Neighboring Samples Research Articles

High-Order Correlation Preserved Incomplete Multi-View Subspace Clustering.

Incomplete multi-view clustering aims to exploit the information of multiple incomplete views to partition data into their clusters. Existing methods only utilize the pair-wise sample correlation and pair-wise view correlation to improve the clustering performance but neglect the high-order correlation of samples and that of views. To address this issue, we propose a high-order correlation preserved incomplete multi-view subspace clustering (HCP-IMSC) method which effectively recovers the missing views of samples and the subspace structure of incomplete multi-view data. Specifically, multiple affinity matrices constructed from the incomplete multi-view data are treated as a third-order low rank tensor with a tensor factorization regularization which preserves the high-order view correlation and sample correlation. Then, a unified affinity matrix can be obtained by fusing the view-specific affinity matrices in a self-weighted manner. A hypergraph is further constructed from the unified affinity matrix to preserve the high-order geometrical structure of the data with incomplete views. Then, the samples with missing views are restricted to be reconstructed by their neighbor samples under the hypergraph-induced hyper-Laplacian regularization. Furthermore, the learning of view-specific affinity matrices as well as the unified one, tensor factorization, and hyper-Laplacian regularization are integrated into a unified optimization framework. An iterative algorithm is designed to solve the resultant model. Experimental results on various benchmark datasets indicate the superiority of the proposed method. The code is implemented by using MATLAB R2018a and MindSpore library: https://github.com/ChangTang/HCP-IMSC.

Discriminative Spectral–Spatial Feature Extraction-Based Band Selection for Hyperspectral Image Classification

Recently, some spectral–spatial band selection (BS) strategies have become hugely popular as they fuse the spectral information of the pixels and the spatial relationship with the neighboring pixels to enhance the performance of classification methods. However, being unsupervised in nature, these methods do not utilize the class information of the training samples which could substantially empower the capabilities of such spectral–spatial BS methods. To circumvent this limitation, a supervised spectral–spatial BS method based on component loadings obtained from the principal components of spectral–spatial principal component analysis (PCA) and using a novel super-pixel based graph Laplacian embedding is proposed. The methodology attempts to unify the two strategies of dimensionality reduction, i.e., BS and feature extraction (FE), so that the benefits from both of them can be combined. The importance of each band is estimated in terms of its component loadings along the principal components which are estimated from a unified objective function consisting of three terms: data fidelity, classification error term, and spatial prior. Additionally, the spatial relationship among the neighboring samples is characterized using a novel superpixel-based graph model. An objective comparison of the proposed approach with several widely used, state-of-the-art BS methods demonstrates a significant improvement in the classification accuracy.

SODAR: Segmenting Objects by Dynamically Aggregating Neighboring Mask Representations.

Recent state-of-the-art one-stage instance segmentation model SOLO divides the input image into a grid and directly predicts per grid cell object masks with fully-convolutional networks, yielding comparably good performance as traditional two-stage Mask R-CNN yet enjoying much simpler architecture and higher efficiency. We observe SOLO generates similar masks for an object at nearby grid cells, and these neighboring predictions can complement each other as some may better segment certain object part, most of which are however directly discarded by non-maximum-suppression. Motivated by the observed gap, we develop a novel learning-based aggregation method that improves upon SOLO by leveraging the rich neighboring information while maintaining the architectural efficiency. The resulting model is named SODAR. Unlike the original per grid cell object masks, SODAR is implicitly supervised to learn mask representations that encode geometric structure of nearby objects and complement adjacent representations with context. The aggregation method further includes two novel designs: 1) a mask interpolation mechanism that enables the model to generate much fewer mask representations by sharing neighboring representations among nearby grid cells, and thus saves computation and memory; 2) a deformable neighbour sampling mechanism that allows the model to adaptively adjust neighbor sampling locations thus gathering mask representations with more relevant context and achieving higher performance. SODAR significantly improves the instance segmentation performance, e.g., it outperforms a SOLO model with ResNet-101 backbone by 2.2 AP on COCO test set, with only about 3% additional computation. We further show consistent performance gain with the SOLOv2 model.

