International Journal of Pattern Recognition and Artificial Intelligence

In the last few years, ensemble learning has received more interest primarily for the task of classification. It is based on the postulation that combining the output of multiple experts is better than the output of any individual expert. Ensemble feature selection may improve the performance of the learning algorithms and has the ability to obtain more stable and robust results. However, during the process of feature aggregation and selection, selected feature subset may contain high levels of inter-feature redundancy. To address this issue, a novel algorithm based on feature rank aggregation and graph theoretic technique for ensemble feature selection (R-GEFS) with the fusion of Pearson and Spearman correlation metrics is proposed. The method works by aggregation of the profile of preferences of five feature rankers as the base feature selectors. Then similar features are grouped into clusters using graph theoretic approach. The most representative feature strongly co-related to target decision classes is drawn from each cluster. The efficiency and effectiveness of the R-GEFS algorithm are evaluated through an empirical study. Extensive experiments on 15 diverse benchmark datasets are carried out to compare R-GEFS with seven state-of-the-art feature selection models with respect to four popular classifiers, namely decision tree, k nearest neighbor, random forest, and support vector machine. The proposed method turns out to be effective by selecting smaller feature subsets with lesser computational complexities and it assists in increasing the classification accuracy.

Graph attention networks (GATs)-based method performs well in Human–Object Interaction (HOI) detection due to its ability to aggregate contextual information. However, the traditional GATs-based methods are computationally intensive, and the graph’s structure cannot effectively represent the HOI in an image. Meanwhile, the graph models are unable to predict the interactions that involve less contextual information correctly. In this paper, we design a method based on graph models called V-SGATs, which stands for Visual Branch and Star Graph Attention Networks. The human-centric star graph and object-centric star graph are adopted to reduce computational complexity, and the logical structure of the star graph can represent HOI more reasonably. Meanwhile, the visual branch that recognizes the interaction without utilizing the contextual information is designed to aid the graph model in predicting the interactions that involve less contextual information. Experiments are carried out on two large-scale HOI public benchmarks V-COCO and HICO-DET, and the results show that the proposed method performs better than most of the existing methods based on GATs.

Improving the reliability of nighttime pedestrian detection is a crucial challenge towards the design of robust autonomous systems. Not surprisingly, most pedestrian fatalities occur in low-illumination settings, thus emphasizing the need for new algorithmic advances. This work presents a novel pedestrian detection approach that makes a number of crucial modifications to the state-of-the-art YOLOV5-PANet architecture, in order to improve the reliability of features extracted from nighttime images. More specifically, the proposed architecture systematically incorporates powerful shuffle attention mechanisms and a transformer module to improve the feature learning pipeline. Instead of advocating the use of other sensing modalities that are better suited for nighttime detection, our approach relies only on conventional RGB cameras and is hence broadly applicable. Our empirical studies with nighttime pedestrian detection benchmarks show that with only minimal increase in model complexity, our approach provides significant improvements in detection efficacy over existing solutions. Finally, we explore the impact of post-hoc network pruning on the speed-accuracy trade-off of our approach and demonstrate that it is well suited for reduced memory/compute requirements.

There are many problems such as strong subjectivity, complex calculations, and lack of intelligence in most of the current energy system evaluation models, so a design of an evaluation system for urban smart energy systems based on an improved genetic algorithm-back propagation (BP) neural network is proposed. First of all, the hierarchical structure of the indicator evaluation system for energy system was established, analytic hierarchy process (AHP) was used to assign weights to each indicator, and data samples were classified. Then, SeqGAN was used to expand the data set, which solved the difficult problem of data acquisition. Finally, the genetic algorithm was used to determine the best initial values of the weights and thresholds of the BP neural network structure parameters designed for this research. The simulation experiment results showed that the method in this paper has higher classification accuracy than the traditional method, and comprehensive evaluation model proposed can effectively evaluate the urban smart energy system.

Graph neural network (GNN) is a powerful representation learning framework for graph-structured data. Some GNN-based graph embedding methods, including variational graph autoencoder (VGAE), have been presented recently. However, existing VGAE-based methods typically focus on reconstructing the adjacent matrix, i.e. topological structure, instead of the node features matrix, this strategy makes graphical features difficult to be fully learned, which weakens and restricts the capacity of a generative network to learn higher-quality representations. To address the issue, we use a contrastive estimator on the representation mechanism, i.e. on the encoding process under the framework of VGAE. In particular, we maximize the mutual information (MI) between encoded latent representation and node attributes which acts as a regularizer forcing the encoder to select the most informative with respect to the node attributes. Additionally, we also solve another key question how to effectively estimate the mutual information by drawing samples from the joint and marginal, and explain why the maximization of MI can contribute to the encoder obtaining more node feature information. Ultimately, extensive experiments on three citation networks and four web-age networks show that our method outperforms contemporary popular algorithms (such as DGI) on node classifications and clustering tasks, and the best result is an $8.28\%$ increase on node clustering task.

