, it creates asymmetric subspaces with enhanced computational efficiency. However, existing research lacks standardized metrics for evaluating the effectiveness of rule set partitioning schemes. The purpose of this paper is to investigate the impact of partitioning on algorithm performance. Based on community structure theory, we construct a weighted graph model for rule sets and propose a disassortative modularity metric to evaluate the effectiveness of rule set partitioning. This metric not only examines intra-community connections but also emphasizes the asymmetric connections between communities. By quantifying these structural features, it provides a novel perspective on rule set partitioning strategies. The experimental results demonstrate a significant positive correlation between disassortative modularity and classification throughput. This metric offers valuable guidance for packet classification partitioning techniques, highlighting the practical significance of symmetry and asymmetry in algorithm design. Full article

In this paper, we focus on a challenging, but interesting, task in speech emotion recognition (SER), i.e., cross-corpus SER. Unlike conventional SER, a feature distribution mismatch may exist between the labeled source (training) and target (testing) speech samples in cross-corpus SER because they come from different speech emotion corpora, which degrades the performance of most well-performing SER methods. To address this issue, we propose a novel transfer subspace learning method called multiple distribution-adapted regression (MDAR) to bridge the gap between speech samples from different corpora. Specifically, MDAR aims to learn a projection matrix to build the relationship between the source speech features and emotion labels. A novel regularization term called multiple distribution adaption (MDA), consisting of a marginal and two conditional distribution-adapted operations, is designed to collaboratively enable such a discriminative projection matrix to be applicable to the target speech samples, regardless of speech corpus variance. Consequently, by resorting to the learned projection matrix, we are able to predict the emotion labels of target speech samples when only the source label information is given. To evaluate the proposed MDAR method, extensive cross-corpus SER tasks based on three different speech emotion corpora, i.e., EmoDB, eNTERFACE, and CASIA, were designed. Experimental results showed that the proposed MDAR outperformed most recent state-of-the-art transfer subspace learning methods and even performed better than several well-performing deep transfer learning methods in dealing with cross-corpus SER tasks. Full article

A brain-computer interface (BCI) translates a user’s thoughts such as motor imagery (MI) into the control of external devices. However, some people, who are defined as BCI illiteracy, cannot control BCI effectively. The main characteristics of BCI illiterate subjects are low classification rates and poor repeatability. To address the problem of MI-BCI illiteracy, we propose a distribution adaptation method based on multi-kernel learning to make the distribution of features between the source domain and target domain become even closer to each other, while the divisibility of categories is maximized. Inspired by the kernel trick, we adopted a multiple-kernel-based extreme learning machine to train the labeled source-domain data to find a new high-dimensional subspace that maximizes data divisibility, and then use multiple-kernel-based maximum mean discrepancy to conduct distribution adaptation to eliminate the difference in feature distribution between domains in the new subspace. In light of the high dimension of features of MI-BCI illiteracy, random forest, which can effectively handle high-dimensional features without additional cross-validation, was employed as a classifier. The proposed method was validated on an open dataset. The experimental results show that that the method we proposed suits MI-BCI illiteracy and can reduce the inter-domain differences, resulting in a reduction in the performance degradation of both cross-subjects and cross-sessions. Full article

Microwave remote sensing is one of the main approaches to glacier monitoring. This paper provides a comparative analysis of how different types of radar information differ in identifying debris-covered alpine glaciers using machine learning algorithms. Based on Sentinel-1A data, three data suites were designed: A backscattering coefficient (BC)-based data suite, a polarization decomposition parameter (PDP)-based data suite, and an interference coherence coefficient (ICC)-based data suite. Four glaciers with very different orientations in different climatic zones of the Tibetan Plateau were selected and classified using an integrated machine learning classification approach. The results showed that: (1) The boosted trees and subspace k-nearest neighbor algorithms were optimal and robust; and (2) the PDP suite (63.41–99.57%) and BC suite (55.85–99.94%) both had good recognition accuracy for all glaciers; notably, the PDP suite exhibited better rock and debris recognition accuracy. We also analyzed the influence of the distribution of glacier surface aspect on the classification accuracy and found that the more asymmetric it was about the sensor orbital plane, the more difficult it was for the BC and PDP suites to recognize the glacier, and a large slope could further reduce the accuracy. Our results suggested that during the inventory or classification of large-scale debris-covered alpine glaciers, priority should be given to polarization decomposition features and elevation information, and it is best to divide the glaciers into multiple subregions based on the spatial relationship between glacier surface aspect and radar beams. Full article