Multilevel Dual-Direction Modifying Variational Autoencoders for Hyperspectral Feature Extraction

Hyperspectral images (HSIs) provide abundant high-quality spectral information through an immense number of spectral channels, which can be used to classify on-ground objects for Earth observation accurately. However, these highly correlated channels and complex informative features always limit the application of HSIs. In this letter, we propose a multilevel dual-direction modifying variational autoencoder (MD²MVAE) for hyperspectral feature extraction. Its architecture is inspired by the spectral–spatial coherence in HSIs. Our motivation is to modify spectral sequential features by spatial sequential features in a multilevel dual-direction network. The dual-direction strategy has two implications: 1) the spatial continuity is captured by flattening neighboring samples in dual classic directions and 2) the spectral–spatial continuities are captured in the forward and backward directions. This multilevel dual-direction network provides a feasible way to avoid the loss of spatial information when flattening samples in the spatial domain, without using any convolutional layers. Inspired by our previous work, a variational autoencoder (VAE)-based network is used to enhance the noise immunity of latent features. To preserve the consistency of spectral–spatial sequential features, a combined loss function based on the solid angle on a unit sphere is proposed for parameter optimization. Two typical datasets are selected as benchmarks to show the effectiveness of MD²MVAE, compared with the state-of-the-art methods.

Video synopsis algorithm based on two-stage target tubes grouping

Video synopsis generates a concentrated video that can be browsed quickly. With the increase of condensation ratio, more pseudo collisions between target tubes will occur. To solve this problem, this paper proposed a video synopsis algorithm based on two-stage target tubes grouping. In the first stage, using the hypergraph to analyze the collision relationship between target tubes, and the target tubes are grouped according to the hyper-edges. In the second stage, a clustering algorithm based on equal distance nearest neighbor sampling is proposed to group the target tubes. Then, selecting target tubes according to the selection principle of quantity priority between groups and length priority within groups (QPB-LPG) . Finally, these target tubes are rearranged to generate concentrated videos with smaller pseudo collisions. The experimental results show that this algorithm can significantly reduce the pseudo collision between target tubes without reducing the frame condensation ratio and frame compact rate compared with existing video concentration algorithms, and the feasibility of the method is fully verified.

Complex Human Activity Recognition Using a Local Weighted Approach

Due to the life expectancy increase, there will be a workforce shortage in elderly care sector in forthcoming years. Ambient Assisted Living (AAL) systems can cope with this issue. A subset of AAL, Human activity recognition (HAR) provides an efficient way to tackle this issue. It can help with evaluating general health and welfare of elderly by automatically tracking their activities. Lifelogging and home diary applications will reduce the load on physicians and caregivers. On the other hand, complex activities play a vital role as they have high level semantic characteristics that truly represent daily life of the user. The main objective is to track these high-level semantic motions with low-cost single sensor systems with efficient machine learning frameworks. To achieve this objective, a framework is proposed to predict complex human activities from a single sensor using a machine learning approach. Time and frequency features are extracted from PAAL ADL Accelerometry Dataset and fed to Locally Weighted Random Forest (LWRF) machine learning algorithm. This algorithm is a hybrid structure that utilizes local weighting by introducing neighboring samples on Random Forest tree building phases. Proposed approach achieved 91% accuracy for HAR and 91.3% for gender recognition, outperforming other machine learning algorithms and previous study on the same dataset. This is the first study that utilize a local weighted approach for accelerometer signal domain. For prospective application, proposed framework can be embedded in lifelogging and home diary applications in home environments to track mental status of elderlies.

