Residual Connections Research Articles

Artificial Intelligence-Based Segmentation and Classification of Plant Images with Missing Parts and Fractal Dimension Estimation

Existing research on image-based plant classification has demonstrated high performance using artificial intelligence algorithms. However, limited camera viewing angles can cause parts of the plant to be invisible in the acquired images, leading to an inaccurate classification. However, this issue has not been addressed by previous research. Hence, our study aims to introduce a method to improve classification performance by taking these limitations into account; specifically, we incorporated both segmentation and classification networks structured as shallow networks to expedite the processing times. The proposed shallow plant segmentation network (Shal-PSN) performs adversarial learning based on a discriminator network; and a shallow plant classification network (Shal-PCN) with applied residual connections was also implemented. Moreover, the fractal dimension estimation is used in this study for analyzing the segmentation results. Additionally, this study evaluated the performance of the proposed Shal-PSN that achieved the dice scores (DSs) of 87.43% and 85.71% with PlantVillage and open leaf image (OLID-I) open datasets, respectively, in instances where 40–60% of plant parts were missing. Moreover, the results demonstrate that the proposed method increased the classification accuracy from 41.16% to 90.51% in the same instances. Overall, our approach achieved superior performance compared to the existing state-of-the-art classification methods.

Multimodal Connectivity‐Guided Glioma Segmentation From Magnetic Resonance Images via Cascaded 3D Residual U‐Net

ABSTRACTGlioma is a type of brain tumor with a high mortality rate. Magnetic resonance imaging (MRI) is commonly used for examination, and the accurate segmentation of tumor regions from MR images is essential to computer‐aided diagnosis. However, due to the intrinsic heterogeneity of brain glioma, precise segmentation is very challenging, especially for tumor subregions. This article proposed a two‐stage cascaded method for brain tumor segmentation that considers the hierarchical structure of the target tumor subregions. The first stage aims to identify the whole tumor (WT) from the background area; and the second stage aims to achieve fine‐grained segmentation of the subregions, including enhanced tumor (ET) region and tumor core (TC) region. Both stages apply a deep neural network structure combining modified 3D U‐Net with a residual connection scheme to tumor region and subregion segmentation. Moreover, in the training phase, the 3D masks generation of subregions with potential incomplete connectivity are guided by the completely connected regions. Experiments were performed to evaluate the performance of the methods on both area and boundary accuracy. The average dice score of the WT, TC, and ET regions on BraTS 2020 dataset is 0.9168, 0.0.8992, 0.8489, and the Hausdorff distance is 6.021, 9.203, 12.171, respectively. The proposed method outperforms current works, especially in segmenting fine‐grained tumor subregions.

A Non-Contact AI-Based Approach to Multi-Failure Detection in Avionic Systems

The increasing electrification and integration of advanced controls in modern aircraft designs have significantly raised the number and complexity of installed printed circuit boards (PCBs), posing new challenges for efficient maintenance and rapid failure detection. Despite self-diagnostic features in current avionics systems, circuit damage and multiple simultaneous failures may arise, compromising safety and diagnostic accuracy. To address these challenges, this paper aims to develop a fast, accurate, and non-destructive, multi-failure diagnosis algorithm for PCBs. The proposed method combines a self-attention mechanism with an adaptive graph convolutional neural network to enhance diagnostic precision. A convolutional neural network with residual connections extracts features from scalar magnetic field data, ensuring robust input diversity. The model was tested on a typical dual-phase amplitude boosting circuit with up to four different simultaneous failures, achieving the experimental results of 99.08%, 98.50%, 98.78%, 98.01%, 98.93%, 98.25%, 97.03%, and 99.77% across metrics including overall precision, per-class precision, overall recall, per-class recall, overall F1 measure, and per-class F1 measure. The results demonstrated its effectiveness and feasibility in diagnosing complex PCBs with multiple failures, indicating the algorithm’s potential to improve failure diagnosis performance and offer a promising PCB diagnosis solution in aerospace applications.

