Drug repositioning, exploring new indications for existing drugs, is emerging as a promising approach to accelerate drug discovery and reduce research risk of failure. Recent advances in this topic by applying graph neural networks have enabled researches to achieve significant results by extracting latent features from the original data. However, the previous studies have not fully considered the distinctive information embedded within different construction graphs, which may lead to insufficient classification performance due to the lack of more detailed features. This work therefore proposes a novel approach, namely MVGF DR, which leverages graph network construction and multi view graph feature fusion for drug repositioning. MVGF-DR built a comprehensive graph network from both similarity and association information, i.e., a similarity graph network is constructed with drug-drug and disease-disease similarities where similarity information are extracted by graph isomorphism networks, and an association graph network with drug-disease associations where drug-disease relationships are explored by graph convolutional networks. Additionally, a maximum value selection strategy is introduced to filter features from different channels for feature fusion and noise reduction. The average AUROC and AUPR achieved by MVGF-DR across the three datasets reached 95.38% and 51.20%, respectively, outperforming the other five state-of-the-art models. Multiple experiments further also demonstrated the flexibility and practical applicability of MVGF-DR.

Drug target interactions (DTIs) play a crucial role in drug discovery and development. The prediction of DTIs based on computational method can effectively assist the experimental techniques for DTIs identification, which are time-consuming and expensive. However, the current computational models suffer from low accuracy and high false positive rate in the prediction of DTIs, especially for datasets with extremely unbalanced sample categories. To accurately identify the interaction between drugs and target proteins, a variety of descriptors that fully show the characteristic information of drugs and targets are extracted and applied to the integrated method random forest (RF) in this work. Here, the random projection method is adopted to reduce the feature dimension such that simplify the model calculation. In addition, to balance the number of samples in different categories, a down sampling method NearMiss (NM) which can control the number of samples is used. Based on the gold standard datasets (nuclear receptors, ion channel, GPCRs and enzymes), the proposed method achieves the auROC of 92.26%, 98.21%, 97.65%, 99.33%, respectively. The experimental results show that the proposed method yields significantly higher performance than that of state-of-the-art methods in predicting drug target interaction.

Surface defect detection is critical to the quality assurance of hot-rolled strip. Existing detection models suffer from high miss rates due to size differences and low contrast of defects, and the calculation is complex, resulting in slow inspection speed. To address these issues, a novel fast and high-precision detection network (FPDNet) based on YOLOv8 is proposed. Firstly, FPDNet employs High Performance GPU Net-V2 (HGNetv2) as the backbone network, exploiting its excellent feature extraction capabilities and combining it with GhostConv to lighten the model. Secondly, the Csp Efficient Deformable Layer Attention Module (CEDLAM) is proposed to dynamically perceive spatial information, adaptively capture variations in target shape and scale, and enhance multilinear feature relationships to focus on critical regions of the target. Finally, The SIoU loss function is employed to speed up model convergence and enhance the accuracy of localization. Experimental results on the NEU-DET dataset indicate an 80.8% mAP@IoU=0.5, 5.7% higher than the baseline, with a detection speed of 107.3 FPS, balancing accuracy and speed to meet practical use requirements. The dataset and code are available at https://github.com/jianghhhhhh/FPDNet .

The surface defect of hot-rolled strip is a significant factor that impacts the performance of strip products. In recent years, convolutional neural networks (CNNs) have been extensively used in strip surface defect recognition to ensure product quality. However, the existing CNNs-based methods confront the challenges of high complexity, difficult deployment and slow inference speed. Accordingly, this work proposes a soft optimization knowledge distillation (SOKD) scheme to distill the ResNet-152 large model and extract a compact strip surface recognition model. The SOKD scheme utilizes Kullback-Leibler (KL) divergence to minimize the error between the soft probability distributions of the student network and the teacher network, and gradually reduces the weight of "Hard loss" during the training process. The operation significantly reduces the learning constraints that the prior knowledge of the teacher network on the student network in the original KD, which improves the recognition performance of the model. Additionally, SOKD is applicable to most CNNs for identify surface defect of hot-rolled strip. The experimental results on NEU-CLS dataset show that the SOKD outperforms state-of-the-art methods.