Adaptive Dual-Path Collaborative Learning for PAN and MS Classification

Due to the limitation of sensor technology, researchers tend to obtain high-quality image information from panchromatic (PAN) images and multispectral (MS) images with different resolutions. Therefore, the classification of remote sensing images of PAN and MS have become a research hotspot. In this paper, we propose an adaptive dual-path collaborative learning method for PAN and MS classification. In the stage of sample generation and training, we propose an adaptive neighborhood sample grading (ANSG) strategy in the establishing sample stage so that each pixel to be classified can obtain neighborhood information suitable for itself. Further, to simulate biological cognitive mechanisms, we divide the samples into different levels, and design the self-paced progressive loss (SPL), thus allowing the network to do preference training in different stages. The network’s training can quickly reach the optimal of the current stage and the overall convergence is more thorough. In the network structure, we propose a dual-path module (DPM) to effectively alleviate the gradient degradation in the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">residual path</i> , while ensuring maximum gradient loss information flow between every two layers in the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">densely connected path</i> . This module can extract more robust features to cope with the complex characteristics of remote-sensing images. Moreover, using the characteristics of the dual path to better fuse the features by the gradual collaborative fusion (GCF) way. The experimental results and theoretical analysis have demonstrated the proposed approach’s effectiveness, feasibility, and robustness. Our model are available at https://github.com/AIpy-nan/DBFI-Net.

GraphComm: Efficient Graph Convolutional Communication for Multiagent Cooperation

Artificial intelligence-empowered smart things (e.g., robots, autonomous vehicles, and unmanned aerial vehicles) have been transforming the world. The “brains” of smart things can be abstracted as the agents or cybertwins residing on end devices and edge servers. The next-generation communication networks (i.e., 6G) will become the nervous system for these agents and natively support multiagent cooperation. By sharing local observations and intentions via communication channels, the agents could better understand the environments and make right decisions. Due to the limited channel bandwidth, the communication is considered as a bottleneck of multiagent cooperation. In this article, we propose a graph convolutional communication method (GraphComm) for multiagent cooperation to relive the bottleneck. Specifically, a variational information bottleneck is used to encode the observations and intentions compactly. Furthermore, a graph information bottleneck with the attention-based neighbor sampling mechanism is utilized to improve the effectiveness and robustness of the multiround communication process. The experimental results show that GraphComm can improve the effectiveness, robustness, and efficiency of communication in multiagent cooperative tasks as compared to baseline methods.

Multiple infill criterion-assisted hybrid evolutionary optimization for medium-dimensional computationally expensive problems

Surrogate-assisted evolutionary algorithms have been paid more and more attention to solve computationally expensive problems. However, model management still plays a significant importance in searching for the optimal solution. In this paper, a new method is proposed to measure the approximation uncertainty, in which the differences between the solution and its neighbour samples in the decision space, and the ruggedness of the objective space in its neighborhood are both considered. The proposed approximation uncertainty will be utilized in the surrogate-assisted global search to find a solution for exact objective evaluation to improve the exploration capability of the global search. On the other hand, the approximated fitness value is adopted as the infill criterion for the surrogate-assisted local search, which is utilized to improve the exploitation capability to find a solution close to the real optimal solution as much as possible. The surrogate-assisted global and local searches are conducted in sequence at each generation to balance the exploration and exploitation capabilities of the method. The performance of the proposed method is evaluated on seven benchmark problems with 10, 20, 30 and 50 dimensions, and one real-world application with 30 and 50 dimensions. The experimental results show that the proposed method is efficient for solving the low- and medium-dimensional expensive optimization problems by compared to the other six state-of-the-art surrogate-assisted evolutionary algorithms.

Effect of street network design on traffic congestion and traffic safety

We examine the effects of street network design on congestion levels and crash rates in neighborhoods across Utah's Wasatch Front. We employ propensity score matching to select pairwise neighborhood samples that have other similar characteristics but differ greatly in street network design. Our results show that denser and more connected neighborhoods have significantly lower congestion levels, but they do not have measurably lower (or higher) crash rates, presumably due to the prevalence of four-way intersections. This study can help guide data-driven decision making on street network design standards for many of the growing urban areas across the country and globe.

