Source Model Research Articles

Dual Classifier Adaptation: Source-Free UDA via Adaptive Pseudo-Labels Learning

Different from Unsupervised Domain Adaptation (UDA), Source-Free Unsupervised Domain Adaptation (SFUDA) transfers source knowledge to target domain without accessing the source data, using only the source model, has attracted much attention recently. One mainstream SFUDA method fine-tunes the source model by self-training to generate pseudo-labels of the target data. However, due to the significant differences between different domains, these target pseudo-labels often contain some noise, and it will inevitably degenerates the target performance. For this purpose, we propose an innovative SFUDA method with adaptive pseudo-labels learning named Dual Classifier Adaptation (DCA). In DCA, a dual classifier structure is introduced to adaptively learn target pseudo-labels by cooperation between source and target classifiers. Simultaneously, the minimax entropy is introduced for target learning, in order to adapt target data to source model, while capture the intrinsic cluster structure in target domain as well. After compared our proposed method DCA with a range of UDA and SFUDA methods, DCA achieves far ahead performance on several benchmark datasets.

A Generic Model for a Persistent Radio Source around Fast Radio Bursts

A Generic Model for a Persistent Radio Source around Fast Radio Bursts

Harmonizing financial futures: AI enhancement and methodical fusion in banking risk oversight

The paper will deal with the link between finance and artificial intelligence (AI) in the context of banking, including how AI was introduced from simple tasks to more complex ones involving risk management and fraud detection. The phenomenon displays use of AI by banks as exemplified by First Financial Bank, which was able to open new accounts through intelligent digital assistants. The article focuses on the question of ethical AI that aims to create trust and compliance. This article has explored the position that big language models (LLMs) hold in enhancing virtual assistants in banking and the significance of open source models, as they contribute to collaboration and innovation in the financial industry.

Analysis of the Occurrent Models of Potential Debris-Flow Sources in the Watershed of Ching-Shuei River

The areas around the Ching-Shuei River saw numerous landslides (2004–2017) after the Jiji earthquake, profoundly harming the watershed’s geological environment. The 33 catchment areas in the Ching-Shuei River watershed and five typhoon and rainstorm events, with a total of 165 occurrences and non-occurrences, were analyzed, and the training and validation were categorized into 70% training and 30% validation. A landslide disaster is deemed, for the purposes of this research, to have taken place if SPOT satellite images taken before and after an incident show a Normalized Difference Vegetation Index difference larger than 0.25, a slope of less than 30 degrees, and a number of connected grids greater than 10. The analysis was carried out using the instability index method analysis with Rogers regression analysis and artificial neural network. The accuracy rates of neural network, logit regression, and instability index analyses were, respectively, 93.3%, 80.6%, and 70.9%. The neural network’s area under the curve was 0.933, indicating excellent discrimination ability; that of the logit regression analysis was 0.794, which is considered good; and that of the instability index analysis was 0.635, or fair. This suggests that any of the three models are suitable for the danger assessment of large post-earthquake debris flows. The results of this study also provide a reference and evidence for specific sites’ potential susceptibility to debris flows.

Transfer‐Learning‐Enabled 3D reconfigurable broadband solar metamaterial absorbers design

Research on metamaterials shows excellent potential in the field of solar energy harvesting. In recent years, the design of broadband solar metamaterial absorbers (SMAs) has attracted significant interest with the wide application of deep learning methods. This paper proposes a deep neural network (DNN) to realize forward prediction and inverse design of reconfigurable 3D SMAs. In the inverse design, a polarization-insensitive broadband SMA with an absorption bandwidth of 2.7 μm and an average absorption rate of 97.6% with an adjustable bandwidth range of 369 nm is successfully designed. The design of SMAs with different structures is also realized by a transfer learning method to improve the training speed and performance further. Using the transfer learning approach, the training speed of the neural network target model can be accelerated, and its training performance can be improved on small datasets by utilizing the trained neural network source model. Meanwhile, using the trained inverse design target model, a polarization-insensitive broadband SMA was designed with an absorption bandwidth of 2.7 μm, an average absorption of 97.9%, and an adjustable bandwidth range of 141 nm. Finally, we verified the solar energy harvesting capability of the designed broadband SMAs under real-world conditions using air mass (AM) 1.5, and they were calculated to be capable of harvesting the vast majority of the energy. The method is instructive in the design process of SMAs and can be effectively used to explore multifunctional complex nanophotonic devices.

