Dual Modulation Research Articles

Unraveling the synergism mechanistic insight of O-vacancy and interfacial charge transfer in WO3-x decorated on Ag2CO3/BiOBr for photocatalysis of water pollutants: Based on experimental and density functional theory (DFT) studies

Photocatalysis has been widely used as one of the most promising approaches to remove various pollutants in liquid or gas phases during the last decade. The main emphasis of the study is on the synergy of vacancy engineering and heterojunction formation, two widely used modifying approaches, to significantly alter photocatalytic performance. The vacancy-induced Ag2CO3/BiOBr/WO3-x heterojunction system has been fabricated using a co-precipitation technique to efficiently abate methylene blue (MB) dye and doxycycline (DC) antibiotic. The as-fabricated Ag2CO3/BiOBr/WO3-x heterojunction system displayed improved optoelectronic characteristic features because of the rational combination of dual charge transferal route and defect modulation. The Ag2CO3/BiOBr/WO3-x system possessed 97% and 74% photodegradation efficacy for MB and DC, respectively, with better charge isolation and migration efficacy. The ternary photocatalyst possessed a multi-fold increase in the reaction rate for both MB and DC, i.e., 0.021 and 0.0078 min−1, respectively, compared to pristine counterparts. Additionally, more insightful deductions about the photodegradation routes were made possible by the structural investigations of MB and DC using density functional theory (DFT) simulations. This study advances the understanding of the mechanisms forming visible light active dual Z-scheme heterojunction for effective environmental remediation.

An uncertainty-aware domain adaptive semantic segmentation framework

Semantic segmentation is significant to realize the scene understanding of autonomous driving. Due to the lack of annotated real-world data, the technology of domain adaptation is applied so that the model is trained on the synthetic data and inferred on the real data. However, this domain gap leads to aleatoric and epistemic uncertainty. These uncertainties link to the potential safety issue of autonomous driving in normal weather and adverse weather. In this study, we explore the scientific problem that has received sparse attention previously. We postulate that the Dual Attention module can mitigate the uncertainty in the task of semantic segmentation and provide some empirical study to validate it. Furthermore, the utilization of Kullback-Leibler divergence (KL divergence) helps the estimation of aleatoric uncertainty and boosts the robustness of the segmentation model. Our empirical study on the diverse datasets of semantic segmentation demonstrates the effectiveness of our method in normal and adverse weather. Our code is available at: https://github.com/liubo629/Seg-Uncertainty-dual-attention.

MICCF: A Mutual Information Constrained Clustering Framework for Learning Clustering-Oriented Feature Representations

Deep clustering is a crucial task in machine learning and data mining that focuses on acquiring feature representations conducive to clustering. Previous research relies on self-supervised representation learning for general feature representations, such features may not be optimally suited for downstream clustering tasks. In this paper, we introduce MICCF, a framework designed to bridge this gap and enhance clustering performance. MICCF enhances feature representations by combining mutual information at different levels and employs an auxiliary alignment mutual information module for learning clustering-oriented features. To be specific, we propose a dual mutual information constraints module, incorporating minimal mutual information constraints at the feature level and maximal mutual information constraints at the instance level. This reduction in feature redundancy encourages the neural network to extract more discriminative features, while maximization ensures more unbiased and robust representations. To obtain clustering-oriented representations, the auxiliary alignment mutual information module utilizes pseudo-labels to maximize mutual information through a multi-classifier network, aligning features with the clustering task. The main network and the auxiliary one work in synergy to jointly optimize feature representations that are well-suited for the clustering task. We validate the effectiveness of our method through extensive experiments on six benchmark datasets. The results indicate that our method performs well in most scenarios, particularly on fine-grained datasets, where our approach effectively distinguishes subtle differences between closely related categories. Notably, our approach achieved a remarkable accuracy of 96.4% on the ImageNet-10 dataset, surpassing other comparison methods. The code is available at https://github.com/Li-Hyn/MICCF.git .

