Twisted convolutional networks (TCNs): Enhancing feature interactions for non-spatial data classification.

To apply deep learning algorithms using a conventional convolutional neural network (CNN) and a recurrent CNN to differentiate three breast cancer molecular subtypes on MRI. A total of 244 patients were analyzed, 99 in training dataset scanned at 1.5 T and 83 in testing-1 and 62 in testing-2 scanned at 3 T. Patients were classified into 3 subtypes based on hormonal receptor (HR) and HER2 receptor: (HR+/HER2-), HER2+, and triple negative (TN). Only images acquired in the DCE sequence were used in the analysis. The smallest bounding box covering tumor ROI was used as the input for deep learning to develop the model in the training dataset, by using a conventional CNN and the convolutional long short-term memory (CLSTM). Then, transfer learning was applied to re-tune the model using testing-1(2) and evaluated in testing-2(1). In the training dataset, the mean accuracy evaluated using tenfold cross-validation was higher by using CLSTM (0.91) than by using CNN (0.79). When the developed model was applied to the independent testing datasets, the accuracy was 0.4-0.5. With transfer learning by re-tuning parameters in testing-1, the mean accuracy reached 0.91 by CNN and 0.83 by CLSTM, and improved accuracy in testing-2 from 0.47 to 0.78 by CNN and from 0.39 to 0.74 by CLSTM. Overall, transfer learning could improve the classification accuracy by greater than 30%. The recurrent network using CLSTM could track changes in signal intensity during DCE acquisition, and achieved a higher accuracy compared with conventional CNN during training. For datasets acquired using different settings, transfer learning can be applied to re-tune the model and improve accuracy. • Deep learning can be applied to differentiate breast cancer molecular subtypes. • The recurrent neural network using CLSTM could track the change of signal intensity in DCE images, and achieved a higher accuracy compared with conventional CNN during training. • For datasets acquired using different scanners with different imaging protocols, transfer learning provided an efficient method to re-tune the classification model and improve accuracy.

Transformer fault diagnosis plays a critical role in ensuring the reliability and safety of power systems. Traditional diagno stic methods often suffer from limited feature extraction capability and low adaptability to complex operating conditions. To address these issues, this paper proposes a transformer fault diagnosis approach based on an improved convolutional neural network (CNN). The proposed model introduces an optimized network structure with enhanced feature extraction layers and an adaptive pooling mechanism to capture both local and global fault characteristics from dissolved gas analysis (DGA) and vibration signal datasets. Experimental results on benchmark datasets demonstrate that the improved CNN achieves higher accuracy and robustness compared with conventional CNN and support vector machine (SVM) methods, especially under noisy and unbalanced data scenarios. The proposed approach provides an effective and scalable solution for real-time transformer fault monitoring, contributing to the advancement of intelligent substations and the reliable operation of modern power grids.

Conventional convolutional neural networks (CNNs) predominantly emphasize spatial features of signals and often fall short in prioritizing sequential features. As the number of layers increases, they are prone to issues such as vanishing or exploding gradients, leading to training instability and subsequent erratic fluctuations in loss values and recognition rates. To address this issue, a novel hybrid model, termed one-dimensional (1D) residual network with attention mechanism and bidirectional gated recurrent unit (BGRU) is developed for rotating machinery fault classification. First, a novel 1D residual network with optimized structure is constructed to obtain spatial features and mitigate the gradient vanishing or exploding. Second, the attention mechanism (AM) is designed to catch important impact characteristics for fault samples. Next, temporal features are mined through the BGRU. Finally, feature information is summarized through global average pooling, and the fully connected layer is utilized to output the final classification result for rotating machinery fault diagnosis. The developed technique which is tested on one set of planetary gear data and three different sets of bearing data, has achieved classification accuracy of 98.5%, 100%, 100%, and 100%, respectively. Compared with other methods, including CNN, CNN-BGRU, CNN-AM, and CNN with an AM-BGRU, the proposed technique has the highest recognition rate and stable diagnostic performance.