Quality Variable Prediction for Nonlinear Dynamic Industrial Processes Based on Temporal Convolutional Networks

Soft sensors have been extensively developed to estimate the difficult-to-measure quality variables for real-time process monitoring and control. Process nonlinearities and dynamics are two main challenges for accurate soft sensor modeling. To cope with these problems, two temporal convolutional network (TCN)-based soft sensor models are developed in this paper. With a hierarchy of temporal convolution kernels and large receptive fields, TCN is able to describe the nonlinearities and long dynamic dependence of process variables. Thus, a TCN model is first designed for nonlinear dynamic soft sensing. However, the original TCN model neglects the auto-correlations of the quality variable, as well as the cross-correlations between the quality variable and process variables. Hence, an autoregressive TCN (AR-TCN) is further constructed by integrating the lagged quality data in the past neighboring samples for quality prediction of the current sample. Thus, AR-TCN is able to capture the auto-correlations and cross-correlations between the quality variables and process variables, which are more beneficial for quality prediction. The effectiveness of two proposed models is validated on an industrial debutanizer column and an industrial hydrocracking process.

Local- and regional-scale racial and ethnic disparities in air pollution determined by long-term mobile monitoring.

Disparity in air pollution exposure arises from variation at multiple spatial scales: along urban-to-rural gradients, between individual cities within a metropolitan region, within individual neighborhoods, and between city blocks. Here, we improve on existing capabilities to systematically compare urban variation at several scales, from hyperlocal (<100 m) to regional (>10 km), and to assess consequences for outdoor air pollution experienced by residents of different races and ethnicities, by creating a set of uniquely extensive and high-resolution observations of spatially variable pollutants: NO, NO2, black carbon (BC), and ultrafine particles (UFP). We conducted full-coverage monitoring of a wide sample of urban and suburban neighborhoods (93 km2 and 450,000 residents) in four counties of the San Francisco Bay Area using Google Street View cars equipped with the Aclima mobile platform. Comparing scales of variation across the sampled population, greater differences arise from localized pollution gradients for BC and NO (pollutants dominated by primary sources) and from regional gradients for UFP and NO2 (pollutants dominated by secondary contributions). Median concentrations of UFP, NO, and NO2 are, for Hispanic and Black populations, 8 to 30% higher than the population average; for White populations, average exposures to these pollutants are 9 to 14% lower than the population average. Systematic racial/ethnic disparities are influenced by regional concentration gradients due to sharp contrasts in demographic composition among cities and urban districts, while within-group extremes arise from local peaks. Our results illustrate how detailed and extensive fine-scale pollution observations can add new insights about differences and disparities in air pollution exposures at the population scale.

A novel αDistance Borderline-ADASYN-SMOTE algorithm for imbalanced data and its application in Alzheimer’s disease classification based on Dense Convolutional Network

The classification problem of imbalanced data has become a very important issue in the fields of machine learning and data mining. At present, relatively effective oversampling methods for processing imbalanced data include SMOTE, Borderline-SMOTE, and ADASYN. These algorithms have their own advantages; however, they do not adequately consider the distance factor, which is an important factor for balancing data precisely and reducing the misclassification probability of a minority boundary sample. Therefore, a new algorithm, αDistance Borderline-ADASYN-SMOTE algorithm, is proposed in the paper by combining the optimized Borderline-SMOTE algorithm with the optimized ADASYN algorithm. In the new algorithm, both the amount and the distance distribution of the nearest neighbor samples are considered. A few formulas are created to realize the algorithm. After being balanced by the algorithm, the data obtained from ADNI data set is trained, verified and tested by the Dense Convolutional Network. The experimental results show that the new model improves the classification performance of the Alzheimer’s disease.

Improved fuzzy C-means clustering algorithm based on t-SNE for terahertz spectral recognition