Neutron Activation Analysis Based on AB-BNCT Treatment Room.

Boron neutron capture therapy (BNCT) is an ideal binary targeted radiotherapy for treating refractory tumors. An accelerator-based BNCT (AB-BNCT) neutron source has attracted more and more attention due to its advantages such as higher neutron yield in the keV energy region, less gamma radiation, and higher safety. In addition to 10B, neutrons also react with other elements in the treatment room during BNCT to produce many activation products. Due to the long half-life of some activation products, there will be residual radiation after the end of treatment and the shutdown of the accelerator, which has adverse effects on radiation workers. Therefore, the ambient dose equivalent rate in the treatment room needs to be evaluated. The AB-BNCT neutron source model proposed by Li is studied in this paper. Based on the Monte Carlo method, the Geant4 platform was used to simulate the dose induced by radionuclides near the Beam Shaping Assembly (BSA) of the source. It is concluded that the concrete wall contributed the most to the radiation dose. The dose rate of 2.45 μSv h-1 after 13 min of shutdown meets the dose rate limit of 2.5 μSv h-1, at which point it is safe for workers to enter the treatment room area.

Study on thermal penetration process of oil casing in shallow water environment by plasma arc based on heat source simulation and experiments

Abandoned offshore oil wellheads must be removed and plugged to protect the marine environment. Underwater plasma arc cutting holds the promise of playing a significant role in this work. However, much remains unknown about the thermal burn-off process of underwater plasma arcs for oil well casing. This work investigated the thermal penetration process of oil casing in a shallow water environment through experiments and heat source modeling simulations. A workpiece made of mild steel plate and cement block was considered as a small segment of the oil casing to be penetrated. The hole diameters, depths, and removal volumes of the workpieces were investigated experimentally. A combined heat source consisting of an elliptical cone heat source and a semi-ellipsoidal heat source was established based on the hole shape. The simulated and experimental hole profiles matched well. The results showed that the evolution of the hole shape of the workpiece was related to the motion state of the plasma arc and could be reflected by the variation of the arc voltage. Compared to steel plates, cement materials were easily removed by heat. These findings are favorable to promote the development of underwater plasma arc cutting in oil casing removal.

Probabilistic earthquake-tsunami financial risk evaluation for the District of Tofino, British Columbia, Canada

ABSTRACT Southwestern British Columbia in Canada has a high likelihood of facing significant earthquake threats in the future. It is exposed to significant seismic and tsunami hazards, originating from the Cascadia subduction zone and shallow crustal and deep inslab earthquakes. This study presents a new probabilistic earthquake-tsunami loss model for the Cascadia subduction zone by focusing on the District of Tofino, British Columbia. The earthquake occurrence and rupture models for the Cascadia subduction zone are developed by incorporating the time-dependency of earthquake occurrence and by adopting a stochastic source modelling approach, which allows for consideration of heterogeneous earthquake slip distributions. The results produce single-hazard and multi-hazard exceedance probability loss curves for earthquakes and tsunamis from the Cascadia subduction zone. In addition, earthquake risks due to other seismic sources are integrated with the multi-hazard loss curves due to the Cascadia events. In this way, a complete financial exposure from all major seismic and tsunami sources is evaluated. These loss curves can be used for determining the insurance rates for multi-hazard and multi-source risk coverage. The results are beneficial for providing more risk financing options via insurance against the future Cascadia and other seismic events.

Resilience-Oriented Planning of Urban Distribution System Source–Network–Load–Storage in the Context of High-Penetrated Building-Integrated Resources