Leukemia is a very dangerous kind of malignancy troubling the blood or bone marrow in all age categories, both in adults and children. The deadly and threatening kind of leukemia is named Acute Lymphoblastic Leukemia (ALL). The accurate and automated ALL diagnosis of blood cancer is complex work. Medical experts and hematologists in the bone marrow and blood samples detect it by employing a high-quality microscope. The manual classification is observed as tiresome and is restricted by varying expert considerations and other attributes. Presently, the Convolutional Neural Networks (CNNs) have become an acceptable mechanism for analyzing the medical image. However, for attaining outstanding performance, conventional CNNs normally demand large data sources for better training. Thus, to alleviate the existing complexities, we implemented an effective ALL detection system using deep learning. At first, the necessitated images are aggregated from global resources of data. Further, the garnered images are inputted into the Optimized Trans-Res-Unet+ (OTRUnet+)-based segmentation model. Here, the Fitness-aided Position Updating in the Social engineering Algorithm (FPUSA) for improving the segmentation process’s efficacy optimally tunes the OTRUnet+ technique parameters. In addition, the segmented images are taken to perform the classification process using the Adaptive Multi-Dilated Residual Attention Network (AMDRAN); here several parameters are optimally tuned by the same FPUSA to enrich the classification process. Finally, the suggested AMDRAN technique offered the ALL classified output. The effectiveness of the designed ALL detection system is explored with several existing systems to display its enhanced performance over other models

Electric power load forecasting is a crucial link in ensuring the safe, stable, and economical operation of power systems and achieving sustainable resource allocation. This paper reviews the research status and development trends of electric power load forecasting, classifying it into three categories based on time span: short-term, medium-term, and long-term, suitable for daily grid dispatch, monthly/quarterly operational planning, and long-term planning, respectively. Regarding influencing factors, external meteorological conditions and economic policies, as well as internal user behavior patterns and power system dynamic characteristics, all significantly impact load forecasting. Traditional statistical methods such as time series analysis, regression analysis, and exponential smoothing once dominated but are limited by model assumptions and their ability to handle nonlinear relationships. Machine learning methods like Support Vector Machines (SVM) and Random Forests (RF) overcome traditional limitations through feature interaction and data fusion capabilities. Deep learning methods such as Convolutional Neural Networks (CNN), Transformers, and Graph Neural Networks (GNN) further enable high-dimensional modeling of complex spatiotemporal patterns, driving prediction accuracy to new heights. Future research directions focus on cross-domain collaborative forecasting, breaking disciplinary barriers to build multimodal fusion networks, and responding to carbon neutrality goals by developing net load forecasting models adapted to new power system demands and prediction models coupled with carbon credit mechanisms. In summary, power load forecasting technology is shifting from traditional approaches towards multidisciplinary integrated innovation and will become the core engine for intelligent decision-making in new power systems, aiding the safe, economical, and low-carbon transformation of the grid.

Misfiring in multi-cylinder is caused due to abnormal combustion of the engine. Cylinders produce less power if engine misfires. Engine misfires generate unburned hydrocarbons and increase the load on the catalytic converter. The job of the catalytic converter is to convert harmful pollutants into less harmful emissions before they ever leave the vehicle exhaust system. If the misfire rate is excessive, the catalytic converter can overheat and be damaged. Hence engine misfire becomes a severe issue and should be detected in early stages. During misfire and at normal combustion, every cylinder has its own unique vibration spectrum this will be a very useful phenomenon for misfire detection. In this work, vibration signals are recorded using a piezoelectric accelerometer for normal combustion as well as combustion with a misfire for each cylinder. The obtained vibration signals are used for training and testing of conventional machine learning algorithm as well as the deep learning algorithm. Support Vector Machine (SVM) with statistical feature extraction as conventional machine learning algorithm and Convolutional Neural Network (CNN) as deep learning algorithm are used for misfire detection in multi-cylinder. In this paper, the evaluation parameters of classifiers such as overall accuracy, precision, recall etc. are compared. Deep learning algorithm i.e. CNN without feature Extraction was found to be performing better for multiclass misfire detection compared to CNN with feature extraction and SVM.

