Unsupervised spike extraction and classification based on locality preserving projection

Compared to vector based manifold learning methods, image based methods can reduce the complexity of algorithm, avoid the small sample size problem and give more spatial structural information of image. Two-dimensionality Locality Discriminat Projections (2D-LDP) is proposed, which an effective dimensionality reduction method and benefits from three parts, i.e., Locality Preserving Projections (LPP) algorithm, image based projection and discriminant analysis. In this paper, we apply 2DLPP to plant leaf classification. 2D-LDP can detect the intrinsic class-relationships between the leaf images by incorporating both class label information and neighborhood information. The Experimental results show that 2D-DLPP has better classifying performance than other methods.KeywordsLocality Preserving Projections (LPP)Two-dimensionality Locality Discriminat Projections (2D-LDP)Plant leaf Classification

With regard to the high dimensional and small sample facial feature, this paper introduced the Kernel and Canonical Correlation Analysis (CCA) into the Locality Preserving Projections (LPP) algorithm and proposed a new face recognition algorithm based on the Kernel Base Locality Preserving Canonical Correlation Analysis (KLPCCA) with the derivation process. According to this algorithm, first use CCA to extract nonlinear information of a face image, and then make a linear mapping through LPP, at last conduct the calculation process taking into consideration of kernel and other data obtained before, so as to carry out the face recognition more accurately and simply.

Dimensionality reduction technologies are very important for pattern representation and recognition. Among them, locality preserving projection (LPP) is a manifold dimensionality reduction scheme and has been successfully applied to face recognition. However, LPP is an unsupervised linear approach, its performance will degrade for classification tasks. Especially, when the dimension of input space is greater than the number of training data, singularity problem will occur and LPP cannot be implemented directly. To tackle the draw backs of LPP algorithm, this paper proposes a novel regularized LPP(RLPP) approach using supervised graph and regularization technique. The proposed RLPP method has been tested and evaluated with two public available databases, namely ORL and FERET databases. Experimental results show that the proposed RLPP approach surpasses Laplacianface and Direct-LPP (DLPP) methods.

Locality-preserving projection (LPP) is a promising manifold-based dimensionality reduction and linear feature extraction method for face recognition. However, there exist two main issues in traditional LPP algorithm. LPP does not utilize the class label information at the training stage and its performance will be affected for classification tasks. In addition, LPP often suffers from small sample size (3S) problem, which occurs when the dimension of input pattern space is greater than the number of training samples. Under this situation, LPP fails to work. To overcome these two limitations, this paper presents a novel supervised regularization LPP (SRLPP) approach based on a supervised graph and a new regularization strategy. It theoretically proves that regularization matrix [Formula: see text] approaches to the original one as the regularized parameter tends to zero. The proposed SRLPP method is subsequently applied to face recognition. The experiments are conducted on two publicly available face databases, namely ORL database and FERET database. Compared with some existing LDA-based and LPP-based linear feature extraction approaches, experimental results show that our SRLPP approach gives superior performance.

In flotation production, the surface texture of the flotation froth is an important visual information reflecting the ore grade, which is closely related to the flotation conditions and directly reflects the mineralization degree (grade level) of the froth layer. Aiming at the problem that the texture feature extraction of the existing flotation froth image does not consider the froth surface particle, a new local texture feature extraction method based on PCA (Principle Component Analysis) algorithm is proposed and a new texture feature granularity is defined. Firstly, the froth image is segmented, the region of interest is extracted and the particle factor is defined. Then the texture feature is extracted by PCA algorithm and the correlation between particle size and zinc concentrate taste is analyzed. The experimental results verify the algorithm effectiveness and indicating that it can more accurately regulate the position of minerals and guide flotation production.

In the last two decades, developing countries are facing heavy increase in diabetes among their population that is leading to other severe diseases. Hence, there is a great need to develop some effective prediction methods to prevent diabetes. In this paper an attempt has been made to develop Firefly-BAT (FFBAT) optimized Rule Based Fuzzy Logic (RBFL) prediction algorithm for diabetes. The algorithm has two main steps. First, Locality Preserving Projections (LPP) algorithm is used for feature reduction and then classification of diabetes is done by means of RBFL classifier. LPP algorithm has been used to identify the related attributes and then the fuzzy rules are produced from RBFL. The rules are optimized using FFBAT algorithm. Next, the fuzzy system is designed with the help of optimized fuzzy rules and membership functions that will classify the diabetes data. FFBAT is the optimization algorithm which combines the features of BAT and Firefly (FF) optimization techniques. The experiment analysis shows that the RBFL-FFBAT algorithm outperforms the existing approaches.

Face recognition using Elasticfaces

Classical feature extraction techniques, like PCA and LDA, do not deal properly with multimodal problems. Such techniques create projections that do not preserve the multimodal structure of the original data distribution. Locality Preserving Projection (LPP) is a feature extraction technique which looks for a transformation matrix that minimizes the changes into the structure of the data after the transformation. This local structure is captured by the affinity matrix. However, there many ways to calculate this affinity matrix. The main aim of this paper is to evaluate the influence of different affinity matrices over the LPP accuracy. The experiments showed that the correct choice of the affinity matrix can lead to a performance gain. Among the analyzed affinity matrices, Local Scaling and Nearest Neighbor reached the best results.