Dual modulation of electrolyte inner solvent structure and anode interface for high performance alkaline Al-air battery

In this work, we propose a Water-in-salt electrolyte based on concentrated ionic additive for Al-air batteries, aiming at inhibiting the unexpected severe hydrogen evolution self-corrosion. Computing calculations and experimental characterizations consistently confirm that the additive can significantly affect both the solvent structure and anode interface. A stronger H-bonding network is reconfigured and formed in the electrolyte, thus leading to a dramatic decrease in free water content and activity. Besides, the adsorption behavior of used additive molecules on Al anode surface plays an activating and stabilizing role, which is conducive to promoting the uniform dissolution of Al anode and improving the electrochemical performance. Owing to this dual modulation, the electrolyte system endows Al-air battery with extremely low self-corrosion rate (reduced by 89.6%) and remarkable discharge capacity of 2623 mAh/g.

Multi-scale and multi-path cascaded convolutional network for semantic segmentation of colorectal polyps

Colorectal polyps are structural abnormalities of the gastrointestinal tract that can potentially become cancerous in some cases. The study introduces a novel framework for colorectal polyp segmentation named the Multi-Scale and Multi-Path Cascaded Convolution Network (MMCC-Net), aimed at addressing the limitations of existing models, such as inadequate spatial dependence representation and the absence of multi-level feature integration during the decoding stage by integrating multi-scale and multi-path cascaded convolutional techniques and enhances feature aggregation through dual attention modules, skip connections, and a feature enhancer. MMCC-Net achieves superior performance in identifying polyp areas at the pixel level. The Proposed MMCC-Net was tested across six public datasets and compared against eight SOTA models to demonstrate its efficiency in polyp segmentation. The MMCC-Net's performance shows Dice scores with confidence interval ranging between 77.43 ± 0.12, (77.08, 77.56) and 94.45 ± 0.12, (94.19, 94.71) and Mean Intersection over Union (MIoU) scores with confidence interval ranging from 72.71 ± 0.19, (72.20, 73.00) to 90.16 ± 0.16, (89.69, 90.53) on the six databases. These results highlight the model's potential as a powerful tool for accurate and efficient polyp segmentation, contributing to early detection and prevention strategies in colorectal cancer.

Enhanced thermoelectric and mechanical performance of Bi2S3 bulk materials by compositing with Bi0.33(Bi6S9)Br nanorods

Recently, Bi2S3 has garnered significant interest in the thermoelectric field due to its abundant and low-toxicity constituents. Nevertheless, pure Bi2S3 material has not been utilized in thermoelectric applications because of its low electrical conductivity. This study presents the fabrication of the Bi0.33(Bi6S9)Br-doped Bi2S3 bulk samples with high electrical conductivity and mechanical performance via the melting method in conjunction with spark plasma sintering technology. The increased electron concentration is attributed to the replacement of S2− by Br− and the introduction of extra Bi in the lattice. At 323 K, the electrical conductivity of the Bi2S3 + 3 wt% Bi0.33(Bi6S9)Br sample increased to 208 Scm−1, signifying a 3-order-of-magnitude enhancement compared to the pure sample. The enhanced electrical conductivity led to the optimization of the electrical transport properties. At 573 K, the Bi2S3 + 2 wt% Bi0.33(Bi6S9)Br bulk sample achieved a peak power factor value of 481 μWm−1K−2, which is four times higher than that of the pure sample. Notably, the low lattice thermal conductivity of the Bi2S3 + 5 wt% Bi0.33(Bi6S9)Br sample was 0.56 W−1 m−1K−1 at 673 K. Given the significantly enhanced electrical transport properties and suppressed thermal conductivity, the Bi2S3 + 2 wt% Bi0.33(Bi6S9)Br sample achieved a peak ZT value of 0.45 at 673 K and a high ZTave value of 0.33 from 373 to 673 K. Compared to the pure Bi2S3 sample, these values are 5 times and 3 times higher, respectively. Such advancements can be implemented in the domain of power generation. Additionally, the mechanical properties of the sample exhibited substantial enhancement, and the average hardness of the 2 wt% Bi0.33(Bi6S9)Br-doped sample increased from 2.73 GPa of the pure sample to 3.03 GPa. This novel strategy of dual point defects modulation provides a new pathway to enhance the thermoelectric performance of Bi2S3 and other material systems.