Real-time object detectors deployed on general-purpose graphics processing units (GPUs) or embedded devices allow their mass usage in industrial applications at an affordable cost. However, existing state-of-the-art object detectors are difficult to meet the requirements of high accuracy and low inference latency simultaneously in industrial applications on general-purpose devices. In this work, we propose RDDPA, a fast and accurate defect detection framework. RDDPA adopts a novel end-to-end pruning scheme, which can prune the detection network from scratch and achieve real-time detection on general-purpose devices. Additionally, we have developed a new training scheme to minimize the accuracy loss associated with the pruning process. Experimental results on a standard steel surface defect dataset indicate that our model achieves 79.2% mAP (mean Average Precision) at 103.7 FPS (Frames Per Second) on a single mid- end Titan X GPU as well as 40.1 FPS on a single low- end GTX 960M GPU, and outperforms the state-of-the-art defect detectors by about 20x speedup with considerable or higher accuracy.

Accurately and effectively identifying defects in concrete reinforcement is crucial for assessing the integrity and long-term durability of building structures. However, this task is costly and requires significant human effort due to the varying appearance of different concrete materials caused by changing weather conditions and lighting intensities, especially the various uncertain graffiti markings, and the random overlapping of defect types on the material surface. Current methods mainly focus on single, non-overlapping defects, which cannot accurately extract features for classification of overlapping defects, resulting in decreased classification performance. To accurately classify overlapping defects in structural defects with multi-label classification, this work proposes an innovative network that incorporates a Spatial and Channel Attention Reconstruction Module (SCARM), and a Spatial Pyramid Pooling Fast (SPPF), to extract defect-representative features from concrete defect images based on EfficientNetV2. Among them, SCARM contributes to assisting the network focus on crucial features and suppressing unnecessary ones and SPPF used to aggregate multi-scale features to addresses the issue of inconsistent input image scales. Additionally, a novel Asymmetric loss (ASL) is introduced to address the imbalance between positive and negative samples in the dataset. As a result, our method achieves a multi-label accuracy of 79.91

As a high mortality disease, cancer seriously affects people's life and well-being. Reliance on pathologists to assess disease progression from pathological images is inaccurate and burdensome. Computer aided diagnosis (CAD) system can effectively assist diagnosis and make more credible decisions. However, a large number of labeled medical images that contribute to improve the accuracy of machine learning algorithm, especially for deep learning in CAD, are difficult to collect. Therefore, in this work, an improved few-shot learning method is proposed for medical image recognition. In addition, to make full use of the limited feature information in one or more samples, a feature fusion strategy is involved in our model. On the dataset of BreakHis and skin lesions, the experimental results show that our model achieved the classification accuracy of 91.22% and 71.20% respectively when only 10 labeled samples are given, which is superior to other state-of-the-art methods.

Identifying surface defects is an essential task in the hot-rolled process. Currently, various computer vision-based classification and detection methods have achieved superior results in recognizing surface defects. However, defects typically exhibit irregular shapes caused by intra-class differences. Therefore, these two methods are unable to accurately identify the specific locations of the defects. To address this issue, this work proposes a U-shaped Encoder-Decoder framework called Resformer-Unet, which can effectively detect surface defects of hot-rolled strip at the pixel-level. In this framework, the Convolutional Neural Network (CNN) and Transformer work in parallel to extract multi-scale features from the image, which enhances the ability of network to capture both global and local information. Additionally, feature coupling modules are employed to fuse multi-scale features, thereby compensating for the information loss that occurs during down-sampling. On the SD-saliency-900 dataset for strip steel surface defect segmentation, Resformer-Unet achieves a mean Dice Similarity Coefficient (DSC) of 89.96% and an average Hausdorff Distance of 12.03%. These results outperform those of several advanced methods.

The classification method of steel surface defects with high performance and easy to be embedded in the detection equipment is one of the keys to ensure the quality of hot rolled strip. However, the development of deep convolutional neural networks (CNNs) in many real-world applications is largely hindered by their high computational cost, especially in industrial production, although it has good classification accuracy compared with machine learning-based methods in image recognition. Therefore, in this work, we present a lightweight network FCCNet based on the convolutional neural network to facilitate its application in the detection system. To compensate for the accuracy loss caused by the network downsizing, a knowledge distillation (KD) method using a larger trained network (teacher network) to teach a smaller network (student network) is adopted to improve the performance of our model. As a result, our method achieves a classification accuracy of 99.44%, precision of 99.46%, recall of 99.45%, and an F1 score of 99.45% on the NEU-CLS dataset, using only 0.03 MB parameters. These results show that the FCCNet is lighter than other existing classic CNNs with good performance for surface defects classification of hot-rolled steel strip, and it has the potential to be applied in the actual production line.