Document age estimation using handwritten text line images is useful for several pattern recognition and artificial intelligence applications such as forged signature verification, writer identification, gender identification, personality traits identification, and fraudulent document identification. This paper presents a novel method for document age classification at the text line level. For segmenting text lines from handwritten document images, the wavelet decomposition is used in a novel way. We explore multiple levels of wavelet decomposition, which introduce blur as the number of levels increases for detecting word components. The detected components are then used for a direction guided-driven growing approach with linearity, and nonlinearity criteria for segmenting text lines. For classification of text line images of different ages, inspired by the observation that, as the age of a document increases, the quality of its image degrades, the proposed method extracts the structural, contrast, and spatial features to study degradations at different wavelet decomposition levels. The specific advantages of DenseNet, namely, strong feature propagation, mitigation of the vanishing gradient problem, reuse of features, and the reduction of the number of parameters motivated us to use DenseNet121 along with a Multi-layer Perceptron (MLP) for the classification of text lines of different ages by feeding features and the original image as input. To demonstrate the efficacy of the proposed model, experiments were conducted on our own as well as standard datasets for both text line segmentation and document age classification. The results show that the proposed method outperforms the existing methods for text line segmentation in terms of precision, recall, F-measure, and document age classification in terms of average classification rate.

The rapid development of deep reinforcement learning makes it widely used in multi-agent environments to solve the multi-agent cooperation problem. However, due to the instability of multi-agent environments, the performance is insufficient when using deep reinforcement learning algorithms to train each agent independently. In this work, we use the framework of centralized training with decentralized execution to extend the maximum entropy deep reinforcement learning algorithm Soft Actor-Critic (SAC) and proposes the multi-agent deep reinforcement learning algorithm MASAC based on the maximum entropy framework. Proposed model treats all the agents as part of the environment, it can effectively solve the problem of poor convergence of algorithms due to environmental instability. At the same time, we have noticed the shortcoming of centralized training, using all the information of the agents as input of critics, and it is easy to lose the information related to the current agent. Inspired by the application of self-attention mechanism in machine translation, we use the self-attention mechanism to improve the critic and propose the ATT-MASAC algorithm. Each agent can discover their relationship with other agents through encoder operation and attention calculation as part of the critic networks. Compared with the recent multi-agent deep reinforcement learning algorithms, ATT-MASAC has better convergence effect. Also, it has better stability when the number of agents in the environment increases.

Automated guided vehicle (AGV) is an important transportation equipment, which is widely used in warehouses and factories. In the scenarios of multi-AGVs application, an efficient AGVs task assignment strategy is beneficial for transportation costs, balance of workload and increasing distribution efficiency. The traditional method usually depends on a powerful scheduling system, which solves the task assignment problem in a regular way. In this paper, we present a decentralized framework of multi-task allocation with attention (MTAA) in deep reinforcement learning, which combines with the methods of task assignment in balance and path planning in cooperation for distribution application. As for task assignment balance, we adopt DNN network to achieve task assignment equilibrium. In multi-AGVs path planning, methods of A3C are embedded in MTAA framework, which are instrumental in improving the stationarity and performance in deep reinforcement learning application. In experiments, we designed two different scenarios under different obstacles to verify the performance of MTAA-A3C and MTAA-DQN methods. The experiments show that the proposed approach has feasibility and effectiveness used in multi-AGVs application.

Quantitative phase imaging (QPI) technology is one of the important techniques for nondestructive imaging of a cell’s morphology. In QPI, the decoupling of refractive index (RI) and thickness of a cell is a key problem. We aim at the multiple hypothesis constraints (MHC) shortcomings of the traditional method in the decoupling of multi-media and nonspherical cells. A method is proposed to decouple the RI and the thickness of a cell under three-wavelength phase imaging in this paper. In this method, the sample’s RI and thickness distributions can be obtained by solving three-wavelength phase equations using the approximate setting under dispersion based on the optical dispersion theories. In simulation experiments, the results show that the method can effectively decouple the RI and thickness for homogeneous and multi-media nonspherical cells and has high accuracy. Because the method in this paper is obtained by using few approximate assumptions and strict mathematical analysis, it can decouple the RI and thickness of nuclear cells with any shape, and moreover, it provides an important basis for the technological development of cell morphology analysis.