Enhancing spinal MRI segmentation with an asymmetric U-Net architecture

Spinal diseases are among the most prevalent health issues in modern society, significantly impacting patients’ quality of life. Diagnosing conditions such as disc herniation and spinal deformity requires advanced medical imaging techniques, including X-rays, magnetic resonance imaging (MRI), computed tomography, and nuclear magnetic resonance. Spine MRI is particularly crucial due to its ability to provide high-resolution images of soft tissues, essential for accurate diagnosis. However, the manual segmentation of spine MRI images is labor-intensive and inadequate for large-scale quantitative analysis. Thus, developing automated spinal MRI segmentation methods is critical to alleviating doctors’ workload and enhancing diagnostic efficiency. In this study, we propose a novel asymmetric U-Net architecture designed to improve the precision of reconstructing complex structures and details by increasing the depth of the upsampling side. The model incorporates adjacent-scale skip connections to control parameters while maintaining high segmentation accuracy. In addition, residual connections on the upsampling side prevent gradient vanishing, thereby enhancing the network’s feature learning and representation capabilities. Experimental results indicate that this method significantly reduces training time and increases model accuracy compared to traditional approaches, marking a substantial advancement in automated spinal MRI segmentation. This innovative approach holds promise for improving clinical outcomes and optimizing the workflow in medical imaging departments.

Interweaving Insights: High-Order Feature Interaction for Fine-Grained Visual Recognition

AbstractThis paper presents a novel approach for Fine-Grained Visual Classification (FGVC) by exploring Graph Neural Networks (GNNs) to facilitate high-order feature interactions, with a specific focus on constructing both inter- and intra-region graphs. Unlike previous FGVC techniques that often isolate global and local features, our method combines both features seamlessly during learning via graphs. Inter-region graphs capture long-range dependencies to recognize global patterns, while intra-region graphs delve into finer details within specific regions of an object by exploring high-dimensional convolutional features. A key innovation is the use of shared GNNs with an attention mechanism coupled with the Approximate Personalized Propagation of Neural Predictions (APPNP) message-passing algorithm, enhancing information propagation efficiency for better discriminability and simplifying the model architecture for computational efficiency. Additionally, the introduction of residual connections improves performance and training stability. Comprehensive experiments showcase state-of-the-art results on benchmark FGVC datasets, affirming the efficacy of our approach. This work underscores the potential of GNN in modeling high-level feature interactions, distinguishing it from previous FGVC methods that typically focus on singular aspects of feature representation. Our source code is available at https://github.com/Arindam-1991/I2-HOFI.

CNCAN: Contrast and normal channel attention network for super-resolution image reconstruction of crops and weeds

Numerous studies have been performed to apply camera vision technologies in robot-based agriculture and smart farms. In particular, to obtain high accuracy, it is essential to procure high-resolution (HR) images, which requires a high-performance camera. However, due to high costs it is difficult to widely apply the camera in agricultural robots. To overcome this limitation, we propose contrast and normal channel attention network (CNCAN) for super-resolution reconstruction (SR), which is the first research for the accurate semantic segmentation of crops and weeds even with low-resolution (LR) images captured by low-cost and LR camera. Attention block and activation function that considers high frequency and contrast information of images are used in CNCAN, and the residual connection method is applied to improve the learning stability.As a result of experimenting with three open datasets, namely, Bonirob, rice seedling and weed, and crop/weed field image (CWFID) datasets, the mean intersection of union (MIOU) results of semantic segmentation for crops and weeds with SR images through CNCAN were 0.7685, 0.6346, and 0.6931 in the Bonirob, rice seedling and weed, and CWFID datasets, respectively, confirming higher accuracy than other state-of-the-art methods for SR.

Contour-constrained branch U-Net for accurate left ventricular segmentation in echocardiography.