Piperine induces autophagy of colon cancer cells: Dual modulation of AKT/mTOR signaling pathway and ROS production

BackgroundColorectal cancer (CRC) is a prevalent malignancy and poses a significant clinical challenge. Piperine, an alkaloid molecule extracted from Piper nigrum and Piper longum, has emerged as a promising anticancer agent. However, the molecular mechanisms of piperine’ antitumor effects in CRC need to be further elucidated. MethodsHuman colorectal cancer cells were treated with piperine in vitro. CCK-8 and clone formation assays were adopted to detect cell viability. The accumulation of autophagosomes was assessed by Western blotting and immunofluorescence. Apoptosis and reactive oxygen species (ROS) levels were analyzed by flow. In vivo, a xenograft tumor mouse model was constructed using CT26 cells. ResultsPiperine inhibited CRC cell viability and suppressed tumor weight and volume in a mouse model. Additionally, piperine treatment induced the accumulation of autophagosomes in CRC cells. This effect was attributed to the inhibition of the AKT/mTOR pathway and the accumulation of ROS. activation of AKT or clearance of ROS attenuated piperine-mediated tumor suppression. ConclusionThis study shows that piperine induces autophagy-dependent cell death in CRC cells by increasing ROS production and inhibiting Akt/mTOR signaling.

Taming of trinitromethyl-oxadiazole to access high density and high oxygen balance via a dual modulation strategy

Energetic compounds bearing the trinitromethyl group are garnering broad attraction as potential candidates for a new generation of high energy dense oxidizers. In this work, an effective dual modulation strategy involving both molecular isomerization and crystal morphology control was employed to design and optimize trinitromethyl-oxadiazole with improved comprehensive performance. Utilizing this dual strategy, 3,5-bis(trinitromethyl)-1,2,4-oxadiazole (3) was synthesized, resulting in the formation of two distinct crystal morphologies (needle and sheet) corresponding to two crystal forms (3-a and 3-b). Encouragingly, while maintaining ultra-high oxygen balance (21.73%), 3 achieves impressive densities (1.97–1.98 g/cm3). To our knowledge, the density of 1.98 g/cm3 for 3-a sets a new record among that of nitrogen-rich monocyclic compounds. Notably, practical crystal morphology prediction was creatively introduced to guide the experimental crystallization conditions of 3, increasing the impact sensitivity and friction sensitivity from 1 J to 80 N (3-a) to 10 J and 240 N (3-b), respectively. Additionally, the crystal structural analyses and theoretical calculations were conducted to elucidate the reasons of differences between 3-a and 3-b in density and stability. This work provides an efficient strategy to enhance performance of trinitromethyl derivatives, broadening the path and expanding the toolbox for energetic materials.

Photoisomerization-modulated supramolecular chirality inversion and energy-transfer efficiency in a cyanostilbene-glutamide-based self-assembly system

Manipulating chirality and energy transfer simultaneously in a single self-assembly system is important but remains challenging. In this study, we design a cyanostilbene-glutamide-based amphiphile L-GC bearing a long alkyl chain. It can form gels in methanol and exhibit circular dichroism (CD) signals. Upon ultraviolet-light irradiation, the CD signals are inverted and the morphology transforms from fibers to spheres. Chirality and energy transfer can be achieved when L-GC co-assembles with the energy acceptor DBTOC18, which is conjugated to a long alkyl chain. Moreover, the energy-transfer efficiency of the co-assembly can be modulated via photoirradiation-induced isomerization. Molecular dynamics simulation was conducted to elucidate the photomodulated assembly behavior. This study realizes the dual modulation of chirality and energy and presents some perspectives for information exchange in chiroptical materials.

