International Journal of Pattern Recognition and Artificial Intelligence

The prevalence of safety hazards has intensified, underscoring the pressing need for effective responses to sudden disasters and enhanced emergency management. The traditional forecast-response paradigm is no longer adequately addressing the management and decision-making needs for contemporary emergency incidents. Existing emergency management frameworks often struggle with the systemic characteristics of unexpected events. This study proposes a Bayesian network-based scenario-response framework for situation deduction in Emergency Management, taking the torrential rain disaster in Henan Province as a point of reference. In this context, the fundamental units of disaster-forming environment (E), emergency measures (M), and scenario state (S) are employed to construct a coherent and dynamic scenario evolution chain, which effectively depicts the progression of the scenario linkage process. This approach facilitates a deeper understanding of the evolution of emergencies and enhances predictive capacity.

The widespread application of cloud services make users pay more attention to Quality of Service (QoS). Generally, the user cannot call all services simultaneously to obtain corresponding QoS data and can only choose a service from a few known data, thus it’s critical to predict unknown QoS values. A third-order tensor can model temporal patterns of QoS data, and studies indicate that the tensor latent factor analysis models based on Canonical Polyadic (CP) decomposition can effectively capture temporal patterns to predict unknown data in QoS. However, the existing CP decomposition-based models limit their learning ability since rank-one tensors contain less structure information, which results in low prediction accuracy. Therefore, this paper proposes a Biased Block Term Tensor Decomposition (BBTTD) model to achieve high accuracy for temporal pattern-aware QoS prediction. It mainly adopts the following three-fold ideas: (a) implementing a tensor learning model by adopting the block term decomposition in rank-($L_r$, $L_r$, 1) terms; (b) proposing the bias block term tensors to enhance the model’s prediction accuracy; (c) designing a nonnegative multiplication update algorithm to learning model parameters. Extensive experiments on two public dynamic QoS datasets demonstrate that BBTTD has higher prediction accuracy compared with several QoS prediction models.

Malicious firmware upgrading represents a critical security vulnerability in Internet of Things (IoT) devices. This study introduces HyCNNAt, a novel hybrid deep learning network for IoT malware detection that synergistically combines Convolutional Neural Networks (CNNs) with transformer attention mechanisms. HyCNNAt’s architecture vertically and horizontally stacks convolution and attention layers, enhancing the network’s generalization capabilities, capacity, and overall effectiveness. We evaluated HyCNNAt using a publicly available IoT firmware dataset, where it demonstrated superior performance with the highest accuracy ($97.11\%\pm 1.02\%$), F1-score ($99.992\%\pm 0.004\%$), and recall ($97.48\%\pm 2.6556\%$), highlighting its robust classification capabilities, although its precision ($91.27\%\pm 45.08\%$) exhibited variability compared to state-of-the-art models such as CoAtNet, MobileViT, MobileNet, and MobileNet variants using transfer learning. These results underscore HyCNNAt’s potential as a robust solution for addressing the pressing challenge of IoT malware detection.

Accurately and rapidly detecting traffic object has been attracted intensive attention due to its potential applications in the fields of autonomous driving, traffic flow monitoring, augmented reality (AR) and so on. However, there are many difficulties in the process of traffic object detection indeed, such as occlusion and aggregation between objects, insufficient feature extraction of objects, in particular the presence of a large number of small objects, which bring great challenges to these traffic objects detection. In this paper, an improved traffic object detection model based on You-Only-Look-Once version 5 small (YOLOv5s) is proposed to address the issues. By utilizing spatial pyramids to extract multi-scale spatial features and applying Squeeze-and-Excitation (SE) channel attention to capture more global and local semantic features, especially by designing a sub-network in the neck to fuse high-resolution information in shallow layers with more accurately semantic information in deep layers, the detection sensitivity of object features is enhanced. More importantly, by explanting decoupled-head into the network, outstanding performance of the model with high detection accuracy and rapid detection speed is realized. The experimental results on the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) and Laboratory for Intelligent and Safe Automobiles (LISA) traffic signs datasets both show that the modified model significantly improves the detection accuracy. Meanwhile, the high real-time performance is still maintained. Undoubtedly, the modified model proposed in this paper can effectively address many difficulties in traffic object detection under various complex scenes, which would be greatly helpful for its potential applications in the future.

In this paper, we introduce a novel system to assist 6D object pose estimation network training, which is only deployed in the training progress to optimize the network parameters, and does not work in the testing stage, called Online training and offline testing system (OTOT). OTOT consists of two modules: a feature fusion module and a supervision module. The feature fusion module fuses several feature maps from the pose estimation network in a specified order to obtain a fused feature. Then, the supervision module uses the encoder–decoder structure network to implicitly extract useful features from the fused feature and optimizes the pose estimation network online through the back-propagation mechanism. OTOT can be migrated to any network with encoder–decoder structure. The network trained with OTOT achieves 56.11% accuracy in terms of the VSD metric on the TLESS dataset using RGB inputs, compared to the 46.70% accuracy of the original network trained without OTOT. Experiments show that OTOT greatly improves the accuracy of the pose estimation network, and since OTOT is not deployed in the testing stage, it does not increase any parameters during testing and affect the original speed of the network.

