Abundant Features Research Articles

Generation of magnonic frequency combs via a two-tone microwave drive

Magnonic frequency combs, a direct analog of optical frequency combs in the field of spin waves, have recently received considerable attentions due to their potential applications in high-precision magnonic frequency metrology. However, due to the weak nonlinear interaction of magnons, it is difficult to generate a wide-bandwidth magnonic frequency comb under a low power drive. Here, we present an efficient mechanism for the generation of a wide-bandwidth magnonic frequency comb via a two-tone microwave driving in a magnomechanical system. Numerical simulations show that the magnetostrictive effect can be greatly enhanced by the beat frequency signal from the two-tone microwave driving field, and a robust magnonic frequency comb can be observed at low power. Furthermore, abundant nonperturbative features appear in the magnonic spectrum, implying that the magnons as bosons may also be similar to photons in atom-molecular systems. Our scheme thus provides a pathway for the generation of a flat magnonic frequency comb under a low-power driving condition that may be beneficial for precision metrology based on magnonic platforms as well as the understanding of nonlinear magnomechanic dynamics.

Enhanced Pseudo-Capacitance Process in Nanoarchitectural Layered Double Hydroxide Nanoarrays Hollow Nanocages for Improved Capacitive Deionization Performance

Layered double hydroxides (LDHs) are perceived as a hopeful capacitive deionization (CDI) faradic electrode for Cl- insertion due to its tunable composition, excellent anion exchange capacity, and fast redox activity. Nevertheless, the self-stacking and inferior electrical conductivity of the two-dimensional structure of LDH lead to unsatisfactory CDI performance. Herein, the three-dimensional (3D) hollow nanocage structure of CoNi-layered double hydroxide/carbon composites is well designed as a CDI anode by cation etching of the pre-carbonized ZIF-67 template. C/CoNi-LDH has a unique 3D hollow nanocage structure and abundant pore features, which can effectively suppress the self-stacking of LDH sheets and facilitate the transport of ions. Moreover, the introduced amorphous carbon layer can act as a conductive network. When employed as the CDI anode, C/CoNi-LDH exhibited a high Cl- removal capacity of 60.88 mg g-1 and a fast Cl- removal rate of 18.09 mg g-1 min-1 at 1.4 V in 1000 mg L-1 NaCl solution. The mechanism of the Cl- intercalation pseudo-capacitance reaction of C/CoNi-LDH is revealed by electrochemical kinetic analysis and ex situ characterization. This study provides vital guidance for the design of high-performance electrodes for CDI.

Utah geosite—the Salina Canyon unconformity, a classic example of missing time

Salina Canyon, Utah, reveals a spectacular angular unconformity along an east-west transect through the southern part of the Wasatch Plateau. This region of Utah is well known as the eastern extent of Sevier orogenesis, but it also includes subsequent extensional overprinting. Earliest descriptions of this unconformity were published by Dutton (1880) and Spieker (1946, 1949), and work continues today. Field relationships expose many classic stratigraphic and sedimentologic features of erosional surfaces. Due to the geometry of the progressive unconformity onto the topographic high of the Sanpete-Sevier Valley antiform, the angular discordance of strata results in a gap in time of greater than 107 million years in the west, decreasing toward the east to about 39 million years and finally to less than 17 million years. Paleosols and small-scale channels/scours with infilled basal conglomerates are also prominent along the unconformity, as are several mine adits. Because of its abundant geologic features, the Salina Canyon unconformity is a superb teaching and learning space for geoscientists and outdoor naturalists.

Automatic cloud detection method for optical remote sensing images combining channel attention mechanism and multi-feature fusion

Cloud detection is a crucial step in remote sensing image pre-processing, but there are still issues such as cloud miss detection and false detection. Therefore, this work suggests a multi-feature fusion and channel attention mechanism combined approach for automatically recognizing clouds. Based on DeepLabV3+, to make the feature extraction faster and more accurate, a lightweight network structure MobileNetV2 is used as the backbone network. To effectively distinguish snow and clouds, channel attention is added to the low-level semantic information, which makes the network more able to focus on essential features. Meantime, mid-level semantic features are added to the decoder module to make use of more abundant features, thus reducing the false detection of mist. Experimental findings on the GF1 WHU dataset demonstrate that the method’s detection accuracy is improved by 2.1% when compared to DeepLabV3+, and it performs well in the cloud detection job of optical remote sensing images.