Short-term subway passenger flow forecasting approach based on multi-source data fusion

Accurate short-term subway passenger flow forecasting is essential in improving the efficiency of daily operation and the management of subway system. Existing research often overlooks the impacts of randomness and unevenness in passenger flows caused by population migrations, also predominantly emphasizes feature selection while neglecting effective feature fusion. To address this gap, we propose a novel hybrid subway passenger flow forecasting approach adopting multi-source data fusion technique. Firstly, the dual processing module consisting of Grey Relation Analysis and SHapley Additive exPlanations is used to gather influencing features. Then, the multivariate empirical mode decomposition decomposes the selected features into multidimensional intrinsic mode functions. To discern and extract the effective components from the decomposed intrinsic mode functions, secondary feature selection module based on Spearman Correlation Coefficient is performed. Finally, the determined components are categorized into three classes by combining recurrence plot and picture information entropy, each class is assigned suitable predictor to ensure accurate forecasting. Experimental results based on three real subway passenger flow datasets from Guangzhou, Beijing, and Chengdu verifies the effectiveness and robustness of the proposed approach. Therefore, the introduction of population migration index and the effective fusion of information are significant to improve subway passenger flow forecasting.

New insights and reaction mechanisms in the design of catalysts for the synergistic removal of NOx and VOCs from coke oven flue gas: Dual regulation of oxidative properties and acidic sites

The acid and redox sites of the MnCo catalysts are simultaneously fine-tuned by the addition of V. A dual-function catalyst, designated as V0.5Mn5Co5, has been constructed for the synergistic removal of NOx and volatile organic compounds under coke-oven flue gas conditions, which exhibits > 95 % NOx conversion and > 80 % N2 selectivity at 180–300 °C. Meanwhile, it removes 70 % of ethylene at 240 °C. Besides it has excellent sulfur and water resistance. The characterization results indicate that this acid-redox dual sites modulation strategy appropriately weakens the oxidation capacity of the catalysts while increasing the surface acidity of the catalysts. The catalyst mainly performs SCR reaction through the E-R mechanism, and N2O is generated through the transition dehydrogenation of NH3 and NSCR reaction. Ethylene is first adsorbed on the catalyst surface then oxidized to form carbonate species, and finally decomposed to CO2. Ethylene oxidation follows the MvK mechanism. There is a competitive adsorption between NH3 and C2H4, and a mutual inhibition between the SCR reaction and the ethylene oxidation reaction. V0.5Mn5Co5 exhibits excellent synergistic removal of NOx and VOCs in coke oven flue gas compared with commercial VWTi catalysts, which indicates great promise for industrial application.

Generation and autocorrelation traces investigation of Nyquist pulse sequences

We propose a novel approach for generating Nyquist pulse sequences using a dual parallel Mach-Zehnder modulator (DPMZM) and a phase modulator (PM). The generated Nyquist pulse sequence achieves a repetition frequency of 10 GHz and a full width at half maximum (FWHM) of 4.22 ps. The time-domain waveforms of these sequences are retrieved using the phase and intensity from cross-correlation and spectrum only (PICASO) method. Further investigation into the autocorrelation traces reveals a pulse width conversion factor of approximately 0.75 between the Nyquist pulse sequences and their autocorrelation traces. Our work offers an alternative approach to generating and measuring Nyquist pulse sequences.

Enhancing cognitive recovery in chronic traumatic brain injury through simultaneous allosteric modulation of α7 nicotinic acetylcholine and α5 GABAA receptors

Traumatic brain injury (TBI) leads to changes in the neural circuitry of the hippocampus that result in chronic learning and memory deficits. However, effective therapeutic strategies to ameliorate these chronic learning and memory impairments after TBI are limited. Two pharmacological targets for enhancing cognition are nicotinic acetylcholine receptors (nAChRs) and GABAA receptors (GABAARs), both of which regulate hippocampal network activity to form declarative memories. A promising compound, 522–054, both allosterically enhances α7 nAChRs and inhibits α5 subunit-containing GABAARs. Administration of 522–054 enhances long-term potentiation (LTP) and cognitive functioning in non-injured animals. In this study, we assessed the effects of 522–054 on hippocampal synaptic plasticity and learning and memory deficits in the chronic post-TBI recovery period. Adult male Sprague Dawley rats received moderate parasagittal fluid-percussion brain injury or sham surgery. At 12 wk after injury, we assessed basal synaptic transmission and LTP at the Schaffer collateral-CA1 synapse of the hippocampus. Bath application of 522–054 to hippocampal slices reduced deficits in basal synaptic transmission and recovered TBI-induced impairments in LTP. Moreover, treatment of animals with 522–054 at 12 wk post-TBI improved cue and contextual fear memory and water maze acquisition and retention without a measurable effect on cortical or hippocampal atrophy. These results suggest that dual allosteric modulation of α7 nAChR and α5 GABAAR signaling may be a potential therapy for treating cognitive deficits during chronic recovery from TBI.

