Critical Feature Research Articles

Melanoma classification using generative adversarial network and proximal policy optimization.

In oncology, melanoma is a serious concern, often arising from DNA changes caused mainly by ultraviolet radiation. This cancer is known for its aggressive growth, highlighting the necessity of early detection. Our research introduces a novel deep learning framework for melanoma classification, trained and validated using the extensive SIIM-ISIC Melanoma Classification Challenge-ISIC-2020 dataset. The framework features three dilated convolution layers that extract critical feature vectors for classification. A key aspect of our model is incorporating the Off-policy Proximal Policy Optimization (Off-policy PPO) algorithm, which effectively handles data imbalance in the training set by rewarding the accurate classification of underrepresented samples. In this framework, the model is visualized as an agent making a series of decisions, where each sample represents a distinct state. Additionally, a Generative Adversarial Network (GAN) augments training data to improve generalizability, paired with a new regularization technique to stabilize GAN training and prevent mode collapse. The model achieved an F-measure of 91.836% and a geometric mean of 91.920%, surpassing existing models and demonstrating the model's practical utility in clinical environments. These results demonstrate its potential in enhancing early melanoma detection and informing more accurate treatment approaches, significantly advancing in combating this aggressive cancer.

HP-YOLOv8: High-Precision Small Object Detection Algorithm for Remote Sensing Images.

YOLOv8, as an efficient object detection method, can swiftly and precisely identify objects within images. However, traditional algorithms encounter difficulties when detecting small objects in remote sensing images, such as missing information, background noise, and interactions among multiple objects in complex scenes, which may affect performance. To tackle these challenges, we propose an enhanced algorithm optimized for detecting small objects in remote sensing images, named HP-YOLOv8. Firstly, we design the C2f-D-Mixer (C2f-DM) module as a replacement for the original C2f module. This module integrates both local and global information, significantly improving the ability to detect features of small objects. Secondly, we introduce a feature fusion technique based on attention mechanisms, named Bi-Level Routing Attention in Gated Feature Pyramid Network (BGFPN). This technique utilizes an efficient feature aggregation network and reparameterization technology to optimize information interaction between different scale feature maps, and through the Bi-Level Routing Attention (BRA) mechanism, it effectively captures critical feature information of small objects. Finally, we propose the Shape Mean Perpendicular Distance Intersection over Union (SMPDIoU) loss function. The method comprehensively considers the shape and size of detection boxes, enhances the model's focus on the attributes of detection boxes, and provides a more accurate bounding box regression loss calculation method. To demonstrate our approach's efficacy, we conducted comprehensive experiments across the RSOD, NWPU VHR-10, and VisDrone2019 datasets. The experimental results show that the HP-YOLOv8 achieves 95.11%, 93.05%, and 53.49% in the mAP@0.5 metric, and 72.03%, 65.37%, and 38.91% in the more stringent mAP@0.5:0.95 metric, respectively.

Rapid computation of Hopf bifurcation points of continuous and discrete systems through minimization

For an autonomous nonlinear system, the Hopf bifurcation point along the equilibrium path is a critical feature that indicates whether the values of the parameters change from exhibiting fixed-point behavior to having a periodic orbit. To solve these problems, we developed a method of transforming an eigenvalue problem based on the Jacobian matrix at equilibrium into a minimization problem, enabling the rapid identification of a solution. Specifically, this generalized eigenvalue problem is solved by identifying the vector variable after reducing the number of eigenequations by one in the nonhomogeneous linear system. This can be achieved by normalizing the value of a selected nonzero component of the eigenvector and then moving the column containing this component to the other side of the equation. An appropriate merit function was established in terms of the Euclidean norm of the eigenequation, and this merit function was minimized using the golden section search algorithm to determine the eigenparameters of the bifurcation point. The accuracy of the method for identifying the parameter values and the corresponding imaginary eigenvalues at the Hopf bifurcation points was evaluated for numerous examples for both the continuous and discrete systems. The method was both fast and accurate. Moreover, its stability in the presence of noise was investigated, and the method was robust.

