Linear discriminant analysis

About: Linear discriminant analysis is a research topic. Over the lifetime, 18361 publications have been published within this topic receiving 603195 citations. The topic is also known as: Linear discriminant analysis & LDA.

Human Carrying Status in Visual Surveillance

Abstract: A person’s gait changes when he or she is carrying an object such as a bag, suitcase or rucksack. As a result, human identification and tracking are made more difficult because the averaged gait image is too simple to represent the carrying status. Therefore, in this paper we first introduce a set of Gabor based human gait appearance models, because Gabor functions are similar to the receptive field profiles in the mammalian cortical simple cells. The very high dimensionality of the feature space makes training difficult. In order to solve this problem we propose a general tensor discriminant analysis (GTDA), which seamlessly incorporates the object (Gabor based human gait appearance model) structure information as a natural constraint. GTDA differs from the previous tensor based discriminant analysis methods in that the training converges. Existing methods fail to converge in the training stage. This makes them unsuitable for practical tasks. Experiments are carried out on the USF baseline data set to recognize a human’s ID from the gait silhouette. The proposed Gabor gait incorporated with GTDA is demonstrated to significantly outperform the existing appearance-based methods.

Particle Swarm Optimization-Based Hyperspectral Dimensionality Reduction for Urban Land Cover Classification

Abstract: A particle swarm optimization (PSO)-based dimensionality reduction approach is proposed to use a simple searching criterion function, called minimum estimated abundance covariance (MEAC), requiring class signatures only. It has low computational cost, and the selected bands are independent of the detector or classifiers used in the following data analysis step. With such an efficient criterion, PSO can find a global optimal solution much more efficiently, compared with other frequently used searching strategies. Its performance is evaluated by support vector machine (SVM)-based classification for urban land cover mapping. In our experiments, SVM classification accuracy using PSO-selected bands is greatly higher than using all of the original bands or dimensionality-reduced data from principal component analysis (PCA) or linear discriminant analysis (LDA). In addition, the improvement on SVM accuracy can bring out even more significant improvement in classifier fusion.

Performance analysis of LDA, QDA and KNN algorithms in left-right limb movement classification from EEG data

Abstract: Brain Computer Interface (BCI) improve the lifestyle of the normal people by enhancing their performance levels. It also provides a way of communication for the disabled people with their surrounding who are otherwise unable to physically communicate. BCI can be used to control computers, robots, prosthetic devices and other assistive technologies for rehabilitation. The dataset used for this study has been obtained from the BCI competition II 2003 databank provided by the University of Technology, Graz. After pre-processing of the signals from their electrodes (C3 & C4), the wavelet coefficients, Power Spectral Density of the alpha and the central beta band and the average power of the respective bands have been employed as features for classification. In one of the approaches we fed all the extracted features individually and in the other approach we considered all features together and submitted them to LDA, QDA and KNN algorithms distinctly to classify left and right limb movement. The aim of this study is to analyze the performance of linear discriminant analysis (LDA), quadratic discriminant analysis (QDA) and K-nearest neighbor (KNN) algorithms in differentiating the raw EEG data obtained, into their associative movement, namely, left-right movement. Also the importance of the feature vectors selected is highlighted in this study. The total set to feature vector comprising all the features (i.e., wavelet coefficients, PSD and average band power estimate) performed better with the classifiers without much deviation in the classification accuracy, i.e., 80%, 80% and 75.71% with LDA, QDA and KNN respectively. Wavelet coefficients performed best with QDA classifier with an accuracy of 80%. PSD vector resulted in superior performance of 81.43% with both QDA and KNN. Average band power estimate vector showed highest accuracy of 84.29% with KNN algorithm. Our approach presented in this paper is quite simple, easy to execute and is validated robustly with a large dataset.

So you think you can PLS-DA?

Abstract: Partial Least-Squares Discriminant Analysis (PLS-DA) is a popular machine learning tool that is gaining increasing attention as a useful feature selector and classifier. In an effort to understand its strengths and weaknesses, we performed a series of experiments with synthetic data and compared its performance to its close relative from which it was initially invented, namely Principal Component Analysis (PCA). We demonstrate that even though PCA ignores the information regarding the class labels of the samples, this unsupervised tool can be remarkably effective as a feature selector. In some cases, it outperforms PLS-DA, which is made aware of the class labels in its input. Our experiments range from looking at the signal-to-noise ratio in the feature selection task, to considering many practical distributions and models encountered when analyzing bioinformatics and clinical data. Other methods were also evaluated. Finally, we analyzed an interesting data set from 396 vaginal microbiome samples where the ground truth for the feature selection was available. All the 3D figures shown in this paper as well as the supplementary ones can be viewed interactively at http://biorg.cs.fiu.edu/plsda Our results highlighted the strengths and weaknesses of PLS-DA in comparison with PCA for different underlying data models.