UCP2-ACGAN: An adaptive condition GAN guided by U-shaped context perceptual processor for PET/CT images fusion

In order to better highlight the lesion area information and edge detailed information of the fused images, an adaptive condition Generative Adversarial Network (GAN) guided by U-shaped context perceptual processor (UCP2-ACGAN) for Positron Emission Tomography (PET) and Computed Tomography (CT) images fusion is proposed in this paper. Firstly, the network architecture of single generator and dual relative discriminators with weights sharing is used in UCP2-ACGAN. Secondly, a multi-granularity adaptive feature extraction module from coarse to fine (MGAFE) is proposed in generator, it is used to extract the valid information contained in the source images. Thirdly, a dual branch feature interactive fusion module guided by U-shaped context perceptual processor (UCP2-DBFIF) is proposed in generator. The context perceptual feature maps are obtained by U-shaped context perceptual processor (UCP2) and multiply them with the feature extraction results as the guiding conditions in the fusion process. In addition, in order to supplement the effective information lost during the feature fusion process, this paper converges the feature information of different granularity during the feature extraction stage. Finally, compared to the best of the 6 comparison methods, the results of comparison experiments show that UCP2-ACGAN is effective on several evaluation metrics values. For example, on the evaluation metrics of Average Gradient (AG), Edge Intensity (EI), Spatial Frequency (SF), Visual Information Fidelity (VIF) and Mutual Information (MI), the metrics values of CT mediastinal window images and PET fused images average increased by 30.74%, 29.35%, 40.24%, 35.47% and 26.18%.

Novel high-valent copper species-involved photo-Fenton-like process boosted by nitrogen/oxygen dual defects modulation

High-valent metal species generated during peroxymonosulfate (PMS) activation have demonstrated significant potential in wastewater treatment. However, the rapid generation of high-valent metal species, especially high-valent copper (Cu(III)), has been challenging. Moreover, the mechanisms behind Cu(III) generation and its role in the PMS-based Fenton-like process have not been fully understood. Herein, a novel N, O dual defects co-modified 0D/2D hybrid of CuO nanodots (NDs)/g-C3N4 nanosheets was constructed as an integrated photocatalytic platform to efficiently activate PMS, which exhibited excellent catalytic activity and stability towards tetracycline degradation through consecutive generation of surface Cu(III) as the key active species. The N, O dual-defects and strong interfacial coupling of NDs/g-C3N4 markedly increased charge separation and induced electron-rich Cu sites. This led to a continuous and rapid Cu(II)/Cu(I) conversion, further triggering persistent Cu(III) generation through Cu(I)-oxidation with a two-electron transfer. The work provides new insights into exploring the properties of Cu(III) species during PMS activation and offers encouraging guidance for their application in pollutant removal.

Performance evaluation of attention-deep hashing based medical image retrieval in brain MRI datasets

BackgroundBrain MRI images pose significant challenges due to their complexity and voluminous data, which often hinder the accuracy of traditional image retrieval methods. In response, this research delves into a novel medical image retrieval approach grounded in attention mechanisms and deep hashing techniques. MethodologyThe study enhances the network's edge perception capability by incorporating a dual mixed attention module (CBAM-MA dual attention) into the convolutional neural network architecture. This addition optimally captures edge saliency and distributional features by introducing an extra attention layer atop CBAM. Furthermore, a novel triple loss function is introduced, amalgamating Euclidean and cosine distances. This fusion exploits the strengths of both distance metrics to comprehensively consider sample geometric attributes, thereby generating more robust and expressive feature representations crucial for preserving categorical details. ResultsExperimental evaluations showcase significant performance advancements in MRI image retrieval tasks using the proposed method. The average hit rate stands at 0.7783, average precision at 0.752246, and average inverse rank at 0.956721. These metrics collectively underscore the method's efficacy in enhancing the quality and expediting the diagnosis of brain diseases. ConclusionThe research underscores the critical role of attention and deep hashing techniques in augmenting the accuracy and efficiency of medical image retrieval, particularly in the context of brain MRI images. The proposed method's notable performance improvements pave the way for enhanced diagnostic capabilities, signaling a significant step forward in improving healthcare outcomes related to brain diseases.