Decentralized Retrofit Model Predictive Control of Inverter-Interfaced Small-Scale Microgrids

In recent years, small-scale microgrids have become popular in the power system industry because they provide an efficient electrical power generation platform to guarantee autonomy and independence from the power grid, which is a critical feature in cases of catastrophic events or remote areas. On the other hand, due to the short distances among multiple distribution generation systems in small-scale microgrids, the interconnection couplings among them increase significantly, which jeopardizes the stability of the entire system. Therefore, this work proposes a novel method to design decentralized robust controllers based on a retrofit model predictive control scheme to tackle the issue of instability due to the short distances among generation systems. In this approach, the retrofit model predictive controller receives the measured feedback signal from the interconnection current and generates a control command signal to limit the interconnection current to prevent instability. To design a retrofit controller, only the model of a robust closed-loop system, as well as an interconnection line, is required. The model predictive control signal is added in parallel to the control signal from the existing robust voltage source inverter controller. Simulation results demonstrate the superior performance of the proposed technique as compared with the virtual impedance and retrofit linear quadratic regulator techniques (benchmarks) with respect to peak-load demand, plug-and-play capability, nonlinear load, and inverter efficiency.

Depth estimation of pipe wall thinning using multifrequency reflection coefficients of T(0,1) mode-guided waves with supervised multilayer perceptron

This study entailed the development of a novel method for estimating the depth of wall thinning of pipes using multifrequency (30–65 kHz) reflection coefficients (MRCs) of the T(0,1) mode-guided waves and a multilayer perceptron (MLP). First, this study established why MRCs are a critical feature of the input layer of the MLP for the defect depth estimation of wall thinning. Further, a mathematical model that can quickly collect large amounts of training data was used to calculate the reflection waveforms. The depths of artificial and actual wall thinning were estimated using the MLP based on the MRCs and the mathematical model. Experiments were conducted using the T(0,1) mode-guided waves to obtain the MRCs for 21 artificial and 6 actual wall thinnings to estimate the defect depths. A maximum of 8347 training data points were prepared using the mathematical model. Because the optimization of the MLP strongly depended on the initial weights and biases, 100 random initial values were prepared to evaluate the average estimations and their standard deviations. The classification scheme of the MLP was used, with classification step widths of 0.5 and 0.25 mm. The correct answer rates for the 21 artificial defects were 93%, with a tolerance of ±0.5 mm for the 0.5 mm classification scheme; those for the 0.25 mm classification scheme were 89%. For the six actual defects, the correct answer rates were 100% with a tolerance of ±0.5 mm for both the 0.5 and 0.25 mm classification schemes. Sufficiently high correct answer rates were obtained in all the cases.

Exploring the Association Between Systemic Lupus Erythematosus and High-Density Lipoproteins: A Systematic Review and Meta-Analysis.

Systemic lupus erythematosus (SLE) is an autoimmune disease with inflammation as a critical feature. Recently, high-density lipoprotein cholesterol (HDLc) have been evidenced to have anti-inflammatory effects, suggesting a potential link between HDL and SLE that needs to be thoroughly studied. The aim was to explore the association between SLE and HDLc through a systematic review with meta-analysis. A systematic review with meta-analysis was conducted to assess mean differences in HDL levels between patients with SLE and healthy controls. Both qualitative and quantitative syntheses were performed, including an assessment of heterogeneity using I2, a publication bias evaluation, a methodologic quality assessment, and a forest plot under a random effects model. Subgroup analyses were conducted based on disease activity and the report of corticosteroid dosage. A total of 53 studies were included in the qualitative synthesis, and 35 studies were included in the quantitative synthesis, comprising 3,002 patients with SLE and 2,123 healthy controls. Mean HDL levels were found to be lower in patients with SLE as follows: in the meta-analysis including all articles -6.55 (95% confidence interval [CI] -8.77 to -4.33); in patients with mild disease activity -5.46 (95% CI -8.26 to -2.65); in patients with moderate or severe disease activity -9.42 (95% CI -15.49 to -3.34); in patients using corticosteroids -5.32 (95% CI -10.35 to -0.29); and in studies with excellent methodologic quality -8.71 (95% CI -12.38 to -5.03). HDL levels appear to be quantitatively altered in patients with SLE, suggesting a potential contribution to immune dysregulation, highlighting the importance of HDL in autoimmune diseases.