Building-integrated flexible resources can offer economical availability to accommodate high-penetrated renewable energy sources (RESs), which can be potentially coordinated to achieve cost-effective supply. This paper proposes a resilience-oriented planning model of urban distribution system source–network–load–storage in the context of high-penetrated building-integrated resources. In this model, source–network–load–storage resources are cost-optimally planned, including the lines, soft open point (SOP), building-integrated photovoltaics (BIPVs), building-integrated wind turbine (BIWT), building-integrated energy storage system (ESS), etc. To enhance fault recovery capability during extreme faults, fault scenarios are incorporated into the distribution system operation via coupled multiple recovery stages. The resilience-oriented planning is a thorny problem due to its source–network–load–storage couplings, normal-fault couplings, etc. The original resilience-oriented planning is reformulated as a mixed-integer linear programming (MILP) problem, which can then be solved with a two-stage method and evaluated via a multi-dimensional evaluation metrics. The proposed planning methodology is benchmarked over a Portugal 54-node urban distribution system to verify the superiority and effectiveness on the system economy and resilience levels. Case studies show that the proposed methodology can exploit the optimal synergies of different source–network–load–storage components and enhance system dispatchability.

Static performance prediction of long-pulse negative ion based neutral beam injection experiment

The mission of negative ion-based neutral beam injection (NNBI) is to conduct experiments with pulses lasting thousands of seconds. It is crucial to develop a simplified physical calculation model for the long-pulse negative ion source in the current NNBI device. This model will be used to evaluate the advantages and disadvantages of the selected parameters prior to the experiment, and to assist in adjusting and establishing the experimental parameters for the long-pulse ion source experiment. This paper presents the development of a static performance prediction model using a back propagation neural network. The model assesses the yield of negative hydrogen ions and the quantity of electrons in the ion source under specific parameter conditions, utilizing various experimental parameters as input. The experimental data used for this model are derived from historical data generated during the operation of the 2022 NNBI experiment. The test results indicate that under the current optimal hyperparameter condition, the prediction accuracy of H− ion current (I_H−) is 80.84%, and the prediction accuracy of extraction grid electronic current (I_EG) is 77.57%. This can effectively prevent invalid shots, accurately assess the advantages and disadvantages of the input parameters, and enhance the performance of the long-pulse NNBI device.

Acceleration mechanisms of energetic ion debris in laser-driven tin plasma EUV sources

Laser-driven tin plasmas are driving new-generation nanolithography as sources of extreme ultraviolet (EUV) radiation centered at 13.5 nm. A major challenge facing industrial EUV source development is predicting energetic ion debris produced during the plasma expansion that may damage the sensitive EUV channeling multilayer optics. Gaining a detailed understanding of the plasma dynamics and ion acceleration mechanisms in these sources could provide critical insights for designing debris mitigation strategies in future high-power EUV sources. We develop a fully kinetic model of tin-EUV sources using one-dimensional particle-in-cell simulations to study ion debris acceleration, which will be valuable for cross-validation of radiation-hydrodynamic simulations. An inverse-bremsstrahlung heating operator is used to model the interaction of a tin target with an Nd:YAG laser, and thermal conduction is included through a Monte Carlo Coulomb collision operator. While the large-scale evolution is in reasonable agreement with analogous hydrodynamic simulations, the significant timescale for collisional equilibration between electrons and ions allows for the development of prominent two-temperature features. A collimated flow of energetic ions is produced with a spectrum that is significantly enhanced at high energies compared to fluid simulations. The dominant acceleration mechanism is found to be a large-scale electric field supported mainly by the electron pressure gradient, which is enhanced in the kinetic simulations due to the increased electron temperature. We discuss the implications of these results for future modeling of tin-EUV sources and the development of debris mitigation schemes.

Numerical analysis of static and dynamic heat source model approaches in laser micro spot welding

Numerical analysis of static and dynamic heat source model approaches in laser micro spot welding

RFL-CDNet: Towards accurate change detection via richer feature learning

Change Detection is a crucial but extremely challenging task of remote sensing image analysis, and much progress has been made with the rapid development of deep learning. However, most existing deep learning-based change detection methods mainly focus on intricate feature extraction and multi-scale feature fusion, while ignoring the insufficient utilization of features in the intermediate stages, thus resulting in sub-optimal results. To this end, we propose a novel framework, named RFL-CDNet, that utilizes richer feature learning to boost change detection performance. Specifically, we first introduce deep multiple supervision to enhance intermediate representations, thus unleashing the potential of backbone feature extractor at each stage. Furthermore, we design the Coarse-To-Fine Guiding (C2FG) module and the Learnable Fusion (LF) module to further improve feature learning and obtain more discriminative feature representations. The C2FG module aims to seamlessly integrate the side prediction from previous coarse-scale into the current fine-scale prediction in a coarse-to-fine manner, while LF module assumes that the contribution of each stage and each spatial location is independent, thus designing a learnable module to fuse multiple predictions. Experiments on several benchmark datasets show that our proposed RFL-CDNet achieves state-of-the-art performance on WHU cultivated land dataset and CDD dataset, and the second best performance on WHU building dataset. The source code and models are publicly available at https://github.com/Hhaizee/RFL-CDNet.