MGML: Multigranularity Multilevel Feature Ensemble Network for Remote Sensing Scene Classification.

Remote sensing (RS) scene classification is a challenging task to predict scene categories of RS images. RS images have two main issues: large intraclass variance caused by large resolution variance and confusing information from large geographic covering area. To ease the negative influence from the above two issues. We propose a multigranularity multilevel feature ensemble network (MGML-FENet) to efficiently tackle the RS scene classification task in this article. Specifically, we propose multigranularity multilevel feature fusion branch (MGML-FFB) to extract multigranularity features in different levels of network by channel-separate feature generator (CS-FG). To avoid the interference from confusing information, we propose a multigranularity multilevel feature ensemble module (MGML-FEM), which can provide diverse predictions by full-channel feature generator (FC-FG). Compared to previous methods, our proposed networks have the ability to use structure information and abundant fine-grained features. Furthermore, through the ensemble learning method, our proposed MGML-FENets can obtain more convincing final predictions. Extensive classification experiments on multiple RS datasets (AID, NWPU-RESISC45, UC-Merced, and VGoogle) demonstrate that our proposed networks achieve better performance than previous state-of-the-art (SOTA) networks. The visualization analysis also shows the good interpretability of MGML-FENet.

Position-Wise Gated Res2Net-Based Convolutional Network with Selective Fusing for Sentiment Analysis.

Sentiment analysis (SA) is an important task in natural language processing in which convolutional neural networks (CNNs) have been successfully applied. However, most existing CNNs can only extract predefined, fixed-scale sentiment features and cannot synthesize flexible, multi-scale sentiment features. Moreover, these models' convolutional and pooling layers gradually lose local detailed information. In this study, a new CNN model based on residual network technology and attention mechanisms is proposed. This model exploits more abundant multi-scale sentiment features and addresses the loss of locally detailed information to enhance the accuracy of sentiment classification. It is primarily composed of a position-wise gated Res2Net (PG-Res2Net) module and a selective fusing module. The PG-Res2Net module can adaptively learn multi-scale sentiment features over a large range using multi-way convolution, residual-like connections, and position-wise gates. The selective fusing module is developed to fully reuse and selectively fuse these features for prediction. The proposed model was evaluated using five baseline datasets. The experimental results demonstrate that the proposed model surpassed the other models in performance. In the best case, the model outperforms the other models by up to 1.2%. Ablation studies and visualizations further revealed the model's ability to extract and fuse multi-scale sentiment features.

Identification of Protein Complexes by Integrating Protein Abundance and Interaction Features Using a Deep Learning Strategy.

Many essential cellular functions are carried out by multi-protein complexes that can be characterized by their protein-protein interactions. The interactions between protein subunits are critically dependent on the strengths of their interactions and their cellular abundances, both of which span orders of magnitude. Despite many efforts devoted to the global discovery of protein complexes by integrating large-scale protein abundance and interaction features, there is still room for improvement. Here, we integrated >7000 quantitative proteomic samples with three published affinity purification/co-fractionation mass spectrometry datasets into a deep learning framework to predict protein-protein interactions (PPIs), followed by the identification of protein complexes using a two-stage clustering strategy. Our deep-learning-technique-based classifier significantly outperformed recently published machine learning prediction models and in the process captured 5010 complexes containing over 9000 unique proteins. The vast majority of proteins in our predicted complexes exhibited low or no tissue specificity, which is an indication that the observed complexes tend to be ubiquitously expressed throughout all cell types and tissues. Interestingly, our combined approach increased the model sensitivity for low abundant proteins, which amongst other things allowed us to detect the interaction of MCM10, which connects to the replicative helicase complex via the MCM6 protein. The integration of protein abundances and their interaction features using a deep learning approach provided a comprehensive map of protein-protein interactions and a unique perspective on possible novel protein complexes.