Image and video analysis using graph neural network for Internet of Medical Things and computer vision applications

Cardiotocography (CTG) is a method of monitoring fetal heart rate and uterine contractions during pregnancy. CTG is a methodology used to measure the fetal well-being in a pregnant woman. The objective of the paper is to reduce the fetal mortality. Data is being evaluated by applying pre-processing techniques, followed by Convolutional Neural Network (CNN) and dimensionality reduction using Principal Component Analysis (PCA). The approach adopted in the proposed method for detecting the fetal heart rate is evaluated using two methods, namely, conventional CNN and conventional CNN integrated with PCA. Using CNN algorithm around 77% has been achieved and CNN integrated with .PCA gave around 95% accuracy.

Conventional Convolutional Neural Networks (CNNs), which are realized in spatial domain, exhibit high computational complexity. This results in high resource utilization and memory usage and makes them unsuitable for implementation in resource and energy-constrained embedded systems. A promising approach for low-complexity and high-speed solution is to apply CNN modeled in the spectral domain. One of the main challenges in this approach is the design of activation functions. Some of the proposed solutions perform activation functions in spatial domain, necessitating multiple and computationally expensive spatial-spectral domain switching. On the other hand, recent work on spectral activation functions resulted in very computationally intensive solutions. This paper proposes a complex-valued activation function for spectral domain CNNs that only transmits input values that have positive-valued real or imaginary component. This activation function is computationally inexpensive in both forward and backward propagation and provides sufficient nonlinearity that ensures high classification accuracy. We apply this complex-valued activation function in a LeNet-5 architecture and achieve an accuracy gain of up to 7% for MNIST and 6% for Fashion MNIST dataset, while providing up to 79% and 85% faster inference times, respectively, over state-of-the-art activation functions for spectral domain.

Cardiac Magnetic Resonance (CMR) imaging is widely used in the clinic to assess the patient-specific cardiac structure and function. However, the manual analysis of the CMR data is tedious and subjective. In this work, we developed a fully automatic segmentation system for the left ventricle (LV) myocardium from MR cine images. The system consists of three major components. Firstly, a conventional convolutional neural network (CNN) was trained to detect the global region of interest (ROI) of LV. Secondly, a novel multi-scope CNN was proposed to segment the LV myocardium from the reduced ROI, taking advantage of the image context in different scopes, such that both the local accuracy and global consistency can be implicitly learned by the CNN. Finally the results were pruned with simple morphological filtering preserving the largest component. With a relatively small training set of 200 MR cine images, the method achieved an average segmentation accuracy of 0.71 as expressed by the Dice overlap index. The proposed method can be applied to automatically segment the LV from MR images with the reasonable accuracy, or as a proper initialization for local shape methods to achieve further refined results.

Graph Neural Networks (GNNs) have proven effective and typically benefit from pre-training on accessible graphs to enhance performance on tasks with limited labeled data. However, existing GNNs are constrained by the ``one-domain-one-model'' limitation, which restricts their effectiveness across diverse graph domains. In this paper, we tackle this problem by developing a method called Multi-Domain Pre-training for a Unified GNN Model (MDP-GNN). This method is based on the philosophical notion that everything is interconnected, suggesting that a latent meta-domain exists to encompass the diverse graph domains and their interconnections. MDP-GNN seeks to identify and utilize this meta-domain to train a unified GNN model through three core strategies. Firstly, it integrates node feature semantics from different domains to create unified representations. Secondly, it employs a bi-level learning strategy to build a domain-synthesized network that identifies latent connections to facilitate cross-domain knowledge transfer. Thirdly, it uses Wasserstein distance to map diverse domains into the common meta-domain for graph distribution alignment. We validate the effectiveness of MDP-GNN through theoretical analysis and extensive experiments on four real-world graph datasets, showing its superiority in enhancing GNN performance across diverse domains.

To accurately predict the coal-fired generating units? thermal efficiency under deep peak shaving conditions, a dynamic prediction method of boiler thermal efficiency is proposed. Determine the auxiliary variables affecting boiler thermal efficiency by analyzing the proportions of heat loss in the anti-balance method, and using the random forest algorithm to carry out supervised dimensionality reduction of auxiliary variables affecting the thermal efficiency. On this basis, the convolutional neural network (CNN) with dynamic modelling function is selected as the infrastructure, and to ensure the lightweight model, the cross-channel communication unit is inserted into the conventional CNN with only three convolutional layers to solve the problem of no interaction of feature maps in the same convolutional layer, and a boiler thermal efficiency predictive model based on the cross-channel communication CNN (C3-CNN) is constructed. Simulation experiments were conducted on the actual operation data for a 600 MW boiler, the results show that the developed method is equally suitable for both transient and steady-state conditions.