Printed Circuit Board (PCB) is a significant component of the power system, and their surface defects may hinder electrical performance. Therefore, developing an efficient and precise PCB surface defect detection method is crucial for ensuring the state of the entire power system. In recent years, there has been growing interest in lightweight attention mechanisms that aim to achieve high accuracy with minimal computational cost.In this work, a single-stage object detection network based on YOLO v5m is proposed, which incorporates and compare 3 attention mechanisms to enhance the detection capabilities of the model, namely Coordinate Attention (CA), Convolutional Block Attention Module (CBAM), and Squeeze-and-Excitation (SE), In addition, the evaluation indicator Wise IoU (WIoU) has also been used to replace traditional IoU. Experimental results indicate that the proposed approach achieves mean Average Precision (mAP) of 97.8% and a frame rate of 80.1.Surpassing the performance of other compared models. The proposed approach has the potential to be deployed on edge device in the future.

The quality of concrete is crucial for the safety of facilities. Specifically, the ex-posed surface defects of the bridge seriously affect its strength and aesthetics. However, due to the influence of weather and light, different types of defects on the concrete surface may potentially overlap, making it difficult for classification algorithms to identify concrete surface defects. Traditional recognition methods based on human observation are unreliable and time-consuming, while automatic recognition methods based on computer vision have limitations in identifying multiple defects simultaneously. In this work, a multi-classification network based on improved EfficientNetV2 [1] is proposed to identify multiple defects simultaneously, in which EfficientNetV2 was used as the backbone to ensure the accuracy of feature extraction, and the spatial pyramid pool structure was combined to achieve multiple label classifications [2]. The results show that the accuracy of the concrete defect multi classification network based on EfficientNetV2 reaches 77.6%, with an average classification accuracy of over 94%. This emphasizes the effectiveness of our method in concrete defect recognition.

Corneal ulcer is a common disease located in the eye. If not detected and treated in a timely manner, it is highly likely to cause irreversible damage to the patient’s eyes, and even lead to blindness. Traditional detection methods have drawbacks such as complex steps and painful inspection processes. So there is an urgent need to develop a fast, convenient, and accurate detection method. In this article, we propose an improved Mobile ViT framework for corneal ulcer classification. Our method is based on the original Mobile ViT model, but we integrated an attention module to enhance feature extraction functionality. In addition, we have improved the connection method of the Mobile ViT module by fusing local features with global features to improve feature extraction capabilities and information exchange between images. Subsequent experiments showed that our modifications were effective and feasible. By using these two improvement strategies, the accuracy was first in-creased from 88.7% to 90.8%, and then from 90.8% to 91.5%, while the number of parameters remained basically. Visualize the extracted feature map using Grad-CAM (Gradient-weighted Class Activation Mapping), and the results show that the extracted features exist in the ulcer area.

The Coronavirus Disease 2019 (COVID-19) is the pandemic that has had the greatest impact on world economic development in recent years. Early detection is critical to identify patients with COVID-19, chest x-ray is used for early detection is a rapid, extensive and cost-effective method. The existing technology use deep learning methods, and have achieved very good results. However, the training time of deep learning method is long, and the model size makes it difficult to deploy on hardware system. In this work, we have proposed an attention-based ResNet50v2 network, and taken the network as the teacher network to transfer the knowledge to the student network by knowledge distillation. Thus, the student network has higher accuracy and sensitivity to the positive samples of COVID-19 under the condition of low model parameters, high training speed. The experimental results show that our network of teacher and student have achieved 100% accuracy and sensitivity in both COVID-19 and Normal binary classification. In addition, the accuracy rate of teacher network is 98.20%, the sensitivity is 99.58%, the accuracy rate of student network is 97.68%, the sensitivity is 99.17% in the COVID-19, Viral pneumonia and Normal multiple classification, and the parameters of the student network are only 0.269M.

The computational methods of protein-protein interaction sites prediction can effectively avoid the shortcomings of high cost and time in traditional experimental approaches. However, the serious class imbalance between interface and non-interface residues on the protein sequences limits the prediction performance of these methods. This work therefore proposed a new strategy, NearMiss-based under-sampling for unbalancing datasets and Random Forest classification (NM-RF), to predict protein interaction sites. Herein, the residues on protein sequences were represented by the PSSM-derived features, hydropathy index (HI) and relative solvent accessibility (RSA). In order to resolve the class imbalance problem, an under-sampling method based on NearMiss algorithm is adopted to remove some non-interface residues, and then the random forest algorithm is used to perform binary classification on the balanced feature datasets. Experiments show that the accuracy of NM-RF model reaches 87.6% and 84.3% on Dtestset72 and PDBtestset164 respectively, which demonstrate the effectiveness of the proposed NM-RF method in differentiating the interface or non-interface residues.