Application of long fibrous coconut silk-based porous carbon in flexible supercapacitor

All-solid-state flexible supercapacitors are considered to be one of the ideal candidates for energy storage in the next generation of wearable electronic devices due to their high capacitance performance and excellent mechanical flexibility. Biomass-based carbon materials are regarded as the ideal precursors for carbon-based electrode materials due to their wide sources, low cost, natural and abundant biological features. In this work, long-fiber coconut silk with vascular bundle structure was selected as the precursor, the controllable adjustment of the pore structure and conductive characteristics of the electrode material was achieved by changing the KOH activation temperature. The specific surface area of the activated carbon electrode material etched with KOH at 900 °C can reach 2794 m2·g−1. The high specific surface area and reasonable pore size distribution provide abundant active sites for the adsorption of electrolyte ions, which leads to the excellent electrode specific capacitance (2 mV·s−1，634 F·g−1，10.73 Wh·kg−1) of the symmetric flexible supercapacitor combined with PVA (polyvinyl alcohol)/H2SO4 gel electrolyte. At the same time, the long-fiber carbon skeleton has intrinsically high Young's modulus, which enables the flexible all-solid-state supercapacitor to maintain better electrochemical stability and mechanical durability under mechanical deformation, and it's retention rate can reach 101.5 % after 10,000 bending experiments at an inner angle of 140°. It proves the great advantages and potential of coconut silk-based biomass carbon materials in the field of preparing flexible supercapacitors, and promotes the development of energy storage units for next-generation wearable electronic devices.

BejaGNN: behavior-based Java malware detection via graph neural network.

As a popular platform-independent language, Java is widely used in enterprise applications. In the past few years, language vulnerabilities exploited by Java malware have become increasingly prevalent, which cause threats for multi-platform. Security researchers continuously propose various approaches for fighting against Java malware programs. The low code path coverage and poor execution efficiency of dynamic analysis limit the large-scale application of dynamic Java malware detection methods. Therefore, researchers turn to extracting abundant static features to implement efficient malware detection. In this paper, we explore the direction of capturing malware semantic information by using graph learning algorithms and present BejaGNN (Behavior-based Java malware detection via Graph Neural Network), a novel behavior-based Java malware detection method using static analysis, word embedding technique, and graph neural network. Specifically, BejaGNN leverages static analysis techniques to extract ICFGs (Inter-procedural Control Flow Graph) from Java program files and then prunes these ICFGs to remove noisy instructions. Then, word embedding techniques are adopted to learn semantic representations for Java bytecode instructions. Finally, BejaGNN builds a graph neural network classifier to determine the maliciousness of Java programs. Experimental results on a public Java bytecode benchmark demonstrate that BejaGNN achieves high F1 98.8% and is superior to existing Java malware detection approaches, which verifies the promise of graph neural network in Java malware detection.

Exploring the microbiome of oral epithelial dysplasia as a predictor of malignant progression

A growing body of research associates the oral microbiome and oral cancer. Well-characterized clinical samples with outcome data are required to establish relevant associations between the microbiota and disease. The objective of this study was to characterize the community variations and the functional implications of the microbiome in low-grade oral epithelial dysplasia (OED) using 16S rRNA gene sequencing from annotated archival swabs in progressing (P) and non-progressing (NP) OED. We characterised the microbial community in 90 OED samples — 30 swabs from low-grade OED that progressed to cancer (cases) and 60 swabs from low-grade OED that did not progress after a minimum of 5 years of follow up (matched control subjects). There were small but significant differences between P and NP samples in terms of alpha diversity as well as beta diversity in conjunction with other clinical factors such as age and smoking status for both taxa and functional predictions. Across all samples, the most abundant genus was Streptococcus, followed by Haemophilus, Rothia, and Neisseria. Taxa and predicted functions were identified that were significantly differentially abundant with progression status (all Ps and NPs), when samples were grouped broadly by the number of years between sampling and progression or in specific time to progression for Ps only. However, these differentially abundant features were typically present only at low abundances. For example, Campylobacter was present in slightly higher abundance in Ps (1.72%) than NPs (1.41%) and this difference was significant when Ps were grouped by time to progression. Furthermore, several of the significantly differentially abundant functions were linked to the Campylobacteraceae family in Ps and may justify further investigation. Larger cohort studies to further explore the microbiome as a potential biomarker of risk in OED are warranted.