BackgroundThis study was conducted to alleviate a common difficulty in chest X-ray image diagnosis: The attention region in a convolutional neural network (CNN) does not often match the doctor’s point of focus. The method presented herein, which guides the area of attention in CNN to a medically plausible region, can thereby improve diagnostic capabilities.MethodsThe model is based on an attention branch network, which has excellent interpretability of the classification model. This model has an additional new operation branch that guides the attention region to the lung field and heart in chest X-ray images. We also used three chest X-ray image datasets (Teikyo, Tokushima, and ChestX-ray14) to evaluate the CNN attention area of interest in these fields. Additionally, after devising a quantitative method of evaluating improvement of a CNN’s region of interest, we applied it to evaluation of the proposed model.ResultsOperation branch networks maintain or improve the area under the curve to a greater degree than conventional CNNs do. Furthermore, the network better emphasizes reasonable anatomical parts in chest X-ray images.ConclusionsThe proposed network better emphasizes the reasonable anatomical parts in chest X-ray images. This method can enhance capabilities for image interpretation based on judgment.

This study reports results of the Extreme Gradient Boosting (XGBoost) algorithm in comparison to Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN) to generate cloud mask comprising of six classes: low cloud, high cloud, cloud shadow, ground, snow, and water. Here, Baetens-Hagolle dataset and WHUS2-CD dataset was considered. Various texture features derived using GLCM, morphological profile analysis, bilateral filtering, and deep features derived using Residual Network (Resnet) were used in combination with spectral information. The first phase of the study suggested both XGBoost and RF exhibit comparable performance for both traditional texture features and deep features, the second phase highlighted that XGBoost showed better generalization capabilities with respect to the different environmental conditions, and finally, comparison with threshold-based methods (Function of Mask [Fmask], Sen2cor, and Maccs-Atcor Joint Method [MAJA]) suggested that XGBoost and RF based cloud detection is a good alternative of existing state-of-the-art cloud detection.

This study presents a novel approach using Wasserstein Generative Adversarial Networks with Gradient Penalty (WGAN-GP) to generate synthetic electroencephalography (EEG) and electrocardiogram (ECG) waveforms. The synthetic EEG data represent concentration and relaxation mental states, while the synthetic ECG data correspond to normal and abnormal states. By addressing the challenges of limited biophysical data, including privacy concerns and restricted volunteer availability, our model generates realistic synthetic waveforms learned from real data. Combining real and synthetic datasets improved classification accuracy from 92% to 98.45%, highlighting the benefits of dataset augmentation for machine learning performance. The WGAN-GP model achieved 96.84% classification accuracy for synthetic EEG data representing relaxation states and optimal accuracy for concentration states when classified using a fusion of convolutional neural networks (CNNs). A 50% combination of synthetic and real EEG data yielded the highest accuracy of 98.48%. For EEG signals, the real dataset consisted of 60-s recordings across four channels (TP9, AF7, AF8, and TP10) from four individuals, providing approximately 15,000 data points per subject per state. For ECG signals, the dataset contained 1200 real samples, each comprising 140 data points, representing normal and abnormal states. WGAN-GP outperformed a basic generative adversarial network (GAN) in generating reliable synthetic data. For ECG data, a support vector machine (SVM) classifier achieved an accuracy of 98% with real data and 95.8% with synthetic data. Synthetic ECG data improved the random forest (RF) classifier’s accuracy from 97% with real data alone to 98.40% when combined with synthetic data. Statistical significance was assessed using the Wilcoxon signed-rank test, demonstrating the robustness of the WGAN-GP model. Techniques such as discrete wavelet transform, downsampling, and upsampling were employed to enhance data quality. This method shows significant potential in addressing biophysical data scarcity and advancing applications in assistive technologies, human-robot interaction, and mental health monitoring, among other medical applications.