Using echocardiography to assess the left ventricular function is one of the most crucial cardiac examinations in clinical diagnosis, and LV segmentation plays a particularly vital role in medical image processing as many important clinical diagnostic parameters are derived from the segmentation results, such as ejection function. However, echocardiography typically has a lower resolution and contains a significant amount of noise and motion artifacts, making it a challenge to accurate segmentation, especially in the region of the cardiac chamber boundary, which significantly restricts the accurate calculation of subsequent clinical parameters. In this paper, our goal is to achieve accurate LV segmentation through a simplified approach by introducing a branch sub-network into the decoder of the traditional U-Net. Specifically, we employed the LV contour features to supervise the branch decoding process and used a cross attention module to facilitate the interaction relationship between the branch and the original decoding process, thereby improving the segmentation performance in the region LV boundaries. In the experiments, the proposed branch U-Net (BU-Net) demonstrated superior performance on CAMUS and EchoNet-dynamic public echocardiography segmentation datasets in comparison to state-of-the-art segmentation models, without the need for complex residual connections or transformer-based architectures. Our codes are publicly available at Anonymous Github https://anonymous.4open.science/r/Anoymous_two-BFF2/ .

Research on Speech Emotion Recognition Method Based on ResSE_CNN1D

This article examines a speech emotion recognition (SER) technique based on the enhanced one-dimensional convolution neural network ResSE_CNN1D. With the artificial intelligence developing rapidly, SER has a profound impact in many areas. The model of this article is used to extract the characteristics of the input data through opensmile, and is sent into the ResSE_CNN1D model, which is eventually classified by the softmax activation function and obtains the final results. The key to this model is efficient learning of decimal sets and the rapid deployment in resource-constrained environments. The ResSE_CNN1D model improves the performance of the model by adding the residual connection and the SE module on the basis of cnn1d. This increasing the accuracy of the recognition and preventing the fitting problem. After the model was created, the study adopted the audio sampling and training of the casia data concentration. The final accuracy was 0.900, which increased the accuracy of 2.9 percent compared to the cnn1d method. And by the analysis of the relationship diagram of the confusion matrix and the accuracy and loss rate relative to the number of training, the model has a high robustness and effectively prevents the appearance of the fitting problem. And it also can achieve high precision and achieve a lightweight goal relative to less training.

Comparative Analysis of Deep Learning Models for Building Extraction from High-resolution Satellite Imagery

In this research, an approach to extract buildings from Google's satellite imagery was proposed. The performances of various deep learning models (U-Net, RIU-Net, U-Net++, Res-U-Net, and DeepLabV3+) on pre-processed datasets were compared. The models were trained using the similarity metrics of Intersection over Union (IoU) and Dice Similarity Coefficient (DSC). The best-performing models among the segmentation techniques were Res-U-Net and DeepLabV3+. Res-U-Net, an enhanced version of the traditional U-Net model that incorporates residual connections for improved feature propagation, achieved an F1 score of 85.43% when using the RGB dataset. Similarly, DeepLabV3+ also achieved high performance on the Enhanced RGB dataset, obtaining an F1 score of 85.18% after applying pre-processing techniques. This research highlights the significance of color as a dominant feature for accurate building extraction from satellite images. The findings contribute to improved methodologies for building identification, benefiting urban planning, and disaster management applications.

Chain-aware graph neural networks for molecular property prediction.

Predicting the properties of molecules is a fundamental problem in drug design and discovery, while how to learn effective feature representations lies at the core of modern deep-learning-based prediction methods. Recent progress shows expressive power of graph neural networks (GNNs) in capturing structural information for molecular graphs. However, we find that most molecular graphs exhibit low clustering along with dominating chains. Such topological characteristics can induce feature squashing during message passing and thus impair the expressivity of conventional GNNs. Aiming at improving node features' expressiveness, we develop a novel chain-aware graph neural network model, wherein the chain structures are captured by learning the representation of the center node along the shortest paths starting from it, and the redundancy between layers are mitigated via initial residual difference connection (IRDC). Then the molecular graph is represented by attentive pooling of all node representations. Compared to standard graph convolution, our chain-aware learning scheme offers a more straightforward feature interaction between distant nodes, thus it is able to capture the information about long-range dependency. We provide extensive empirical analysis on real-world datasets to show the outperformance of the proposed method. The MolPath code is publicly available at https://github.com/Assassinswhh/Molpath.