Ocular Surface Fluid: More than a Matrix.

Although the eye can be subjected to therapeutic manipulation, some of its structures are highly inaccessible. Thus, conventional therapeutic administration pathways, such as topical or systemic routes, usually show significant limitations in the form of low ocular penetration or the appearance of side effects linked to physiology, among others. The critical feature of many xenobiotics is the drug gradient from the concentrated tear reservoir to the relatively barren corneal and conjunctival epithelia, which forces a passive route of absorption. The same is true in the opposite direction, towards the ocular surface (OS). With the premise that tears can be regarded as equivalent to or a substitute for plasma, researchers may determine drug concentrations in the OS fluid. Within this framework, a survey of scholarly sources on the topic was conducted. It provided an overview of current knowledge, allowing the identification of relevant theories, methods, and gaps in the existing research that can be employed in subsequent research. OS fluid (tears particularly) has enormous potential as a source of biological material for external drug screening and as a biomarker of various systemic diseases. Given the numerous alternate matrices, knowledge of their properties is very important in selecting the most appropriate specimens in toxicological analyses.

Research on the Method of Foreign Object Detection for Railway Tracks Based on Deep Learning.

Addressing the limitations of current railway track foreign object detection techniques, which suffer from inadequate real-time performance and diminished accuracy in detecting small objects, this paper introduces an innovative vision-based perception methodology harnessing the power of deep learning. Central to this approach is the construction of a railway boundary model utilizing a sophisticated track detection method, along with an enhanced UNet semantic segmentation network to achieve autonomous segmentation of diverse track categories. By employing equal interval division and row-by-row traversal, critical track feature points are precisely extracted, and the track linear equation is derived through the least squares method, thus establishing an accurate railway boundary model. We optimized the YOLOv5s detection model in four aspects: incorporating the SE attention mechanism into the Neck network layer to enhance the model's feature extraction capabilities, adding a prediction layer to improve the detection performance for small objects, proposing a linear size scaling method to obtain suitable anchor boxes, and utilizing Inner-IoU to refine the boundary regression loss function, thereby increasing the positioning accuracy of the bounding boxes. We conducted a detection accuracy validation for railway track foreign object intrusion using a self-constructed image dataset. The results indicate that the proposed semantic segmentation model achieved an MIoU of 91.8%, representing a 3.9% improvement over the previous model, effectively segmenting railway tracks. Additionally, the optimized detection model could effectively detect foreign object intrusions on the tracks, reducing missed and false alarms and achieving a 7.4% increase in the mean average precision (IoU = 0.5) compared to the original YOLOv5s model. The model exhibits strong generalization capabilities in scenarios involving small objects. This proposed approach represents an effective exploration of deep learning techniques for railway track foreign object intrusion detection, suitable for use in complex environments to ensure the operational safety of rail lines.

Residual squeeze-and-excitation convolutional auto-encoder for fault detection and diagnosis in complex industrial processes

Recently, deep learning has attracted increasing attention in process monitoring. However, it is still a big challenge to extract effective features of process data for fault detection and diagnosis (FDD) by using deep learning-based methods. In this study, a novel deep learning-based model, residual squeeze-and-excitation convolutional auto-encoder (RSECAE) is proposed for FDD in complex industrial processes. A hybrid structure that integrates convolution calculation and auto-encoder is proposed to construct multiple convolution layers and deconvolution layers in the encoder and decoder respectively. Thus, RSECAE can learn effective features from the process signals in an unsupervised manner. At the same time, multiple residual squeeze-and-excitation networks (RSENets) are embedded in the deep network, which effectively preserves the critical feature representations of the convolution layer and transmits them to the decoder in the form of residuals. Finally, maximum mean discrepancy (MMD) term is utilized in RSECAE to reduce the distribution difference between the process signals and the learned features. This enables RSECAE to extract effective features with prior distribution information. RSECAE provides a new solution for process FDD for complicated industrial processes. Two industrial cases are adopted to validate the effectiveness of the RSECAE in process FDD. The experimental results demonstrate the effectiveness and superiority of RSECAE in process fault detection and fault diagnosis.