This paper introduces a fault diagnosis method for synchronous traction machine demagnetization based on the Informer model. The method utilizes the Informer model to analyze and model the sensor data of the synchronous traction machine, adaptively learn the feature representation of time series data, and predict the future states in order to accurately identify and predict demagnetization faults. The specific steps include: (1) collecting sensor data of the synchronous traction machine, including parameters such as current, voltage, and rotational speed; (2) preprocessing the data, including denoising, normalization, and feature extraction; (3) constructing the Informer model and training and optimizing it using the preprocessed sensor data; and (4) using the trained model to predict and determine new sensor data, thereby achieving an accurate diagnosis of demagnetization faults. The advantages of this method are as follows: (1) automatic learning and extraction of important features of time series data without the need for manual feature design, improving diagnostic accuracy; (2) ability to handle long sequence data and strong modeling capability for time dependencies, better predicting future states and fault occurrences; and (3) adaptability to the characteristics and data features of different elevator systems and strong generalization capability.

Image classification has surpassed human performance in numerous domains. However, in real-world scenarios, challenges such as storage constraints, privacy concerns, and commercial data protection necessitate the enhancement of existing models to accommodate new class classifications — a critical area of research known as class incremental learning. Data replay stands out as a pivotal technique within this domain. The DeepInversion algorithm ingeniously employs classification networks to generate training set images of notable quality. In this study, we improve the DeepInversion algorithm, leveraging a pre-trained model to yield superior quality playback images for class incremental learning. Within the context of class incremental learning, we introduce a cosine orthogonal classification loss function, formulated on the basis of linear layer analysis, to guide image generation and augment class incremental learning. This loss function is designed to ensure the mutual orthogonality of all class centers while minimizing intra-class distances. In the realm of knowledge distillation, we harness a combination of limited real images, a profusion of synthesized images, and Mix virtual images to facilitate feature cosine distance distillation. Sufficient comparative experiments and analyses with similar latest methods underscore the efficacy of our proposed approach, and obtain the SOTA results. The code for the paper can be found at http://github.com/YunXiaooooo/Class-incremental-learning-based-on-playback-images-generated-by-classification-network

There are lots of gravel layers gradually increasing from east to west of Kuqa Piedstone in Tarim Basin, especially the thickest gravel layer, high gravel content, large particle size, high compressive strength and poor drillability in Bozi block, which are the key factors that result in low drilling efficiency and complex drilling accidents. Therefore, this paper presents a design scheme based on lithology characteristics deep learning and artificial intelligence analysis of bit failure mode and characteristics of large section gravel layer in fore-salt. The design scheme of a PDC special-shaped tooth bit with strong impact resistance has been formed through the structural design of anti-impact aggressive PDC teeth, rock breaking efficiency study, combined with the optimization design of bit layout. The design scheme not only improves the anti-impact but also improves the anti-eddy performance of the PDC special-shaped tooth bit. It effectively expands both the application range of PDC special tooth bit in the gravel layer and prolongs the bit life. The artificial intelligence optimized PDC special-shaped tooth bit has obtained a great effect in deep gravel stratum, in the meantime, it successfully solved the problems of slow drilling rate and short footage of single bit in Kuqa formation and other formations.

The acquisition of clear images is a critical aspect in various fields including computer vision, aerial detection, and medical imaging. The issue of image blur caused by object motion poses a challenge in obtaining clear images. To address this, an improved AT-DGAN network model is proposed in this paper. This model integrates the pyramid generator module of the DeblurGAN-v2 network with a self-attention mechanism. The feature pyramid is employed for image feature extraction and representation, while the self-attention mechanism dynamically adjusts the weight of important features in each pyramid layer and performs weighted fusion, thereby compensating for the information loss during feature extraction in the feature pyramid network. Additionally, a hinge loss function is designed for the proposed model to balance the discriminator and the generator, enhancing the stability and training efficiency of the generative adversarial network. The experimental results show that compared to other algorithms of the same type, this improved algorithm has increased the Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) of restored images by 0.58dB and 1.5%, respectively.

This paper first proposes the notion of the intuitionistic fuzzy sets on inclusion degree, furthermore, a couple of dual operators’ lower approximations, and upper approximations of fuzzy inclusion approximate space are provided, thus, a probabilistic intuitionistic fuzzy set model that stemmed from inclusion degree was obtained. A rough intuitionistic fuzzy set and histogram equalization-based image enhancement algorithm is proposed to address the shortcomings of excessive enhancement and loss of image detail information in fuzzy enhancement that cannot improve image contrast and histogram equalization enhancement. The fusion of rough intuitionistic blur enhancement and histogram equalization focuses on rough intuitionistic blur enhancement while suppressing the shortcomings of histogram equalization, which not only enhances the detailed information of the image but also improves its contrast. Finally, its effectiveness is verified through typical image enhancement examples.