UTran-DSR: a novel transformer-based model using feature enhancement for dysarthric speech recognition

Over the past decade, the prevalence of neurological diseases has significantly risen due to population growth and aging. Individuals suffering from spastic paralysis, brain attack, and idiopathic Parkinson’s disease (PD), among other neurological illnesses, commonly suffer from dysarthria. Early detection and treatment of dysarthria in these patients are essential for effectively managing the progression of their disease. This paper provides UTrans-DSR, a novel encoder-decoder architecture for analyzing Mel-spectrograms (generated from audios) and classifying speech as healthy or dysarthric. Our model employs transformer encoder features based on a hybrid design, which includes the feature enhancement block (FEB) and the vision transformer (ViT) encoders. This combination effectively extracts global and local pixel information regarding localization while optimizing the mel-spectrograms feature extraction process. We keep up with the original class-token grouping sequence in the vision transformer while generating a new equivalent expanding route. More specifically, two unique growing pathways use a deep-supervision approach to increase spatial data recovery and expedite model convergence. We add consecutive residual connections to the system to reduce feature loss while increasing spatial data retrieval. Our technique is based on identifying gaps in mel-spectrograms distinguishing between normal and dysarthric speech. We conducted several experiments on UTrans-DSR using the UA speech and TORGO datasets, and it outperformed the existing top models. The model performed significantly in pixel’s localized and spatial feature extraction, effectively detecting and classifying spectral gaps. The Tran-DSR model outperforms previous research models, achieving an accuracy of 97.75%.

Optimisation of U-Net Semantic Segmentation Model Based on Residual Connectivity and Attention Mechanisms

Optimisation of U-Net Semantic Segmentation Model Based on Residual Connectivity and Attention Mechanisms

A semantic segmentation method integrated convolutional nonlinear spiking neural model with Transformer

Semantic segmentation is a critical task in computer vision, with significant applications in areas like autonomous driving and medical imaging. Transformer-based methods have gained considerable attention recently because of their strength in capturing global information. However, these methods often sacrifice detailed information due to the lack of mechanisms for local interactions. Similarly, convolutional neural network (CNN) methods struggle to capture global context due to the inherent limitations of convolutional kernels. To overcome these challenges, this paper introduces a novel Transformer-based semantic segmentation method called NSNPFormer, which leverages the nonlinear spiking neural P (NSNP) system—a computational model inspired by the spiking mechanisms of biological neurons. The NSNPFormer employs an encoding–decoding structure with two convolutional NSNP components and a residual connection channel. The convolutional NSNP components facilitate nonlinear local feature extraction and block-level feature fusion. Meanwhile, the residual connection channel helps prevent the loss of feature information during the decoding process. Evaluations on the ADE20K and Pascal Context datasets show that NSNPFormer achieves mIoU scores of 53.7 and 58.06, respectively, highlighting its effectiveness in semantic segmentation tasks.

Sequential Recommendation Model Based on Deep Residual Recurrent Neural Network

In order to solve the problem that the sequence recommendation model based on RNN (Recurrent Neural Network) is prone to gradient vanishing or exploding when processing long sequences, which leads to the instability of the recommendation model training process, residual connection, layer normalization and feedforward neural network modules are introduced on the basis of the traditional gated recurrent unit (GRU), and a sequence recommendation model DeepGRU based on deep residual recurrent neural network is proposed. It is verified on 3 public datasets. The experimental results show that the DeepGRU has obvious advantages over the most advanced sequence recommendation methods (the average recommendation accuracy is improved 8.68%). The ablation experiment verifies the effectiveness of the introduced residual connection and other modules under the DeepGRU framework. In addition, the DeepGRU effectively alleviates the problem of unstable training process when processing long sequences.