Language-vision matching for text-to-image synthesis with context-aware GAN

Text-to-image generation (T2I) aims to produce visually compelling images while maintaining a high degree of semantic consistency with textual descriptions. Despite the impressive progress made by existing methods, there are problems with limited details of synthesized images and insufficient correlation between the provided text description and the generated images. To address these issues, we propose a Context-Aware Generative Adversarial Network (CA-GAN), which generates images aligned with the input text representations. Specifically, the Context-Aware Block (CA-Block) learns a semantic-adaptive transformation based on text style, enabling the effective fusion of text descriptions and image features for high-quality image generation with better language-vision matching. Furthermore, we propose an Attention Convolution Module (ACM) to identify greater representative traits and avoid the inability to capitalize on non-local contextual information, which enables our model to generate images that exhibit numerous detailed attributes while maintaining high-quality and semantic consistency. Thereafter, we integrate self-attention with convolution to enhance feature maps to reinforce the semantic information in the discriminator, emphasizing critical feature channels while suppressing extraneous information, ultimately yielding more detailed and richer images. The experimental results demonstrate the superiority of our method over the SOTA approaches. In addition, further studies confirm the effectiveness of the generated visual details, which exhibit a high degree of alignment with the input text descriptions. Notably, our attention mechanism showcases cooperative effects contributing to overall performance improvement. The code is available at: https://github.com/hylneu/CAGAN.

Metallurgical Copper Recovery Prediction Using Conditional Quantile Regression Based on a Copula Model

This article proposes a novel methodology for estimating metallurgical copper recovery, a critical feature in mining project evaluations. The complexity of modeling this nonadditive variable using geostatistical methods due to low sampling density, strong heterotopic relationships with other measurements, and nonlinearity is highlighted. As an alternative, a copula-based conditional quantile regression method is proposed, which does not rely on linearity or additivity assumptions and can fit any statistical distribution. The proposed methodology was evaluated using geochemical log data and metallurgical testing from a simulated block model of a porphyry copper deposit. A highly heterotopic sample was prepared for copper recovery, sampled at 10% with respect to other variables. A copula-based nonparametric dependence model was constructed from the sample data using a kernel smoothing method, followed by the application of a conditional quantile regression for the estimation of copper recovery with chalcocite content as secondary variable, which turned out to be the most related. The accuracy of the method was evaluated using the remaining 90% of the data not included in the model. The new methodology was compared to cokriging placed under the same conditions, using performance metrics RMSE, MAE, MAPE, and R2. The results show that the proposed methodology reproduces the spatial variability of the secondary variable without the need for a variogram model and improves all evaluation metrics compared to the geostatistical method.

A Security Domain Feature Pattern Recognition Method Based on Capsule Neural Network

Abstract In this paper, a security domain feature pattern recognition method based on capsule neural network is proposed. Firstly, Halton sampling is performed on the active output of all generators except for the balancing machine to obtain a uniform sample in the coverage space. Then, the support vector machine based recursive feature elimination (SVM-RFE) method is used to select the critical pre fault feature quantities of the system. Finally, based on the selected active features, one-dimensional data is converted into two-dimensional data and input into the capsule neural network. Dynamic security domain boundaries are constructed in both two-dimensional and three-dimensional spaces, which can divide the feature space into stable and unstable subspaces.

Modeling lung endothelial dysfunction in sepsis-associated ARDS using a microphysiological system.