Terahertz (THz) waves characterized by low energy, instantaneity and proficiency in spectral analysis, have promising future in material identification. The main dimensionality reduction methods of Terahertz-based material identification, including Principal Component Analysis (PCA), Local Preserving Projection (LPP) and Local Linear Embedding (LLE), are sensitive to the number of nearest neighbor samples and neglect difference among classes, thus making it difficult to design the subsequent clustering model or leading to incorrect clustering. The t-distributed Stochastic Neighbor Embedding (t-SNE), which regards the sample distribution in the high-dimension as a Gaussian distribution and the coordinates in the low-dimension as t-distribution, makes the distance of clusters with long distance longer and then relieve their congestion. Besides, in the traditional Fuzzy C-means (FCM) clustering methods, the initial clustering center is randomly determined, so the problem of local optimum is easy to appear, which is prone to cause wrong recognition. To solve the above problems, an improved FCM algorithm is proposed in this paper for Terahertz spectral recognition. Firstly, t-SNE method is used for dimensional reduction and for the selection of initial clustering center for a more accurate clustering effect. On this basis, classical FCM clustering is used to recognize different substances through Terahertz spectrum. The algorithm can not only relieve the congestion among classes in the process of clustering, but also reflect the distance there for an appropriate cluster center in samples. In order to verify the reliability of the proposed method, the Terahertz time-domain spectroscopy is used to detect three genetically modified cotton seeds of lumianyan28, lumianyan29 and lumianyan36 respectively, obtaining their time-domain spectral data. It is the proposed method, which is used to analyze the spectral data, that successfully distinguishes three different types of transgenic cotton seeds, with a total accuracy of 0.9668. The result shows that the clustering method proposed in the paper has a bright prospect in identifying the Terahertz spectrum of materials.

An algorithm to identify the appropriate size of a neighbourhood when analysing spatially distributed field experiments by local cokriging

Farm management decisions based on multiple treatments or yield differences with measurable statistical uncertainty are effective and popular. The “oft” (on-farm-tools) software developed in this study provides an easy access to identify local neighbourhood points for non-collocated multiple treatment response differences. Methods for analysing multiple treatments based on multivariate geostatistics are generally computationally expensive for relatively large samples, such as data coming from modern sensors. Localised methods for multiple treatments reduce the computational complexity and can address local variability; however, their performance depends on the choice of the number of neighbours. A common practice for localised methods is to pre-specify a neighbourhood radius or number of neighbourhood samples based on analyst or farmer experience while tolerating the computational costs. Hence, introducing the “oft” package, we propose an adaptive neighbourhood point algorithm to identify the neighbouring samples for the localised analysis in order to achieve a good balance between computation time and accuracy. We explore an experimental case study dataset obtained from a vineyard in Langhorne Creek, South Australia, which is a particular application of the general problem of the neighbourhood selection for localised multivariate geostatistics.

Neural-Network-Based Cross-Channel Intra Prediction

To reduce the redundancy among different color channels, e.g., YUV, previous methods usually adopt a linear model that tends to be oversimple for complex image content. We propose a neural-network-based method for cross-channel prediction in intra frame coding. The proposed network utilizes twofold cues, i.e., the neighboring reconstructed samples with all channels, and the co-located reconstructed samples with partial channels. Specifically, for YUV video coding, the neighboring samples with YUV are processed by several fully connected layers; the co-located samples with Y are processed by convolutional layers; and the proposed network fuses the twofold cues. We observe that the integration of twofold information is crucial to the performance of intra prediction of the chroma components. We have designed the network architecture to achieve a good balance between compression performance and computational efficiency. Moreover, we propose a transform domain loss for the training of the network. The transform domain loss helps obtain more compact representations of residues in the transform domain, leading to higher compression efficiency. The proposed method is plugged into HEVC and VVC test models to evaluate its effectiveness. Experimental results show that our method provides more accurate cross-channel intra prediction compared with previous methods. On top of HEVC, our method achieves on average 1.3%, 5.4%, and 3.8% BD-rate reductions for Y, Cb, and Cr on common test sequences, and on average 3.8%, 11.3%, and 9.0% BD-rate reductions for Y, Cb, and Cr on ultra-high-definition test sequences. On top of VVC, our method achieves on average 0.5%, 1.7%, and 1.3% BD-rate reductions for Y, Cb, and Cr on common test sequences.