Numerical modelling of impact seismic sources using the stress glut theory

Numerical modelling of impact seismic sources using the stress glut theory

On the Stability of Mantle‐Sensitive P‐Wave Interference During a Secondary Microseismic Event

AbstractBody wave extraction from oceanic secondary microseismic sources with seismic interferometry provides alternative information to better constrain the Earth's structure. However, sources' spatiotemporal variations raise concerns about travel time measurement robustness. Therefore, we study the cross‐correlations’ stability during a single oceanic event. This study focuses on 3 days of data and three seismic arrays' combinations between 8 and 11 December 2014 during storm Alexandra, a “weather bomb” event in southern Greenland. We use the WAVEWATCH III hindcast to model P‐wave noise sources and assess the impact of short‐term source variations on cross‐correlations. Model‐based cross‐correlations compared to data show coherent delays to reference 3D Earth models (∼0–3 s) confirming the robustness of the source model which could explain minor travel time variations (≤1 s).

A photon source model for alpha-emitter radionuclides

Objective. A Monte Carlo virtual source model named PHID (photon from Ion decay) that generates photons emitted in the complex decay chain process of alpha-emitter radionuclides is proposed, typically for use during the simulation of SPECT image acquisition. Approach. Given an alpha-emitter radionuclide, the PHID model extracts from Geant4 databases the photon emission lines from all decaying daughters for both isometric transition and atomic relaxation processes. According to a given time range, abundances and activities in the decay chain are considered thanks to the Bateman equations, taking into account the decay rates and the initial abundances. Main results. PHID is evaluated by comparison with analog Monte Carlo simulation. It generates photons with the correct energy and temporal distribution, avoiding the costly simulation of the complete decay chain thus decreasing the computation time. The exact time gain depends on the simulation setup. As an example, it is 30× faster for simulating 1 MBq of 225Ac in water for 1 section Moreover, for 225Ac, PHID was also compared to a simplified source model with the two main photon emission lines (218 and 440 keV). PHID shows that 2 times more particles are simulated and 60% more counts are detected in the images. Significance. PHID can simulate any alpha-emitter radionuclide available in the Geant4 database. As a limitation, photons emitted from Bremsstrahlung are ignored, but they represent only 0.7% of the photons above 30 keV and are not significant for SPECT imaging. PHID is open-source, available in GATE 10, and eases the investigation of imaging photon emission from alpha emitters.

The Association between Parental Child Vaccination Refusal Rate and the Impact of Mass Vaccination against COVID-19 in Kazakhstan: An Interrupted Time Series Analysis with Predictive Modelling of Nationwide Data Sources from 2013 to 2022.

Despite well-established evidence supporting vaccination efficacy in reducing morbidity and mortality among infants and children, there is a global challenge with an increasing number of childhood vaccination refusals. This issue has intensified, especially during the COVID-19 pandemic. Our study aims to forecast mandatory childhood vaccination refusal trends in Kazakhstan until 2030, assessing the impact of mass COVID-19 vaccination on these rates. Utilizing annual official statistical data from 2013 to 2022 provided by the Ministry of Health of Kazakhstan, the study reveals a significant surge in refusals during the pandemic and post-pandemic periods, reaching record levels of 42,282 cases in 2021 and 44,180 cases in 2022. Notably, refusal rates sharply rose in specific regions, like Aktobe (13.9 times increase) and Atyrau (4.29 times increase), emphasizing the need for increased public healthcare attention in these areas. However, despite a decade of data, our forecasting analysis indicates a lack of volatility in childhood vaccine refusal trends for all vaccine types up to 2030, highlighting the statistical significance of the obtained results. The increasing trend in vaccine refusals underscores the necessity to enhance crisis response and support health initiatives, particularly in regions where a substantial rise in refusals has been observed in recent years.