Enhanced methanol electrooxidation catalysis via dual modulation of PtCu alloy and oxygen vacancies

Achieved by introducing defect sites, enhances both the catalytic activity and stability of catalysts. This is crucial for advancing key applications in energy conversion and environmental protection. Here, we introduce a novel Pt-Cu@CeO2/MWCNTs electrocatalyst prepared by dual modulation of the oxygen vacancy anchoring point and the center of the alloying d-band. This catalyst significantly boosts the rate of methanol oxidation. The Pt-Cu@CeO2/MWCNTs catalyst demonstrates a mass activity 3.7 times higher than that of commercial Pt/C, along with exceptional durability. The electron transfer between the PtCu alloy and the oxygen vacancies effectively modulates the d-band center of Pt, leading to a “volcano” relationship between the catalyst’s mass activity and the d-band center. This study advances the development of methanol electrooxidation catalysts by introducing a novel “double enhancement effect” model.

Incidence and time to onset of immunotherapy-related adrenal insufficiency in the I-SPY2 trial.

584 Background: I-SPY2 was the first randomized trial to demonstrate an improvement in pCR and EFS when Immune Checkpoint Inhibitors (ICI) were added to standard neoadjuvant chemotherapy. This finding was validated in the phase 3 KEYNOTE 522 trial for triple-negative breast cancer, leading to approval of pembrolizumab (pembro) in this setting. ICIs add the risk of immune-related adverse events (irAEs); while most irAEs are reversible, endocrinopathies appear to persist. Higher rates of adrenal insufficiency (AI), either from primary adrenalitis or hypophysitis, have been reported with ICI-based therapy (tx) in early breast cancer (EBC) compared to advanced disease. Here we describe the incidence and time of onset of AI with ICI-based tx in I-SPY2. Methods: I-SPY2 is a multicenter, phase 2 trial using response-adaptive randomization within biomarker subtypes defined by hormone receptor, HER2, and MammaPrint status to evaluate novel agents in the neoadjuvant setting in patients (pts) with high-risk EBC. Treatment included weekly paclitaxel (T) plus ICI-based tx x 4 cycles, followed by doxorubicin/cyclophosphamide (AC) every 2-3 weeks x 4. 6 completed investigational arms included ICIs: T+pembro x 4, T+pembro x 8 (+/- AC), T+pembro+SD101 (TLR9 agonist), T+durvalumab+olaparib, T+cemiplimab (cemi), and T+cemi+REGN3767 (anti-LAG-3). AI diagnoses were made by treating providers with review from two board certified endocrinologists. AI was defined as AM serum cortisol level < 3 mcg/dl in absence of recent glucocorticoid treatment. If ACTH was low or normal, pt was dx with hypophysitis; if ACTH was elevated, pt was dx with primary AI. Results: 442 pts in I-SPY2 received ICI-based tx; 11.8% (n=52) developed AI (primary+hypophysitis). Mean age of pts who developed AI was 47.3 vs 48.6 yrs in those who did not (p=0.44). Incidence of AI was highest in arms with dual immune modulation (18.1%) as compared to single ICI tx (8.5%). Median time to onset varied by arm, ranging from 92-204 days from first ICI dose. While some cases occurred during ICI tx (13.5%), the majority occurred between 12-24 wks after tx initiation, with 21.2% diagnosed after completion of all tx. The majority (94.2%) of pts with AI were symptomatic; the most common presenting symptoms included fatigue (59.6%), nausea (48.1%), and vomiting (34.6%). All pts with AI remain on glucocorticoid replacement tx. Conclusions: The incidence of AI in EBC is higher than has previously been reported in advanced disease, with higher rates observed in those tx with dual immune modulation. No correlation was observed with age; correlation of AI with response predictive subtype and other clinicopathologic features is ongoing and will be presented. Given the high incidence of AI in EBC, close monitoring in the peri/postoperative setting and education of pts and providers is critical, as risk persists even after completion of tx. Work to identify risk factors for AI is ongoing. Clinical trial information: NCT01042379 .