This work proposes a new computational framework for learning a structured generative model for real-world datasets. In particular, we propose to learn a Closed-loop Transcriptionbetween a multi-class, multi-dimensional data distribution and a Linear discriminative representation (CTRL) in the feature space that consists of multiple independent multi-dimensional linear subspaces. In particular, we argue that the optimal encoding and decoding mappings sought can be formulated as a two-player minimax game between the encoder and decoderfor the learned representation. A natural utility function for this game is the so-called rate reduction, a simple information-theoretic measure for distances between mixtures of subspace-like Gaussians in the feature space. Our formulation draws inspiration from closed-loop error feedback from control systems and avoids expensive evaluating and minimizing of approximated distances between arbitrary distributions in either the data space or the feature space. To a large extent, this new formulation unifies the concepts and benefits of Auto-Encoding and GAN and naturally extends them to the settings of learning a both discriminative and generative representation for multi-class and multi-dimensional real-world data. Our extensive experiments on many benchmark imagery datasets demonstrate tremendous potential of this new closed-loop formulation: under fair comparison, visual quality of the learned decoder and classification performance of the encoder is competitive and arguably better than existing methods based on GAN, VAE, or a combination of both. Unlike existing generative models, the so-learned features of the multiple classes are structured instead of hidden: different classes are explicitly mapped onto corresponding independent principal subspaces in the feature space, and diverse visual attributes within each class are modeled by the independent principal components within each subspace. Full article

We study certain physically-relevant subgeometries of binary symplectic polar spaces $W(2N-1,2)$ of small rank N, when the points of these spaces canonically encode N-qubit observables. Key characteristics of a subspace of such a space $W(2N-1,2)$ are: the number of its negative lines, the distribution of types of observables, the character of the geometric hyperplane the subspace shares with the distinguished (non-singular) quadric of $W(2N-1,2)$ and the structure of its Veldkamp space. In particular, we classify and count polar subspaces of $W(2N-1,2)$ whose rank is $N-1$. $W(3,2)$ features three negative lines of the same type and its $W(1,2)$’s are of five different types. $W(5,2)$ is endowed with 90 negative lines of two types and its $W(3,2)$’s split into 13 types. A total of 279 out of 480 $W(3,2)$’s with three negative lines are composite, i.e., they all originate from the two-qubit $W(3,2)$. Given a three-qubit $W(3,2)$ and any of its geometric hyperplanes, there are three other $W(3,2)$’s possessing the same hyperplane. The same holds if a geometric hyperplane is replaced by a ‘planar’ tricentric triad. A hyperbolic quadric of $W(5,2)$ is found to host particular sets of seven $W(3,2)$’s, each of them being uniquely tied to a Conwell heptad with respect to the quadric. There is also a particular type of $W(3,2)$’s, a representative of which features a point each line through which is negative. Finally, $W(7,2)$ is found to possess 1908 negative lines of five types and its $W(5,2)$’s fall into as many as 29 types. A total of 1524 out of 1560 $W(5,2)$’s with 90 negative lines originate from the three-qubit $W(5,2)$. Remarkably, the difference in the number of negative lines for any two distinct types of four-qubit $W(5,2)$’s is a multiple of four. Full article

Vehicle detection is a key component of environmental sensing systems for Intelligent Vehicles (IVs). The traditional shallow model and offline learning-based vehicle detection method are not able to satisfy the real-world challenges of environmental complexity and scene dynamics. Focusing on these problems, this work proposes a vehicle detection algorithm based on a multiple feature subspace distribution deep model with online transfer learning. Based on the multiple feature subspace distribution hypothesis, a deep model is established in which multiple Restricted Boltzmann Machines (RBMs) construct the lower layers and a Deep Belief Network (DBN) composes the superstructure. For this deep model, an unsupervised feature extraction method is applied, which is based on sparse constraints. Then, a transfer learning method with online sample generation is proposed based on the deep model. Finally, the entire classifier is retrained online with supervised learning. The experiment is actuated using the KITTI road image datasets. The performance of the proposed method is compared with many state-of-the-art methods and it is demonstrated that the proposed deep transfer learning-based algorithm outperformed existing state-of-the-art methods. Full article