Endothelial dysfunction is a critical feature of acute respiratory distress syndrome (ARDS) associated with higher disease severity and worse outcomes. Preclinical invivo models of sepsis and ARDS have failed to yield useful therapies in humans, perhaps due to interspecies differences in inflammatory responses and heterogeneity of human host responses. Use of microphysiological systems (MPS) to investigate lung endothelial function may shed light on underlying mechanisms and targeted treatments for ARDS. We assessed the response to plasma from critically ill sepsis patients in our lung endothelial MPS through measurement of endothelial permeability, expression of adhesion molecules, and inflammatory cytokine secretion. Sepsis plasma induced areas of endothelial cell (EC) contraction, loss of cellular coverage, and luminal defects. EC barrier function was significantly worse following incubation with sepsis plasma compared to healthy plasma. EC ICAM-1 expression, IL-6 and soluble ICAM-1 secretion increased significantly more after incubation with sepsis plasma compared with healthy plasma. Plasma from sepsis patients who developed ARDS further increased IL-6 and sICAM-1 compared to plasma from sepsis patients without ARDS and healthy plasma. Our results demonstrate the proof of concept that lung endothelial MPS can enable interrogation of specific mechanisms of endothelial dysfunction that promote ARDS in sepsis patients.

Studying the Thermodynamic Phase Stability of Organic-Inorganic Hybrid Perovskites Using Machine Learning.

As an important photovoltaic material, organic-inorganic hybrid perovskites have attracted much attention in the field of solar cells, but their instability is one of the main challenges limiting their commercial application. However, the search for stable perovskites among the thousands of perovskite materials still faces great challenges. In this work, the energy above the convex hull values of organic-inorganic hybrid perovskites was predicted based on four different machine learning algorithms, namely random forest regression (RFR), support vector machine regression (SVR), XGBoost regression, and LightGBM regression, to study the thermodynamic phase stability of organic-inorganic hybrid perovskites. The results show that the LightGBM algorithm has a low prediction error and can effectively capture the key features related to the thermodynamic phase stability of organic-inorganic hybrid perovskites. Meanwhile, the Shapley Additive Explanation (SHAP) method was used to analyze the prediction results based on the LightGBM algorithm. The third ionization energy of the B element is the most critical feature related to the thermodynamic phase stability, and the second key feature is the electron affinity of ions at the X site, which are significantly negatively correlated with the predicted values of energy above the convex hull (Ehull). In the screening of organic-inorganic perovskites with high stability, the third ionization energy of the B element and the electron affinity of ions at the X site is a worthy priority. The results of this study can help us to understand the correlation between the thermodynamic phase stability of organic-inorganic hybrid perovskites and the key features, which can assist with the rapid discovery of highly stable perovskite materials.

Bluish veil detection and lesion classification using custom deep learnable layers with explainable artificial intelligence (XAI)

Melanoma, one of the deadliest types of skin cancer, accounts for thousands of fatalities globally. The bluish, blue-whitish, or blue-white veil (BWV) is a critical feature for diagnosing melanoma, yet research into detecting BWV in dermatological images is limited. This study utilizes a non-annotated skin lesion dataset, which is converted into an annotated dataset using a proposed imaging algorithm (color threshold techniques) on lesion patches based on color palettes. A Deep Convolutional Neural Network (DCNN) is designed and trained separately on three individual and combined dermoscopic datasets, using custom layers instead of standard activation function layers. The model is developed to categorize skin lesions based on the presence of BWV. The proposed DCNN demonstrates superior performance compared to the conventional BWV detection models across different datasets. The model achieves a testing accuracy of 85.71 % on the augmented PH2 dataset, 95.00 % on the augmented ISIC archive dataset, 95.05 % on the combined augmented (PH2+ISIC archive) dataset, and 90.00 % on the Derm7pt dataset. An explainable artificial intelligence (XAI) algorithm is subsequently applied to interpret the DCNN's decision-making process about the BWV detection. The proposed approach, coupled with XAI, significantly improves the detection of BWV in skin lesions, outperforming existing models and providing a robust tool for early melanoma diagnosis.

Comprehensive analytics for virus-cell and cell-cell multinucleation system

Cell-fusion mediated generation of multinucleated syncytia represent critical feature during viral infection and in development. Efficiency of syncytia formation is usually illustrated as fusion efficiency under given condition by quantifying total number of nuclei in syncytia normalized to total number of nuclei (both within syncytia and unfused cell nuclei) in unit field of view. However heterogeneity in multinucleated syncytia sizes poses challenge in quantification of cell-fusion multinucleation under diverse conditions. Taking in-vitro SARS-CoV-2 spike-protein variants mediated virus-cell fusion model and placenta trophoblast syncytialization as cell-cell fusion model; herein we emphasize wide application of simple unbiased detailed measure of virus-cell and cell-cell multinucleation using experiential cumulative distribution function (CDF) and fusion number events (FNE) approaches illustrating comprehensive metrics for syncytia interpretation.