Evaluation of the X-ray/EUV Nanolithography Facility at AS through wavefront propagation simulations.

Synchrotron light sources can provide the required spatial coherence, stability and control to support the development of advanced lithography at the extreme ultraviolet and soft X-ray wavelengths that are relevant to current and future fabricating technologies. Here an evaluation of the optical performance of thesoft X-ray (SXR) beamline of the Australian Synchrotron (AS) and its suitability for developing interference lithography using radiation in the 91.8 eV (13.5 nm) to 300 eV (4.13 nm) range are presented. A comprehensive physical optics model of the APPLE-II undulator source and SXR beamline was constructed to simulate the properties of the illumination at the proposed location of a photomask, as a function of photon energy, collimation and monochromator parameters. The model is validated using a combination of experimental measurements of the photon intensity distribution of the undulator harmonics. It is shown that the undulator harmonics intensity ratio can be accurately measured using an imaging detector and controlled using beamline optics. Finally, the photomask geometric constraints and achievable performance for the limiting case of fully spatially coherent illumination are evaluated.

Submarine landslides and tsunami genesis in Sagami Bay, Japan, caused by the 1923 Great Kanto earthquake

AbstractThe 1923 Great Kanto earthquake occurred on September 1, in Japan, and caused severe damage mainly in the Kanto region. Tsunamis were observed over wide regions from the east coast of the Izu Peninsula to the west coast of the Boso Peninsula, and particularly, the damage in Atami was devastating. Many earthquake fault models including those of Kanamori (1971) and Ando (1971) were proposed based on the records of land deformation. However, such fault models cannot sufficiently explain the tsunami elevation and its initial sea-level motion on the coasts. Hence, the detailed mechanisms remain elusive. This study examines the possibility that a leading mechanism of the tsunami in the 1923 Great Kanto earthquake was a large-scale submarine landslide that occurred at Sagami Bay and at the mouth of Tokyo Bay based on the records of depth data measured by the Imperial Japanese Navy (1924) before and after the earthquake. We first show that the tsunami calculated by each fault model was inconsistent with the waveform at Yokosuka, the coastal tsunami elevations, and initial sea-level motion. Then, based on statistical analysis of the depth changes at the 1923 Great Kanto earthquake, we found that the seafloor bathymetric changes represented large-scale submarine landslides that may correspond to long-runout submarine liquefied sediment flows. The seafloor gradient over a 40 km flow-out distance was equal to or less than 0.4$$^\circ$$ ∘ . Through the identification of the submarine landslide source by tsunami backpropagation analysis and utilizing an analytical solution of a high-density gravity flow and a sensitivity analysis, we conducted a range of numerical simulations of the 1923 Great Kanto earthquake tsunamis using a fault model and a submarine landslide tsunami source model due to a high-density liquefied gravity flow. The results quantitatively accounted for the discrepancy between the observed tsunami records with maximum tsunami elevations over 12 m and the fault-model–based simulations with maximum tsunami elevations of 2 to 5 m and explained consistently the maximum tsunami elevation distributions as well as the time-series tsunami waveforms. These results may thus facilitate and deepen our understanding of the earthquake-induced submarine landslide tsunami risk as cascading multi-geohazards.

Voltage sag source location based on power polarity and twin characteristics matching

AbstractPrecisely locating the voltage sag sources is significant for improving power quality and clarifying the responsibilities of both power supply and consumption. However, the existing location models are complex, computationally intensive, and influenced by the measurement error and fault type, which have low positioning efficiency and large errors. Therefore, this paper proposes a method for locating voltage sag sources based on power polarity and twin characteristics matching. The method firstly uses the polarity of the sequence sag component disturbance reactive power at the monitoring point to determine the sag line. Secondly, the basic realization method of the twin power system is discussed, and the twin characteristics are defined by combining the positive and negative sequence sag component disturbance reactive power and the twin database is established. Thirdly, the spectral norm and Pearson product‐moment correlation coefficient are used to define the data matching degree, and the precise location model of the sag source is established with the objective function of maximizing the data matching between the twin characteristics of the preset and actual sag. Finally, the Improved Whale Optimization Algorithm is used to solve the location model. The correctness and superiority of the proposed method is verified based on IEEE33 node modelling.