Digital histopathology image segmentation can facilitate computer-assisted cancer diagnostics. Given the difficulty of obtaining manual annotations, weak supervision is more suitable for task than full is. However, most weakly supervised models are not ideal handling severe intra-class heterogeneity and inter-class homogeneity in images. Therefore, we propose a novel end-to-end learning framework named WESUP. With only sparse point it performs accurate exhibits good generalizability. The...

PageRank has been widely used to measure the authority or influence of a user in social networks. However, conventional only makes use edge-based relations, which represent first-order relations between two connected nodes. It ignores higher-order that may exist In this article, we propose novel framework, motif-based (MPR), incorporate into computation. Motifs are subgraphs consisting small number We motifs capture nodes network and introduce methods, one linear non-linear, combine conduct...

Tropical cyclone (TC) intensity estimation is an important task in meteorological research. Meanwhile, TC performance can be improved by developing advanced machine learning techniques using the newly emerged high-quality multispectral images (MSIs) acquired FY-4 satellite of China. To this end, article proposes a novel model, tensor-based convolutional neural network (TCNN). Not only being deep entirely formulated tensor algebra, but TCNN also establishes mathematical connections between...

Cell classification based on histopathology images is crucial for tumor recognition and cancer diagnosis. Using deep learning, accuracy hugely improved. Semi-supervised learning an advanced approach that uses both labeled unlabeled data. However, complex datasets comprise diverse patterns may drive models towards harmful features. Therefore, it useful to involve human guidance during training. Hence, we propose a mixed-supervised method incorporating semi-supervision "human-in-the-loop" cell...

Hyperspectral unmixing is an important issue in hyperspectral image processing. In this paper, we transform the problem into a constrained nonlinear least squares (CNLS) by introducing abundance sum-to-one constraint, nonnegative and bound constraints on nonlinearity parameters. The new CNLS-based algorithms assume that mixing mechanism of each observed pixel can be described two forms. One sum linear mixtures endmember spectra variations reflectance, other joint mixture resulting from...

Tropical cyclone (TC) intensity estimation is vital to disastrous weather forecasting. In this paper, the task approached as a classification problem, regarding levels class labels. Multispectral Imagery (MSI) captured by recently launched satellite, No. 4 meteorological satellite (FY-4) of China, used raw data for classification. To solve paper proposes machine learning framework with three major parts: useable band determination, band-wise and fusion. The compatible arbitrary classifiers...

Change detection (CD) for multitemporal hyperspectral images (HSI) can be approached as classification consisting of two steps, change feature extraction and identification. This paper is focused on binary the changed unchanged samples, which essential case detection. Meanwhile, it challenging to extract clean features from heavily corrupted spectral vectors (SCV) HSI. The corruptions characterized gross sample-specific errors, i.e., outliers, small entry-wise noise following Gaussian...

It is challenging to estimate wind speed of tropical cyclones directly using remote sensing image patterns. This paper approaches the task in two major steps: cyclone category estimation and regression. A novel framework based on Tensor Convolutional Neural Network (Tensor CNN) proposed solve problem. Not only does combine analysis for dimensionality reduction deep neural networks pattern recognition, CNN also provides a unitary concise mathematical representation form significant models....

Biomedical image analysis by machine learning enables accurate cell counting and localization, which are essential to clinic diagnosis research. However, high-density population of the cells poses great challenges. The variability intricacy in shapes colours further jeopardize performance existing models. Aiming overcome difficulties, we propose an end-to-end manifold-regularized regression network (MRRN) for localization. backbone is a U-Net tweaked trained density maps preserve spatial...

Nonlinear geometric manifold of hyperspectral data usually makes great trouble for accurate endmember extraction in literature. To address this issue, we propose a novel nonlinear algorithm by building hypergraph and fuzzy assessment strategy. The global change is first measured whose hyperedges correspond to different local pixel subgroups. In contrast edges simple graph, every hyperpath connected multiple instead individual pixels effectively facilitates the determination simplex spanned...

High-molecular weight nylon 66/modified clay (Mclay) nanocomposites with a low apparent viscosity were prepared by in-situ polymerization and post solid-state polycondensation. Thermogravimetric analysis X-ray diffraction patterns of the Mclay revealed that cetyltrimethyl ammonium bromide successfully inserted into interlayers clay. Scanning electron microscope images cross sections showed was well-dispersed in 66 matrix. The effects on mechanical, rheological, thermal properties...

Existing image-to-image (I2I) translation methods achieve state-of-the-art performance by incorporating the patch-wise contrastive learning into Generative Adversarial Networks. However, only focuses on local content similarity but neglects global structure constraint, which affects quality of generated images. In this paper, we propose a new unpaired I2I framework based dual regularization and spectral normalization, namely SN-DCR. To maintain consistency texture, design using different...

Remote sensing change detection (CD) using multitemporal hyperspectral images (HSIs) provides detailed information on spectral–spatial changes and is useful in a variety of applications such as environmental monitoring, urban planning, disaster detection. However, the high dimensionality low spatial resolution HSIs do not only lead to expensive computation but also bring about inter-class homogeneity inner-class heterogeneity. Meanwhile, labeled samples are difficult obtain reality field...

There are two great challenges for classification of hyperspectral images (HSIs): lack in prior knowledge and serious internal-class variability. To address the issues, we propose a novel semisupervised method based on affinity scoring (AS). It can harness fuzzy state contributions spectral spatial features to classification. The consists three major steps: over-segmentation, modification. First, superpixels generated maintain local class consistency, which balance Then unlabeled samples...

Prediction of tropical cyclone (TC) intensity using remote sensing imagery can provide early warning to coastal areas avoid economic losses and human casualties. Accurate TC prediction requires high-quality images. In recent years, deep learning based on images has been applied with great success, but most them ignore the key physical information that affects intensity. To this end, we propose an integrated framework for data enhancement analysis high spatial resolution predict maximum...

To address the issue of human object detection in transmission line inspection, an enhanced single-stage neural network is proposed, which based on improvement YOLOv7-tiny model. Firstly, a lighter GSConv module utilized to optimize original ELAN module, reducing parameters network. In order make less sensitive targets with unconventional pose, CSPNeXt and designed integrated extract deep features from targets. Moreover, WIoU (Wise Intersection over Union) loss function enhance ability model...

Maximum wind speed (MWS) is an important characteristic of tropical cyclone (TC). Estimation MWS with remote sensing images TCs via machine learning a relatively new and challenging task. Here we propose novel effective method, Regularized Tensor Network (RTN), to estimate using multispectral (MSIs). RTN transductive regression model, built on deep (TN) combined two regularizations: manifold categorization error. Experimental results showed that outperformed several classic methods as well...