Evaluating the Improvement of Blend Potency Measurements in the Feed Frame of a Rotary Tablet Press Using Combined NIR and Raman Spectroscopy.

This study investigated the added value of combining both near-infrared (NIR) and Raman spectroscopy into a single NIRaman Combi Fiber Probe for in-line blend potency determination in the feed frame of a rotary tablet press. A five-component platform formulation was used, containing acetylsalicylic acid as the Active Pharmaceutical Ingredient (API). Calibration models for the determination of 1 and 5%w/w label claim tablets were developed using NIR and Raman spectra of powder blends ranging from 0.75 to 1.25%w/w and 3.75 to 6.25%w/w API, respectively. Step-change experiments with deliberate 10% deviation steps from the label claims were performed, from which the collected spectra were used for model validation. For model development and validation, low-level data fusion was explored through concatenation of preprocessed NIR and Raman spectra. Mid-level data fusion was also evaluated, based on extracted features of the preprocessed data. Herewith, score vectors were extracted by transforming preprocessed spectra through Principal Component Analysis, followed by critical feature selection through Elastic Net Regression. Partial Least Squares regression was applied to regress singular, low-level or mid-level fused data versus blend potency. It could be concluded that irrespective of the data fusion technique, an increase in Step-Change Sensitivity (SCS) and decrease in Root Mean Squared Error (RMSE) was observed when predicting the 5%w/w step-change experiment. For the prediction of the 1%w/w step-change experiment, no added benefit with regard to SCS and RMSE was observed due to the addition of the noisy NIR spectra.

Peripheral PD-1+NK cells could predict the 28-day mortality in sepsis patients.

Unbalanced inflammatory response is a critical feature of sepsis, a life-threatening condition with significant global health burdens. Immune dysfunction, particularly that involving different immune cells in peripheral blood, plays a crucial pathophysiological role and shows early warning signs in sepsis. The objective is to explore the relationship between sepsis and immune subpopulations in peripheral blood, and to identify patients with a higher risk of 28-day mortality based on immunological subtypes with machine-learning (ML) model. Patients were enrolled according to the sepsis-3 criteria in this retrospective observational study, along with age- and sex-matched healthy controls (HCs). Data on clinical characteristics, laboratory tests, and lymphocyte immunophenotyping were collected. XGBoost and k-means clustering as ML approaches, were employed to analyze the immune profiles and stratify septic patients based on their immunological subtypes. Cox regression survival analysis was used to identify potential biomarkers and to assess their association with 28-day mortality. The accuracy of biomarkers for mortality was determined by the area under the receiver operating characteristic (ROC) curve (AUC) analysis. The study enrolled 100 septic patients and 89 HCs, revealing distinct lymphocyte profiles between the two groups. The XGBoost model discriminated sepsis from HCs with an area under the receiver operating characteristic curve of 1.0 and 0.99 in the training and testing set, respectively. Within the model, the top three highest important contributions were the percentage of CD38+CD8+T cells, PD-1+NK cells, HLA-DR+CD8+T cells. Two clusters of peripheral immunophenotyping of septic patients by k-means clustering were conducted. Cluster 1 featured higher proportions of PD1+ NK cells, while cluster 2 featured higher proportions of naïve CD4+T cells. Furthermore, the level of PD-1+NK cells was significantly higher in the non-survivors than the survivors (15.1% vs 8.6%, P<0.01). Moreover, the levels of PD1+ NK cells combined with SOFA score showed good performance in predicting the 28-day mortality in sepsis (AUC=0.91,95%CI 0.82-0.99), which is superior to PD1+ NK cells only(AUC=0.69, sensitivity 0.74, specificity 0.64, cut-off value of 11.25%). In the multivariate Cox regression, high expression of PD1+ NK cells proportion was related to 28-day mortality (aHR=1.34, 95%CI 1.19 to 1.50; P<0.001). The study provides novel insights into the association between PD1+NK cell profiles and prognosis of sepsis. Peripheral immunophenotyping could potentially stratify the septic patients and identify those with a high risk of 28-day mortality.