Abstract 6606: Biomarker discovery in non-small-cell lung cancer enabled by deep multi-omics profiling of proteins, metabolites, transcripts, and genes in blood

Abstract Lung cancer is the leading cause of cancer-related deaths in the United States, with estimates of 236,740 new cases and 118,830 deaths in 2022 secondary to the disease. Blood-based liquid biopsies hold promise to reduce morbidity and mortality from lung cancer by enabling early detection to downstage disease at diagnosis, theragnostic identification of patients most likely to be helped or harmed by therapy, monitoring of therapeutic efficacy, and detection of residual disease. PrognomiQ’s multi-omics platform comprehensively profiles proteins, metabolites, lipids, mRNA, and cfDNA in blood samples which can be used for the development of liquid biopsy tests with high sensitivity and specificity for lung cancer. We conducted a case-control study comprising 1031 subjects: 361 subjects with untreated non-small-cell lung cancer (NSCLC) and 670 matched controls which included 340 subjects with salient pulmonary and gastrointestinal co-morbidities. Blood samples from each subject were processed to provide 7 different `omics readouts. LCMS was used to detect and quantify proteins, metabolites, and lipids. In addition, cfDNA and mRNA were assayed using next-generation sequencing. cfDNA reads were analyzed to estimate fragment-lengths, copy-number variation, and CpG site methylation. All molecular data were normalized using standard methods specific to each assay. Univariate analyses of cases vs controls were performed to identify differentially abundant features on all available samples per assay. We detected 9,868 proteins, 605 lipids, 329 metabolites, and 109,070 mRNA transcripts. Of these, 3,098 proteins, 210 lipids, 57 metabolites, and 30,236 mRNA transcripts were significantly different (FWER < 0.05) in cases versus controls. Gene set enrichment analysis on statistically significant transcripts and proteins identified multiple gene-ontology terms associated with cancer including the Wnt signaling process and IgA immunoglobulin complex, respectively. From cfDNA data, we identified 234 non-contiguous genomic regions associated with the fragment-length disorder, 4,790 with copy-number variation, and 74 differentially methylated genomic regions spanning 184 CpG sites (FWER < 0.05). With the premise that deviations from copy number neutrality are more likely to indicate a tumor contribution, we then focused our examination on those differentially expressed proteins that overlap with differentially expressed mRNA transcripts as well as CNV genomic regions. We identified 52 protein coding genes including E-cadherin (associated with EMT) and related binding proteins such as RAB11B, CAPZB, EPS15, FLNB, MYH9, STK24 and YWHAE. Ongoing machine-learning-based classifier training to distinguish between cancer and non-cancer can serve as the basis for the development of high-sensitivity liquid-biopsy tests for lung cancer. Citation Format: Jinlyung Choi, Ajinkya Kokate, Ehdieh Khaledian, Manway Liu, Preethi Prasad, John Blume, Jessica Chan, Rea Cuaresma, Kevin Dai, Manoj Khadka, Thidar Khin, Yuya Kodama, Joon-Yong Lee, Hoda Malekpour, Megan Mora, Nithya Mudaliar, Sara Nouri Golmaei, Madhuvanthi Ramaiah, Saividya Ramaswamy, Peter Spiro, Dijana Vitko, Kavya Swaminathan, James Yee, Brian Young, Chinmay Belthangady, Bruce Wilcox, Brian Koh, Philip Ma. Biomarker discovery in non-small-cell lung cancer enabled by deep multi-omics profiling of proteins, metabolites, transcripts, and genes in blood. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6606.