Exploring dimensionality reduction of EEG features in motor imagery task classification

Moving objects detection in video is essential for many computer vision applications and it is considered as a challenging research issue due to dynamic changes in object size, shape, complex background and illumination changes. In this research article a novel method to detect moving objects in video is proposed grounded on Locality Preserving Projections (LPP) and also thorough analysis of variations in neighbor mode and weight mode for constructing adjacency graph is given. LPP is an unsupervised subspace learning approach used for dimensionality reduction, which preserves the neighborhood structure of dataset while detection, which is vital in further steps accurate tracking and recognition of objects. The proposed method is tested on standard datasets with complex environments and experimental results obtained with variations in neighborhood modes and weight modes are encouraging.

The human face exhibits a high level of complexity when it is regarded as a multidimensional visual model, leading to face recognition systems that require difficult and extensive computations for coding and decoding the face images. A well-established approach in this regard is based on using principle component analysis (PCA) for both feature extraction and face recognition, known as the eigenface approach. This technique, despite a good recognition rate, suffers from the disadvantage of high computation cost due to the complexity of the PCA algorithm. In this paper, we use a geometrical approximated PCA (gaPCA) algorithm for computing the eigenfaces for three different datasets. The face recognition task is performed using a similarity score based on the inverse Euclidean distance for the first two datasets and using a nerual network in the third case. All the results are compared to the case where standard PCA is used. These accuracy results show that gaPCA represents a viable alternative to the classical statistical approach for computing the principal components.

Graph convolutional networks (GCN) excel in rapidly mining semantic information from data with superior capabilities, enabling deep feature extraction and classification for hyperspectral images (HSI). However, supervised classification methods heavily rely on many high-quality labeled samples, which is often a significant limitation in real-world applications. Unsupervised learning, with its unique ability to automatically discover potential structures in data without labeled samples, provides a new approach to address this issue. Thus, this paper proposes an unsupervised dual-masked graph autoencoder feature extraction network (UDMG) for HSI classification. This method first utilizes a masked autoencoder to mask and reconstruct the edges and node features of the graph structure. Meanwhile, the encoder combines GCN and Transformer to capture local and global features in HSI. To validate the performance of UDMG, we employ various evaluation methods and conduct experiments on three publicly available hyperspectral datasets. The results demonstrate that UDMG exhibits superior classification performance compared to existing state-of-the-art unsupervised feature extraction methods.

The current damage is the most stubborn and difficult fault of high-power motor bearings because its vibration characteristics are easily confused with those of ordinary bearing mechanical faults. If it is discriminated as an ordinary mechanical fault without electrical insulation protection, the current damage of bearing shafts will still repeatedly appear. Aiming at the problem that it is difficult to identify the bearing current damage fault under variable working conditions, a bearing shaft current damage identification method based on multiscale feature label propagation and manifold metric transfer (MFLP-MMT) is proposed. Firstly, the multiscale sub-band signal is obtained by wavelet packet decomposition, and the multiscale sub-band fuzzy entropy is obtained by calculating its fuzzy entropy. Then, according to the extracted features, a neighbor graph is constructed on the source domain of the known fault label to obtain the pseudo label of the target domain sample, and the source domain label information is gradually diffused by way of the graph label propagation. The multiscale sub-band fuzzy entropy of the sample is mapped to the low-dimensional manifold space by locality preserving projections (LPP), and the source domain samples close to the target domain are given higher weights by cross-domain density ratio estimation to solve the problem of domain offset. Combined with the label samples of the target domain in label propagation, the manifold distance metric is learned to minimize the intra-class distance and maximize the inter-class distance in the domain and eliminate the overlapping phenomenon in the domain. By increasing the range of label propagation after each iteration, the label propagation error of the leading graph is gradually reduced, and unsupervised metric transfer learning is realized. The experimental results show that the new method is superior to the semi-supervised transfer learning method in fault identification ability; the highest fault identification accuracy can reach 100% and it has a good robustness.

Optimal locality preserving projection for face recognition

Automatic analysis methods of electrocardiograms (ECGs) usually required large-scale annotated training data, but the annotation process is extremely time-consuming. While semi-supervised learning can leverage unlabeled data, its performance depends heavily on the quality of the initial labeled subset. Active learning has been used to identify the most informative samples for annotation, but conventional approaches face three critical limitations: (1) dependency on manual intervention for iterative query design, (2) prohibitive computational costs during sample selection, and (3) limited compatibility with semi-supervised learning frameworks. To address these limitations, we proposed an Unsupervised Active Feature-selective Semi-Supervised Learning (UAFSSL) framework for ECG analysis, including an unsupervised feature selection-based active learning module and a semi-supervised learning module. UAFSSL captures latent data distributions via unsupervised feature extraction, selects diverse and representative samples using pseudo-label clustering, and integrates seamlessly with semi-supervised learning to eliminate human intervention. We validated our algorithm on an ECG waveform segmentation task and an atrial fibrillation detection task. In the waveform segmentation task, our method improved the F1-score for P-wave delineation by 2.4% compared to random sampling, using only 5% of labeled samples. For the atrial fibrillation detection task, we evaluated our method on both the AFDB and a 24-hour dataset collected from 500 atrial fibrillation patients. Using only 200 labeled samples for model training, our method achieved AUC improvements of 2.5% and 2.2% over random sampling in five-fold cross validation. This is the first study to integrate unsupervised active learning with semi-supervised learning for automatic ECG analysis, offering a robust, automated solution to reduce annotation costs while enhancing clinical applicability.