Medical images can be accurately segmented to provide reliable basis for clinical diagnosis and pathology research, and assist doctors to make more accurate diagnosis, as well as deep learning technology can accelerate this process. Convolutional Neural Networks (CNNs) and Transformer have become two mainstream architectures of deep learning in medical image segmentation. However, the Transformer architecture has limited ability to obtain local inductive bias, and the Transformer architecture is at a disadvantage in a small sample data set. Many theories and experiments show that the above problems can be effectively solved by fusing Convolution and Transformer features. In this manuscript, a new U-shaped segmentation model based on Convolution and swin-transformer framework is proposed, which is called CST-UNET. In the encoder part, it combines the advantages of both dilated convolution and Transformer, which can make the model fully obtain semantic inductive bias information and long-term information. At the same time, it has the advantages of fewer parameters and lower Flops. Even if it is trained on a small sample data set, the framework still has strong generalization ability. In addition, on BraTS2021 dataset, the Dice coefficients of ET, TC and WTare 85.46%, 89.38%, 92.35% respectively, and the result of HD95 are 7.95, 5.06 and 4.07 respectively.

热轧带钢是钢铁行业的主要原材料之一,其表面质量控制一直是生产过程智能检测的重点任务.针对带钢表面缺陷自动在线检测逐步取代人工检测的现状,概述带钢表面缺陷检测方法,着重阐述基于机器视觉的表面缺陷检测方法,比较分析传统机器视觉、深度学习方法在带钢表面缺陷检测的应用,探讨带钢表面缺陷检测中存在的关键技术问题,并对其未来发展趋势做展望.传统机器视觉的带钢缺陷检测方法检测速度较高,但精度较低;主流深度学习的缺陷检测方法检测精度高,但速度较慢.如何在保证检测实时性的前提下提高算法的准确性和鲁棒性,不仅是自动化和智能化检测的发展趋势,也是基于机器视觉部署在实际工业现场的关键所在.

Surface defect inspection is a key step to ensure the quality of the hot rolled steel surface. However, current advanced detection (DET) methods have high precision but low detection speed, which hinders the application of the detector in actual production. In this work, a real-time detection network (RDN) focusing on both speed and accuracy is proposed to solve the problem of steel surface defect detection. RDN takes ResNet-dcn, a modular encoding, and decoding network with light weights, as the basic convolutional architecture whose backbone is pretrained on ImageNet. To improve the detection accuracy, a skip layer connection module (SCM) and a pyramid feature fusion module (PFM) are involved into RDN. On the standard dataset NEU-DET, the proposed method can achieve the state-of-the-art recognition speed of 64 frames per second (FPS) and the mean average precision of 80.0% on a single GPU, which fully meets the requirements of the detection accuracy and speed in the actual production line.

Surface defect classification plays an important role in the assessment of production status and analyzing possible defect causes of hot rolled strip steel. It is extremely challenging owing to the rare occurrence and various appearances of defects. In this work, an improved deep learning model is proposed to solve the problem of poor classification accuracy when only a few labeled samples can be available. Different from most inductive small-sample learning methods, a transductive learning algorithm is designed where a new classifier is trained in the test phase and therefore can fit in with the needs of unknown samples. In addition, a simple feature fusion technique is implemented to extract more sample information. Based on a real-world steel surface defect dataset NEU, the proposed method can achieve a high classification accuracy of 97.13% with only one labeled sample. The experimental results show that the improved model is superior to other existing few-shot learning methods for surface defects classification of hot-rolled steel strip.

The change of core temperature of blast furnace reflects the working status of hearth. However, the temperature of core dead stock column can not be measured by sensors directly. Therefore, a prediction model of Core Dead Stock Column Temperature is proposed in this work based on primary component analysis (PCA) and ridge regression algorithms, where PCA and person correlation coefficient are used for feature extraction and ridge regression is employed to solve multi-collinearity problems. Based on an inhouse dataset collected within a successive three months, experimental results show that the R -squared of model on the training data set can achieve 88% and the average relative error on the test data set is only 0.33%, which shows the effectiveness of the proposed model.