Outcomes and complications of vertical parasagittal hemispherotomy in children: a nationwide population-based study.

The goal of this study was to assess the complications associated with vertical parasagittal hemispherotomy (VPH), the impact of incomplete disconnection on long-term seizure freedom, and how VPH impacts cognitive development. A retrospective evaluation was performed in all patients who had undergone VPH during 1991-2022 at the authors' institution. Two-year follow-up data were available for 45 patients, and there were 6-month data for 3 more. All available postoperative MRI studies (31/48, 64.6%) were reviewed. Before 2010, postoperative MRI was only performed if seizures recurred. Primary VPH led to Engel class I in 73% of patients. Acquired etiologies had a higher rate of Engel I compared to developmental and progressive etiologies (96% vs 46% and 44%, p < 0.001). Nearly half of patients (45%) showed improved cognitive trajectories as opposed to their preoperative ones, whereas in 45% trajectories remained unchanged. Additionally, 5 patients (10%) exhibited new major deficits or accelerated cognitive deterioration after VPH. Surgical complications occurred in 14 patients (29%) after the first VPH; 4 cases were classified as transient, resolving during follow-up without surgical intervention. Nontransient complications included 8 cases of hydrocephalus requiring surgical treatment, 1 shunted subdural hygroma, and 1 case of CSF leakage from the wound. Diabetes insipidus occurred in 6 patients, with all resolving spontaneously. Residual connections were present in 16 patients, primarily in the temporomesial region. Seven patients remained seizure free despite visible residual connections. VPH is a highly effective treatment for drug-resistant hemispheric epilepsy, resulting in durable seizure freedom and often favorable cognitive outcomes. Diabetes insipidus in addition to hydrocephalus is a common complication after VPH. Incomplete disconnection does not necessarily preclude seizure freedom.

An Accurate and Explainable Residuum Inspection Network With Residual Block Connection and Hybrid Attention Fusion

An Accurate and Explainable Residuum Inspection Network With Residual Block Connection and Hybrid Attention Fusion

Enhancing the analysis of murine neonatal ultrasonic vocalizations: Development, evaluation, and application of different mathematical models.

Rodents employ a broad spectrum of ultrasonic vocalizations (USVs) for social communication. As these vocalizations offer valuable insights into affective states, social interactions, and developmental stages of animals, various deep learning approaches have aimed at automating both the quantitative (detection) and qualitative (classification) analysis of USVs. So far, no notable efforts have been made to determine the most suitable architecture. We present the first systematic evaluation of different types of neural networks for USV classification. We assessed various feedforward networks, including a custom-built, fully-connected network, a custom-built convolutional neural network, several residual neural networks, an EfficientNet, and a Vision Transformer. Our analysis concluded that convolutional networks with residual connections specifically adapted to USV data, are the most suitable architecture for analyzing USVs. Paired with a refined, entropy-based detection algorithm (achieving recall of 94.9 % and precision of 99.3 %), the best architecture (achieving 86.79 % accuracy) was integrated into a fully automated pipeline capable of analyzing extensive USV datasets with high reliability. In ongoing projects, our pipeline has proven to be a valuable tool in studying neonatal USVs. By comparing these distinct deep learning architectures side by side, we have established a solid foundation for future research.

AGNN: Alternating Graph-Regularized Neural Networks to Alleviate Over-Smoothing.