DGSSC: A Deep Generative Spectral-Spatial Classifier for Imbalanced Hyperspectral Imagery

In recent years, hyperspectral image classification (HSIC) has achieved impressive progress with emerging studies on deep learning models. However, the classification performance downgrades due to the limited number of annotated samples, especially for minority classes. Notably, the imbalanced data dilemma is familiar in remote sensing hyperspectral image because the ground objects are commonly distributed without evenness. Therefore, this paper proposes a novel deep generative spectral-spatial classifier (DGSSC) for addressing the issues of imbalanced HSIC. Specifically, the DGSSC comprises three components, a two-stage encoder, a decoder, and a classifier, which are trained in an end-to-end manner. In particular, to exploit the abundant spectral-spatial features with relatively low computational complexity, the first stage of the encoder comprises successive three-dimensional (3D) and two-dimensional (2D) convolutions, exploring the spectral-spatial and deep spatial information. In addition, the second stage involves the deep latent variable model to achieve minority-class data augmentation. Furthermore, a patch distance-based reconstruction loss function is designed to facilitate the outputs of the decoder being more similar to the input 3D patch samples. The proposed DGSSC can outperform the state-of-the-art methods on three benchmark datasets, especially with its more robust prediction results. For instance, the DGSSC achieves a remarkable 97.85% mean overall accuracy with 0.24% standard deviation over ten independent runs with randomly selected imbalanced 1% training samples on the University of Pavia dataset.

Fusion of Spatial, Temporal, and Spectral EEG Signatures Improves Multilevel Cognitive Load Prediction

Cognitive load prediction is one of the most important issues in the nascent field of neuroergonomics, and it has significant value in real-world applications. Most of the previous studies of cognitive load prediction only utilized electroencephalography (EEG)-based spectral signatures or interchannel connectivity, ignoring abundant temporal microstate features, which may represent the transient topologies of EEG signals. Furthermore, previous studies have mostly focused on the binary-level classification of cognitive load for single-type cognitive tasks. To date, there are few studies on the multilevel prediction of cognitive load during mixed cognitive tasks. Here, we first designed a new paradigm termed the “finding fault game,” mixing multiple tasks of memory, counting, and visual search, and then developed a multidimensional analysis framework to improve cognitive load prediction using a fusion of spatial, temporal, and spectral EEG features. Specifically, EEG-based functional connectivity, microstates and power spectral densities (PSD) were calculated for three cognitive load levels. Twelve adult subjects participated in the study. The experimental results show that increased cognitive load was associated with elevated theta and degraded alpha power and significant changes in interchannel connectivity and microstates, and that fusing the three types of EEG features improved the performance of three-level cognitive load prediction, achieving the accuracies of greater than 80% in the cross-validation, real-time, and over-time prediction. The findings suggest that all three types of EEG features can serve as signatures of cognitive load and that their fusion can improve multilevel prediction.

Designing Lignin-Based Biomaterials as Carriers of Bioactive Molecules.

There is a need to develop circular and sustainable economies by utilizing sustainable, green, and renewable resources in high-tech industrial fields especially in the pharmaceutical industry. In the last decade, many derivatives of food and agricultural waste have gained considerable attention due to their abundance, renewability, biocompatibility, environmental amiability, and remarkable biological features. Particularly, lignin, which has been used as a low-grade burning fuel in the past, recently attracted a lot of attention for biomedical applications because of its antioxidant, anti-UV, and antimicrobial properties. Moreover, lignin has abundant phenolic, aliphatic hydroxyl groups, and other chemically reactive sites, making it a desirable biomaterial for drug delivery applications. In this review, we provide an overview of designing different forms of lignin-based biomaterials, including hydrogels, cryogels, electrospun scaffolds, and three-dimensional (3D) printed structures and how they have been used for bioactive compound delivery. We highlight various design criteria and parameters that influence the properties of each type of lignin-based biomaterial and corelate them to various drug delivery applications. In addition, we provide a critical analysis, including the advantages and challenges encountered by each biomaterial fabrication strategy. Finally, we highlight the prospects and future directions associated with the application of lignin-based biomaterials in the pharmaceutical field. We expect that this review will cover the most recent and important developments in this field and serve as a steppingstone for the next generation of pharmaceutical research.

Dynamics of oral microbiome acquisition in healthy infants: A pilot study.

