Space Domain Research Articles

Multi-feature fusion based face forgery detection with local and global characteristics.

The malicious use of deepfake videos seriously affects information security and brings great harm to society. Currently, deepfake videos are mainly generated based on deep learning methods, which are difficult to be recognized by the naked eye, therefore, it is of great significance to study accurate and efficient deepfake video detection techniques. Most of the existing detection methods focus on analyzing the discriminative information in a specific feature domain for classification from a local or global perspective. Such detection methods based on a single type feature have certain limitations in practical applications. In this paper, we propose a deepfake detection method with the ability to comprehensively analyze the forgery face features, which integrates features in the space domain, noise domain, and frequency domain, and uses the Inception Transformer to learn the mix of global and local information dynamically. We evaluate the proposed method on the DFDC, Celeb-DF, and FaceForensic++ benchmark datasets. Extensive experiments verify the effectiveness and good generalization of the proposed method. Compared with the optimal model, the proposed method with a small number of parameters does not use pre-training, distillation, or assembly, but still achieves competitive performance. The ablation experiments evaluate the role of each component.

The light-dependent pseudo-capacitive charging of conjugated polymer nanoparticles coupled with the depolarization of the neuronal membrane.

The mechanism underlying visual restoration in blind animal models of retinitis pigmentosa using a liquid retina prosthesis based on semiconductive polymeric nanoparticles is still being debated. Through the application of mathematical models and specific experiments, we developed a coherent understanding of abiotic/biotic coupling, capturing the essential mechanism of photostimulation responsible for nanoparticle-induced retina activation. Our modeling is based on the solution of drift-diffusion and Poisson-Nernst-Planck models in the multi-physics neuron-cleft-nanoparticle-extracellular space domain, accounting for the electro-chemical motion of all the relevant species following photoexcitation. Modeling was coupled with electron microscopy to estimate the size of the neuron-nanoparticle cleft and electrophysiology on retina explants acutely or chronically injected with nanoparticles. Overall, we present a consistent picture of electrostatic depolarization of the bipolar cell driven by the pseudo-capacitive charging of the nanoparticle. We demonstrate that the highly resistive cleft composition, due to filling by adhesion/extracellular matrix proteins, is a crucial ingredient for establishing functional electrostatic coupling. Additionally, we show that the photo-chemical generation of reactive oxygen species (ROS) becomes relevant only at very high light intensities, far exceeding the physiological ones, in agreement with the lack of phototoxicity shown in vivo.

Translational symmetry in convolutions with localized kernels causes an implicit bias toward high frequency adversarial examples.

Adversarial attacks are still a significant challenge for neural networks. Recent efforts have shown that adversarial perturbations typically contain high-frequency features, but the root cause of this phenomenon remains unknown. Inspired by theoretical work on linear convolutional models, we hypothesize that translational symmetry in convolutional operations together with localized kernels implicitly bias the learning of high-frequency features, and that this is one of the main causes of high frequency adversarial examples. To test this hypothesis, we analyzed the impact of different choices of linear and non-linear architectures on the implicit bias of the learned features and adversarial perturbations, in spatial and frequency domains. We find that, independently of the training dataset, convolutional operations have higher frequency adversarial attacks compared to other architectural parameterizations, and that this phenomenon is exacerbated with stronger locality of the kernel (kernel size) end depth of the model. The explanation for the kernel size dependence involves the Fourier Uncertainty Principle: a spatially-limited filter (local kernel in the space domain) cannot also be frequency-limited (local in the frequency domain). Using larger convolution kernel sizes or avoiding convolutions (e.g., by using Vision Transformers or MLP-style architectures) significantly reduces this high-frequency bias. Looking forward, our work strongly suggests that understanding and controlling the implicit bias of architectures will be essential for achieving adversarial robustness.

Application of high-order SF-SFDTD scheme to solving a time-dependent SchrFDTD scheme , Wu