The process of utilizing Artificial Intelligence (AI) to identify and label human behaviors from unprocessed activity data gathered from various sources is known as Human Activity Recognition (HAR). Because of its potential applications across multiple areas, computer vision faces a significant challenge in recognizing human actions and the accompanying interactions with objects and the environment. Investigating the temporal and geographical characteristics of the skeleton sequence is essential for this endeavor, according to recent studies. However, efficiently extracting discriminative temporal and spatial information remains a difficult task. This work proposes a novel Human Action Recognition Model exploiting improved Graph Convolutional Network (GCN)-based pose estimation with a Hybrid Classifier (IGCN-HC). The phases carried out in this model are pre-processing, pose estimation, feature extraction, and activity recognition. Initially, the input video will be pre-processed and a frame from the input video stream will be generated. Subsequently, human pose estimation exploiting improved GCN will be accomplished. Further, human skeletal joints’ coordinates in two- or three-dimensional spaces are determined via human pose estimation. Then, Shape Local Binary Texture (SLBT) and an improved hierarchy of skeleton features have been used to detect the variance in different activities. In the last phase, a hybrid classification model with the combination of Deep Maxout and customized CNN has been proposed for the recognition phase. The model utilizes two inputs pose estimation results (skeleton) and the extracted features for training purposes. Finally, the proposed trained model is evaluated for recognition on different test inputs and contrasted with the existing techniques.

Numerous people to determine the location by wireless fingerprint technology indoors at present. The premise of fingerprint positioning is to assume that the received signal strength distance is consistent with the positioning distance. However, due to the interference of obstacles such as walls, desks, chairs, pedestrians, the closest distance selected using the received signal strength distance may not be the closest to the target at its corresponding position, which may lead to large positioning errors. To improve accuracy, this paper analyzes the advantages of the respective features of RSS and CSI, use algorithms to extract them, and proposes a novel WiFi fingerprint positioning algorithm for fusion to estimate the target location. The experimental data show that this method has certain advantages in improving positioning accuracy.

To mitigate the flutter instability observed during lift-type power catwalk operations, a Dual-Loop Error-Driven Adaptive Disturbance Rejection Control (EDDL-ADRC) strategy is proposed. This approach enhances the tracking accuracy and robustness of the electro-hydraulic servo system under variable load and large inertia conditions, effectively alleviating flutter induced by fluctuations in driving speed. Under low-frequency small load conditions, the EDDL-ADRC overcomes the limitations of traditional Active Disturbance Rejection Control (ADRC), particularly those related to system order, and demonstrates superior control performance compared to conventional ADRC. The dual-loop error-driven controller consists of two identical second-order ADRC modules. The primary controller operates in a conventional driving mode, while the error-driven controller utilizes both the system’s output error and control input as driving signals. This error-driven controller effectively compensates for phase lag in the main controller and the limited observation accuracy of the Extended State Observer (ESO), thereby improving overall system performance. The proposed control strategy’s effectiveness is validated via a joint simulation platform that integrates both the electro-hydraulic and mechanical systems of the catwalk mechanism.

The attention mechanism has been successfully used in a sequence consisted of a series of word embeddings to improve the representation of the sequence. Inspired by this, we leverage the attention mechanism on a set of tasks to implement a multi-task transfer learning method called AMP (Attentions between Multiple Prompts). First, we encode a task into a prompt as task representation called task embedding. Second, we learn an attention component on all task embeddings to generate a combined prompt for each task, which is an attention-weighted sum of task embeddings. Each combined prompt incorporates the knowledge of all tasks. The word embedding is a vector, but the task embedding is a 2D matrix. The attention mechanism can be exploited on a set of vectors rather than on a set of matrices. The prior methods employ pooling or flattened method to transform the matrix to the vector for computing the attentions between matrices. We propose a method called DAM (Direct Attention Mechanism) which can compute attentions between matrices directly without transforming. DAM method can more exactly compute the attentions between matrices. Wide experiments demonstrate that AMP outperforms prompt-tuning method and prior prompt transfer methods.

Electric power resources are essential for the efficient and orderly development of society. Accurate power load forecasting is a key driver for the low-carbon upgrade of power systems. Traditional forecasting methods often struggle to capture long-term dependencies. Additionally, extracting complex nonlinear features from data remains a significant challenge, making it challenging to meet the accuracy demands of modern power systems. Besides, current deep learning-based forecasting methods cannot simulate multi-granularity power load data. To address these challenges, this paper presents a Generative Pre-trained Transformer model, GPT4PLTS, designed for power data simulation and fine-grained power load forecasting. The model leverages the Transformer architecture, incorporating the first six layers of the GPT decoder structure. It utilizes a multi-head attention mechanism to extract temporal features and includes a time alignment layer to maintain the sequence of time-series data, addressing both short-term and long-term dependencies. Extensive experiments are conducted on load observations from 2000 enterprises. The results demonstrate that GPT4PLTS achieves high accuracy in data simulation and forecasts across different time granularities, particularly excelling in short and medium-term predictions. Future research could focus on optimizing the model structure to enhance the model’s generalization ability.