The human oral microbiota is one of the most complex bacterial communities in the human body. However, how newborns initially acquire these bacteria remains largely unknown. In this study, we examined the dynamics of oral microbial communities in healthy infants and investigated the influence of the maternal oral microbiota on the acquisition of the infant's oral microbiota. We hypothesized that the infant oral microbial diversity increases with age. One hundred and sixteen whole-salivary samples were collected from 32 healthy infants and their biological mothers during postpartum and 9- and 15-month well-infant visits. Bacterial genomic DNA was extracted and sequenced by Human Oral Microbe Identification using Next Generation Sequencing (HOMINGS) methods. The Shannon index was used to measure the microbial diversity of the infant-mother dyads (alpha diversity). The microbial diversity between the mother-infant dyads (beta-diversity) was calculated using the weighted non-phylogenetic Bray-Curtis distance in QIIME 1.9.1. Core microbiome analysis was performed using MicrobiomeAnalyst software. Linear discriminant analysis coupled with effect size analysis was used to identify differentially abundant features between mother and infant dyads. A total of 6,870,571 16S rRNA reads were generated from paired mother-infant saliva samples. Overall, oral microbial profiles significantly differed between the mother and infant groups (p < 0.001). The diversity of the salivary microbiomes in the infants increased in an age-dependent manner, whereas the core microbiome of the mothers remained relatively stable during the study period. Breastfeeding and gender did not affect the microbial diversity in infants. Moreover, infants had a greater relative abundance of Firmicutes and a lower abundance of Actinobacteria, Bacteroidetes, Fusobacteria, and Proteobacteria than their mothers. The SparCC correlation analysis demonstrated constant changes in infants' oral microbial community network (p < 0.05). This study provides new evidence that the oral cavities of infants are colonized by a distinct group of bacterial species at birth. The acquisition and diversity of changes in oral microbial composition are dynamic during the first year of an infant's life. Before reaching the second birthday, the composition of the oral microbial community could be more similar to that of their biological mothers.

Multiscale cascade recurrent dilation convolution network for fault diagnosis of rolling bearing under cross-load conditions

Recently, deep learning (DL) models based on convolutional neural networks have achieved satisfactory results in rolling bearing fault diagnosis. However, the bearings usually work in variable loading conditions, and their feature distribution could vary with load. The important features cannot be effectively captured in the convolution process using the existing diagnosis models, resulting in poor generalization performance. In this paper, a novel DL model, named multiscale cascade recurrent dilation convolution network, is proposed by introducing the dilated convolution and global average pooling (GAP) layer. Firstly, a new multiscale cascade structure with different convolution kernel sizes is introduced to extract multiscale features contained in the vibration signal. Secondly, a recurrent dilation convolution strategy is designed in each branch of the multiscale cascade structure to extract abundant feature information. Finally, the GAP is employed to reduce redundant feature vectors and output them, while a classifier of multilayer perceptron is used to automatically identify the fault types. The effectiveness of the proposed algorithm is evaluated by two experimental cases. The results show that the proposed method can successfully identify the labels of fault samples under unknown load conditions using the fault samples with labels under existing load conditions. Compared with other methods, this method exhibits excellent robustness and generalization performance for bearing fault diagnosis under cross-load conditions.

Assessment of ensemble learning for object-based land cover mapping using multi-temporal Sentinel-1/2 images

Accurate land cover mapping is challenging in Southeast Asia where cloud coverage is prevalent and landscape is heterogenous. Object-based mapping, multi-temporal images and combined use of optical and microwave data, provide abundant features in spectral, spatial, temporal, geometric and polarimetric dimensions. And random forest is usually employed due to robustness and efficiency in handling high-dimensional and noisy data. This study assesses whether feature selection and ensemble analysis, which are rarely adopted, yield improved result. Recursive feature elimination decreases original 568 features into a subset of 53 features, achieving the optimal combination of features. Ensemble analysis of random forest, support vector machine, and K-nearest neighbors leads to an overall accuracy of 0.816. Comparison experiments demonstrated the merits of the multi-temporal, multi-source approach, feature elimination and ensemble analysis.