Enhanced Wind Field Spatial Downscaling Method Using UNET Architecture and Dual Cross-Attention Mechanism

Before 2008, China lacked high-coverage regional surface observation data, making it difficult for the China Meteorological Administration Land Data Assimilation System (CLDAS) to directly backtrack high-resolution, high-quality land assimilation products. To address this issue, this paper proposes a deep learning model named UNET_DCA, based on the UNET architecture, which incorporates a Dual Cross-Attention module (DCA) for multiscale feature fusion by introducing Channel Cross-Attention (CCA) and Spatial Cross-Attention (SCA) mechanisms. This model focuses on the near-surface 10-m wind field and achieves spatial downscaling from 6.25 km to 1 km. We conducted training and validation using data from 2020–2021, tested with data from 2019, and performed ablation experiments to validate the effectiveness of each module. We compared the results with traditional bilinear interpolation methods and the SNCA-CLDASSD model. The experimental results show that the UNET-based model outperforms SNCA-CLDASSD, indicating that the UNET-based model captures richer information in wind field downscaling compared to SNCA-CLDASSD, which relies on sequentially stacked CNN convolution modules. UNET_CCA and UNET_SCA, incorporating cross-attention mechanisms, outperform UNET without attention mechanisms. Furthermore, UNET_DCA, incorporating both Channel Cross-Attention and Spatial Cross-Attention mechanisms, outperforms UNET_CCA and UNET_SCA, which only incorporate one attention mechanism. UNET_DCA performs best on the RMSE, MAE, and COR metrics (0.40 m/s, 0.28 m/s, 0.93), while UNET_DCA_ars, incorporating more auxiliary information, performs best on the PSNR and SSIM metrics (29.006, 0.880). Evaluation across different methods indicates that the optimal model performs best in valleys, followed by mountains, and worst in plains; it performs worse during the day and better at night; and as wind speed levels increase, accuracy decreases. Overall, among various downscaling methods, UNET_DCA and UNET_DCA_ars effectively reconstruct the spatial details of wind fields, providing a deeper exploration for the inversion of high-resolution historical meteorological grid data.

A robust multi-scale feature extraction framework with dual memory module for multivariate time series anomaly detection

Although existing reconstruction-based multivariate time series anomaly detection (MTSAD) methods have shown advanced performance, most assume the training data is clean. When faced with noise or contamination in training data, they can also reconstruct the anomaly well, weakening the distinction between normal and anomaly. Some probabilistic generation-based methods have been used to address this issue because of their implicit robust structure to noise, but the training process and suppression of anomalous generalization are not stable. The recently proposed explicit method based on the memory module would also sacrifice the reconstruction effect of normal patterns, resulting in limited performance improvement. Moreover, most existing MTSAD methods use a single fixed-length window for input, which weakens their ability to extract long-term dependency. This paper proposes a robust multi-scale feature extraction framework with the dual memory module to comprehensively extract features fusing different levels of semantic information and lengths of temporal dependency. First, this paper designs consecutive neighboring windows as inputs to allow the model to extract local and long-term dependency information. Secondly, a dual memory-augmented encoder is proposed to extract global typical patterns and local common features. It ensures the reconstruction ability of normal data while suppressing the generalization of the anomaly. Finally, this paper proposes a multi-scale fusion module to fuse latent variables representing different levels of semantic information and uses the reconstructed latent variables to reconstruct samples for anomaly detection. Experimental results on five datasets from diverse domains show that the proposed method outperforms 16 typical baseline methods.