The stereo correspondence problem is one of the most pre-eminent problems in a stereo vision system. With the right correspondence, a stereo vision system can help cap over diverse problems, while on the other hand, a wrong correspondence can be costly. While the performance of a feature-based correspondence approach is exceptional, the method can still produce wrong correspondences. This work presents an amalgam of feature-based and correlation-based correspondence, where the local pixels around a feature pair are compared using Structural SIMilarity index (SSIM), enhancing the correspondences, and a semantic-based filtering module, which further filters the obtained corresponding features using semantic data whenever detected in both the stereo image pair. While approaches in the literature are focused towards finding better features and their representation, the proposed approach advocates that correlation-based verification of the features can filter out bad correspondences, and in addition, aided by semantic-level filtering. These two modules establish the novelty of the work. The proposed correspondence matching algorithm is used to solve the problem of Visual Odometry to let a low-cost robot compute its pose in a novel environment. The experimental results show adequate filtering of wrong feature correspondence wherein, different environments with different lighting conditions were also considered. The proposed approach outperformed numerous state-of-the-art approaches available in the literature. The visual odometry algorithm using the proposed correspondence matching is compared against classical methods and a deep learning method, and it is observed that the proposed approach delivers lower trajectory error values in most scenarios on the KITTI dataset sequences.

Musculoskeletal disorders (MSDs) are associated with sitting postures. The assessment and prevention of risk factors for workplace exposure are indispensable aspects of reducing the occurrence of MSDs. This paper proposes an ergonomic assessment method of risk factors for MSDs associated with sitting postures in the actual working conditions. A Kinect sensor with the RULA method was primarily used to collect the data and evaluate the relevant postures. The results obtained were compared with the evaluation results by a human expert. Additionally, we verified the capability and effectiveness of this method. A program system for human joint recognition and acquisition was implemented. The results indicated that the Kinect joint data is generally accurate and can adequately complete the RULA evaluation table. The results from the front and right-hand side obtained by the Kinect were consistent with the results of the expert evaluation, and no significant difference was observed between them ($p>0.05$). However, when the participants faced the Kinect, the sensor performed better, and the evaluation result was more accurate. A high consistency was observed between the evaluation results obtained from the front and the expert (proportion agreement $\mathrm{\text{index}}=0.65$, Cohen’s $\mathrm{\text{kappa}}=0.77$). Only a slight consistency was observed between the evaluation results obtained from the right-hand side and the expert (proportion agreement $\mathrm{\text{index}}=0.41$, Cohen’s $\mathrm{\text{kappa}}=0.08$). This research created a new ergonomic method for the risk assessment of MSDs associated with sitting postures. The combination of theory and practice is crucial in the risk assessment of sitting postures in workplaces.

The distribution transformer voltage may be overloaded, which may lead to the aging of distribution transformer components, shorten the service life of distribution transformer components and even affect the daily life of community residents and the operation of enterprises. A large amount of real data are collected, and the factors that affect the heavy overload of distribution transformer are comprehensively considered from multiple angles, so as to establish a model for future prediction and early maintenance to reduce losses. First, the collected data is analyzed by attributes and preprocessed to improve the quality of the data. Then, the time attributes are generalized according to seasons, months, holidays and weekends. The test results show that the data prediction value is more accurate when generalized according to seasons. For the prediction model, the gradient lifting decision tree algorithm is selected to establish the model, and then the parameters are further optimized, and finally the model is evaluated. Lastly, the prediction accuracy of the model reaches a high level, and it can be determined that the prediction is close to the objective fact. The model can be used to predict the heavy overload of distribution transformer voltage, so as to reduce the loss caused by abnormal conditions of relevant equipment for the enterprises.

In recent years, the study of social-based recommender systems has become an active research topic. We incorporate a combination of static and dynamic interest characteristics to predict users’ real-time dynamic interests, which has rarely been considered in previous studies. In this paper, we propose a graph neural network social recommendation model that integrates static and dynamic feature relationships (FSDFR-GNNSR). The model uses a graph embedding algorithm to extract static features of users and movies, and takes the static features as input to gated recurrent unit (GRU), so that the model can take static features into consideration while modeling user dynamic behavior. Finally, we use graph attention networks to represent the dynamic influence of friends, simplify the update strategy of second-order neighbor nodes. We apply graph pooling operations to improve the generalization ability of the algorithm. Empirical analyses on real datasets show that the proposed approach achieves superior performance to existing approaches.

Decentralized control allows each unmanned surface vehicle (USV) in a flock to be autonomous to some extent. In this paper, the flocking algorithm put forth by Olfati-Saber was improved and applied in the multi-leader scenario based on its original flocking movement. Repulsion was also taken into account in the formation. Subsequently, the concept of neighborhood in finite-range communications was borrowed to control the communication distance of each USV. In this way, USVs could satisfy the restriction of a given communication distance, and avoid collision in motion. USVs might have collaborative tracking of a virtual leader and obtain the distributed cooperative control strategy. Based on the simulation results, the improved flocking control algorithm allowed every USV to accurately track the speed of the virtual leader they followed, respectively. The test results also demonstrated the effectiveness of the algorithm in safe flock navigation, and proved that the proposed control algorithm must be an effective way to ensure the safe flocking of multiple USVs.