Contrasting Features and Volcanostratigraphy of the Mafic-Hosted Mandoos and Shinas Volcanogenic Massive Sulfide Deposits, Samail Ophiolite, Oman

AbstractDespite the substantial amount of research on the Cretaceous Samail ophiolite in Oman, the factors controlling the size and metal endowment of the mafic-hosted, Cu-Au(-Zn-Ag) volcanogenic massive sulfide (VMS) deposits remain elusive. This work shows that the volcanostratigraphic position, hydrothermal venting style, and oxidation processes are critical factors controlling the distinct features of the Shinas and Mandoos deposits.Mandoos is a large (8 Mt, 1.8 wt % Cu, 0.18 g/ton Au) orebody preserving abundant primary vent-related features formed via mound growth and collapse within a wide hydrothermal field, overlying a poorly developed stockwork. The smaller Shinas deposit (0.8 Mt, 2.6 wt % Cu, 0.63 g/t Au) represents a higher-temperature system evolving from low fS2/fO2 conditions, locally sealed by jaspers, to a mound growth stage with widespread subseafloor brecciation/replacement with associated zone refining. Mandoos formed at the onset of the postaxial stage (Geotimes-Tholeiitic Alley transition), and Shinas is hosted within the Alley units. Volcanism in Samail was seemingly continuous, and the low ɛNd and Nb/Ta of the Shinas hanging-wall lavas record the onset of significant modifications of the mantle source during the postaxial stage.Mandoos is enriched in Te + As + Se ± Zn ± Ga ± Sb relative to Shinas, where higher Cu + Au + Tl ± Mo grades possibly reflect leaching of protoarc-like lavas. Rare earth element patterns in the ores mimicking the deposit footwall can be employed to constrain volcanostratigraphic positioning and indicate that the footwall lavas may also represent a source of metals. Formation of metal-rich ochres at the sea floor likely led to Cu + Au upgrading in the ores during seawater-induced oxidation, which was enhanced during subaerial gossan formation.

Predicting epileptic seizures based on EEG signals using spatial depth features of a 3D-2D hybrid CNN.

Epilepsy is a recurrent chronic brain disease that affects nearly 75 million people around the world. Therefore, the ability to reliably predict epileptic seizures would be instrumental for implementing interventions to reduce brain injury and improve patients' quality of life. In addition to classical machine learning algorithms and feature engineering methods, the use of electroencephalography (EEG) to predict seizures has gradually become a mainstream trend. Here, we propose a patient-specific method to predict epileptic seizures based on EEG data acquired using spatial depth features of a three-dimensional-two-dimensional hybrid convolutional neural network (3D-2D HyCNN) model. This method facilitates the acquisition of abundant and reliable deep features from multi-channel EEG signals. We first developed a reliable data preprocessing method to reconstruct time-series EEG signals into 3D feature images. Then, the 3D-2D HyCNN model was used to extract correlation features between multiple channels of EEG signals, which are automatically exploited by the network to improve seizure prediction. The method achieved accuracy of 98.43% and 93.11%, sensitivity of 98.58% and 90.98%, and specificity of 96.86% and 92.39% on the CHB-MIT Scalp EEG dataset and the American Epilepsy Society Epilepsy Prediction Challenge dataset, respectively. The results revealed that the new algorithm is reliable. Graphical Abstract A new patient-specific epilepsy prediction approach.

A novel meta-learning-based hyperspectral image classification algorithm

Aimed at the hyperspectral image (HSI) classification under the condition of limited samples, this paper designs a joint spectral–spatial classification network based on metric meta-learning. First, in order to fully extract HSI fine features, the squeeze and excitation (SE) attention mechanism is introduced into the spectrum dimensional channel to selectively extract useful HSI features to improve the sensitivity of the network to information features. Second, in the part of spatial feature extraction, the VGG16 model parameters trained in advance on the HSRS-SC dataset are used to realize the transfer and learning of spatial feature knowledge, and then, the higher-level abstract features are extracted to mine the intrinsic attributes of ground objects. Finally, the gated feature fusion strategy is introduced to connect the extracted spectral and spatial feature information on HSI for mining more abundant feature information. In this paper, a large number of experiments are carried out on the public hyperspectral dataset, including Pavia University and Salinas. The results show that the meta-learning method can achieve fast learning of new categories with only a small number of labeled samples and has good generalization ability for different HSI datasets.