Graph convolutional network (GCN) with the powerful capacity to explore graph-structural data has gained noticeable success in recent years. Nonetheless, most of the existing GCN-based models suffer from the notorious over-smoothing issue, owing to which shallow networks are extensively adopted. This may be problematic for complex graph datasets because a deeper GCN should be beneficial to propagating information across remote neighbors. Recent works have devoted effort to addressing over-smoothing problems, including establishing residual connection structure or fusing predictions from multilayer models. Because of the indistinguishable embeddings from deep layers, it is reasonable to generate more reliable predictions before conducting the combination of outputs from various layers. In light of this, we propose an alternating graph-regularized neural network (AGNN) composed of graph convolutional layer (GCL) and graph embedding layer (GEL). GEL is derived from the graph-regularized optimization containing Laplacian embedding term, which can alleviate the over-smoothing problem by periodic projection from the low-order feature space onto the high-order space. With more distinguishable features of distinct layers, an improved Adaboost strategy is utilized to aggregate outputs from each layer, which explores integrated embeddings of multi-hop neighbors. The proposed model is evaluated via a large number of experiments including performance comparison with some multilayer or multi-order graph neural networks, which reveals the superior performance improvement of AGNN compared with the state-of-the-art models.

Low‐light image enhancement via lightweight custom non‐linear transform network

AbstractConvolutional neural network (CNN)‐based models have shown significant progress in low light image enhancement. However, many existing models possess a large number of parameters, making them unsuitable for deployment on terminal devices. Moreover, adjustments to brightness, contrast, and colour in images are often non‐linear, and convolution is not the best at capturing complex non‐linear relationships in image data. To address these issues, a model based on an end‐to‐end custom non‐linear transform network (CNTNet) is proposed. CNTNet combines a custom non‐linear transform layer with CNN layers to achieve image contrast and detail enhancement. The CNT layer in this model introduces transformation parameters at multiple scales to manipulate input images within various ranges. CNTNet progressively processes images by stacking multiple non‐linear transform layers and convolutional layers while integrating residual connections to capture and leverage subtle image features. The final output is generated through convolutional layers to obtain enhanced images. Experimental results of CNTNet demonstrate that, while maintaining a comparable level of image quality evaluation metrics to mainstream models, it significantly reduces the parameter count to only 2K.

Improving lithofacies prediction in lacustrine shale by combining deep learning and well log curve morphology in Sanzhao Sag, Songliao Basin, China

The accurate identification of shale lithofacies is crucial for characterizing the hydrocarbon potential of lacustrine shale oil reservoirs. Petrophysical logging, serving as an effective tool for acquiring subsurface lithofacies information, provides a convenient and reliable lithofacies identification solution. Deep learning technology, capable of adapting to the nonlinearity and non-stationarity inherent in geological statistics, exhibits unique advantages in conventional reservoir lithofacies prediction. However, the lithofacies of lacustrine shale formations undergo rapid spatial and temporal changes, rendering lithofacies prediction more complex compared to conventional reservoirs. In this study, the lower Qingshankou member in the Sanzhao Sag was selected as the research target, and the Deep Residual Shrinkage Network (DRSN), known for its ability to handle complex nonlinear relationships and mitigate the effects of noisy data through residual connections and shrinkage mechanisms, was employed as a deep learning framework for predicting lithofacies in lacustrine shale formations for the first time. Well logging data, including natural gamma ray (GR), acoustic (AC), deep investigate double lateral resistivity log (RD), shallow investigate double lateral resistivity log (RS), and corrected compensated neutron log (CNC), were used as input features for the model. The results indicate that the DRSN model achieves an accuracy of 76.3% in predicting lithofacies in lacustrine shale formations. However, the DRSN model still exhibits shortcomings in capturing lithofacies change information. To enhance the model's ability to identify lithofacies change interfaces, this study further explicitly introduces Well Logging Curve Morphological Features (WLCM) as additional features and establishes a recognition method combining DRSN with WLCM. The combined DRSN-WLCM model was validated using a separate test dataset, demonstrating an improved accuracy of 85.5%, using the five well logging attributes and the derivative of the AC as inputs. Furthermore, the study reveals the lithofacies spatial distribution characteristics of the lower Qingshankou member in the Sanzhao Sag. This method can be widely applied to lithofacies delineation in lacustrine shale formations and similar stratigraphic units.