Weak antilocalization and sign inversion of magnetoresistance on CaRuO3 epitaxial films

CaRuO3 is a prototypical perovskite oxide among the Ruddlesden–Popper series with multi-exotic physical properties on the verge of quantum criticality. Unlike the case of the well-known ferromagnetic SrRuO3, the literature on the physical properties of the CaRuO3 system is scarce with no consensus about its magnetic and transport ground states. This work provides a detailed study of the magnetotransport properties of 300 nm CaRuO3 epitaxial films grown on single crystalline SrTiO3 and LaAlO3 substrates. The magnetic and electronic ground states in tensile strained CaRuO3/SrTiO3 films and compressively strained CaRuO3/LaAlO3 films are found to be distinctly different. In particular, in zero magnetic field, the high-temperature resistivity of both films show a T1/2 non-Fermi liquid behavior. Cooling below 30 K, CaRuO3/SrTiO3 undergoes metal–insulator-transition ascribed by 3D weak localization, with signs of weak ferromagnetic-like domains. By contrast, the transport data of the nonmagnetic CaRuO3/LaAlO3 film below 30 K, show T3/2 non-Fermi liquid type behavior. Magnetoresistance in CaRuO3/SrTiO3 shows combination of weak antilocalization and weak localization behaviors deep in the insulating side, emphasizing the existence of spin–orbit coupling and the strong influence of ferromagnetic impurities. Moreover, the nonlinearity signals in CaRuO3/SrTiO3 Hall measurements could be described by anomalous Hall effects. Interestingly, below 30 K, magnetoresistance and Hall in CaRuO3/LaAlO3 change sign, indicating an inversion from hole-like to electron-like behavior. We discuss the sensitivity of the 300 nm CaRuO3/SrTiO3 and CaRuO3/LaAlO3 films to the 30 K temperature, an additional quantum critical feature to the ruthenate oxides.

A new lightweight network for efficient UAV object detection

Optimizing the structure of deep neural networks is essential in many applications. Especially in the object detection tasks of Unmanned Aerial Vehicles. Due to the constraints of the onboard platform, a more efficient network is required to meet practical demands. Nevertheless, existing lightweight detection networks exhibit excessive redundant computations and may yield in a certain level of accuracy loss. To address these issues, this paper proposes a new lightweight network structure named Cross-Stage Partially Deformable Network (CSPDNet). The initial proposal consists of a Deformable Separable Convolution Block (DSCBlock), separating feature channels, greatly reducing the computational load of convolution, and applying adaptive sampling to the separated feature map. Subsequently, to establish information interaction between feature layers, a channel weighting module is proposed. This module calculates weights for the separated feature map, facilitating information exchange across channels and resolutions. Moreover, it compensates for the effect of point-wise (1 ×\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\ imes$$\\end{document} 1) convolutions, filtering out more important feature information. Furthermore, a new CSPDBlock is designed, primarily composed of DSCBlock, establishing multidimensional feature correlations for each separated feature layer. This approach improves the ability to capture critical feature information and reconstruct gradient paths, thereby preserving detection accuracy. The proposed technology achieves a balance between model parameter size and detection accuracy. Furthermore, experimental results on object detection datasets demonstrate that our designed network, using fewer parameters, achieves competitive detection performance results compared to existing lightweight networks YOLOv5n, YOLOv6n, YOLOv8n, NanoDet and PP-PicoDet. The optimization effect of the designed CSPDBlock, using the VisDrone dataset, is validated when incorporated into advanced detection algorithms YOLOv5m, PPYOLOEm, YOLOv7, RTMDetm and YOLOv8m. In more detail, by incorporating the designed modules it was achieved that the parameters were reduced by 10–20% while almost maintaining detection accuracy.