Graph representation learning for road type classification

We present a novel learning-based approach to graph representations of road networks employing state-of-the-art graph convolutional neural networks. Our approach is applied to realistic road networks of 17 cities from Open Street Map. While edge features are crucial to generate descriptive graph representations of road networks, graph convolutional networks usually rely on node features only. We show that the highly representative edge features can still be integrated into such networks by applying a line graph transformation. We also propose a method for neighborhood sampling based on a topological neighborhood composed of both local and global neighbors. We compare the performance of learning representations using different types of neighborhood aggregation functions in transductive and inductive tasks and in supervised and unsupervised learning. Furthermore, we propose a novel aggregation approach, Graph Attention Isomorphism Network, GAIN1. Our results show that GAIN outperforms state-of-the-art methods on the road type classification problem.

A fusion algorithm model based on KNN-SVM to classify and recognize spam

Spam usually has the characteristics of large data dimension and large numbers of samples. At present, the common classification learning algorithms have their own advantages and disadvantages. Aiming at the classification tendency of K nearest neighbor algorithm (KNN) in the face of unbalanced sample data and the long time consuming of support vector machine (SVM) training model, a fusion algorithm of KNN and SVM is proposed. Firstly, the content of spam is extracted, and then the text preprocessing is transformed into sparse matrix. The k nearest neighbor samples of the test sample set are screened out by KNN algorithm, and then the SVM algorithm is used to train these screened samples, establish the separation hyperplane, identify the mail, and judge whether it is spam or not. The results show that the KNN-SVM fusion algorithm can effectively reduce the impact of sample imbalance while maintaining the high classification accuracy of SVM algorithm, and its classification efficiency is slightly lower than that of KNN, algorithm but much higher than that of SVM algorithm.

Inductive and irregular dynamic network representation based on ordinary differential equations

Dynamic network representation has aroused increasing attention from researchers since it reflects the characteristics of real-world networks and has been widely used in various applications. Enormous efforts have been made to this problem. Existing studies often use fixed-time steps. However, in fact, the dynamics in real-world networks are usually irregular over time. To address this problem, in this paper, we employ ordinary differential equations and propose a novel graph representation approach for dynamic networks, namely Graph-ODE. The core idea lies in that it leverages ordinary differential equations to model the continuous latent dynamics of the network and therefore can naturally capture the evolution pattern. Specifically, to deal with the irregularity of node interactions, we first propose a differential form of GRU aggregator to aggregate the chronological neighbors. In addition, we employ neighbor sampling strategies for efficiency. Besides, our inductive algorithm owns the ability to deal with new emerging nodes. Experimental results on various real-world data sets demonstrate that our method significantly outperforms several state-of-the-art algorithms on downstream tasks.

Collaborative Filtering Recommendation Algorithm Based on Class Correlation Distance

Background: A class correlation distance collaborative filtering recommendation algorithm is proposed to solve the problems of category judgment and distance metric in the traditional collaborative filtering recommendation algorithm, which is using the advantage of the distance between the same samples and the class related distance. First, the class correlation distance between the training samples is calculated and stored. Second, the K nearest neighbor samples are selected, the class correlation distance of training samples and the difference ratio between the test samples and training samples are calculated respectively. Finally, according to the difference ratio, we classify the different types of samples. The experimental result shows that the algorithm combined with user rating preference can get lower MAE value, and the recommendation effect is better. With the change of K value, CCDKNN algorithm is obviously better than KNN algorithm and DWKNN algorithm, and the accuracy performance is more stable. The algorithm improves the accuracy of similarity and predictability, which has better performance than the traditional algorithm. Objective: A class correlation distance collaborative filtering recommendation algorithm is proposed to solve the problems of category judgment and distance metric in the traditional collaborative filtering recommendation algorithm, which is using the advantage of the distance between the same samples and the class related distance. Methods: First, the class correlation distance between the training samples is calculated and stored. Second, the K nearest neighbor samples are selected, the class correlation distance of training samples and the difference ratio between the test samples and training samples are calculated respectively. Finally, according to the difference ratio, we classify the different types of samples. Results: The experimental result shows that the algorithm combined with user rating preference can get lower MAE value, and the recommendation effect is better. Conclusion: With the change of K value, CCDKNN algorithm is obviously better than KNN algorithm and DWKNN algorithm, and the accuracy performance is more stable. The algorithm improves the accuracy of similarity and predictability, which has better performance than the traditional algorithm.