The finite-difference time-domain 2, 2 (FDTD (2, 2)) method with second-order numerical accuracy in time and space has been extensively employed in the field of quantum mechanics to solve the Schrödinger equation. Nevertheless, the presence of the Courant-Friedrichs-Lewy (CFL) condition imposes limitations on the grid size in the computational space, thereby constraining the admissible range of time steps. Accordingly, the efficiency of the FDTD(2, 2) method significantly decreases. In addition, the second-order numerical accuracy of the FDTD(2, 2) method both in time domain and in space domain often results in significant error accumulation during calculations, thereby undermining the fidelity of the simulation results. To surmount the constraints imposed by the CFL stability conditions and enhance the accuracy of computations, a novel approach termed SF-SFDTD(3, 4) method has been proposed, with 3 and 4 referring to the accuracy in space and time, respectively. This method combines spatial filtering (SF) with the high-order symplectic finite-difference time-domain (SFDTD) method. Its primary objective is to solve the time-dependent Schrödinger equation while ensuring time stability and scalability. The SF-SFDTD(3, 4) method obviates the need for further deriving the iterative formula employed in the conventional SFDTD(3, 4) method. Therefore, the method under consideration exhibits a remarkable degree of compatibility with its traditional counterpart. It is merely necessary to include a spatial filtering operation during each numerical iteration to eliminate spatial high-frequency components arising from the utilization of time step sizes that fail to satisfy the CFL stability condition, thereby ensuring the stability of the numerical scheme. Moreover, when the time step value satisfies the CFL stability condition, the amplitude of the high-frequency component approaches zero, thereby exerting a minimal influence on the accuracy of the computational results. The adoption of time steps that do not meet the CFL stability conditions leads to an amplification in the amplitude of the high-frequency component. However, this finding solely affects the stability of the computational results, and the elimination of these unstable high-frequency components scarcely affect the accuracy of the computational results. The SF-SFDTD(3, 4) retains the simplicity and efficacy inherent in the traditional SFDTD(3, 4) methods, while enhancing computational efficiency. Additionally, the numerical stability and dispersion error of the SF-SFDTD(3, 4) method are analyzed theoretically . Finally, the validity and efficacy of the proposed method are corroborated through numerical illustrations.

Deep Mixed Domain Generalization Network for Intelligent Fault Diagnosis Under Unseen Conditions

Emerging intelligent fault diagnosis models based on domain adaptation can resolve domain shift problems produced by different working conditions. However, the prerequisite of obtaining target data in advance limits the application of these models to practical engineering scenarios. To address this challenge, a deep mixed domain generalization network (DMDGN) is proposed for intelligent fault diagnosis. In this novel model, data augmentation is applied to both class and domain spaces, adversarial learning is employed to introduce adversarial perturbations, and a domain-based discrepancy metric is used to balance intra-domain and inter-domain distances. The model can effectively learn more domain-invariant and discriminative features from multiple source domains to perform different generalization tasks for different working loads and machines. The feasibility of the DMDGN model is verified on two public datasets and one private dataset collected from practical production processes. Empirical results show that the DMDGN model outperforms several state-of-the-art models.

The new psychopathological paradigm of schizophrenia and schizophrenia spectrum disorders

The basis of the new paradigm, as an attempt to modernize the systematics of psychopathological disorders, is the concept of simultaneous representation of two relatively independent domains (founded by neurobiological processes) in the clinical space of schizophrenia - negative and positive disorders. The study established the spectra of functional activity of endogenous dimensions, as well as the trajectories of their development, which determine the dominance of predominantly negative or positive symptoms (negative/positive schizophrenia) during the course of the disease. The differentiated impact of endogenous domains on constitutional characterological structures accompanied by the formation of pseudopsychiopathies and subpsychotic personality disorders is observed.

Sensitivity studies of PWR MOX fuel management to the plutonium initial vector using Artificial Neural Networks

This paper presents new metamodels based on artificial neural networks trained on full core 3D depletion simulations performed with APOLLO2 and CRONOS2. They are used to estimate the irradiation cycle length, discharge burn-up of each fuel assembly type and radial power factor of a PWR loaded with 30% of MOX fuels, as a function of the initial plutonium composition. They allow to explore the impact of the plutonium isotopic vector on the reactor characteristics and can be used for scenarios studied for future fuel cycle. Some exclusion domains in the plutonium isotopic vector phase space are identified as a function of the cycle length. As an example, the potentialities of such fuel management for plutonium recycling from MOX spent fuel are studied.

Fourier-Based Frequency Space Disentanglement and Augmentation for Generalizable Face Anti-Spoofing.

Generalizing face anti-spoofing (FAS) models to unseen distributions is challenging due to domain shifts. Previous domain generalization (DG) based FAS methods focus on learning invariant features across domains in the spatial space, which may be ineffective in detecting subtle spoof patterns. In this paper, we propose a novel approach called Frequency Space Disentanglement and Augmentation (FSDA) for generalizable FAS. Specifically, we leverage Fourier transformation to analyze face images in the frequency space, where the amplitude spectrum captures low-level texture information that forms distinct visual appearances, and the phase spectrum corresponds to the content information. We hypothesize that the liveness of a face is more related to these low-level patterns rather than high-level content information. To locate spoof traces, we disentangle the amplitude spectrum into domain-related and spoof-related components using either empirical or learnable strategies. We then propose a frequency space augmentation technique that mixes the disentangled components of two images to synthesize new variations. By imposing a distillation loss and a consistency loss on the augmented samples, our model learns to capture spoof patterns that are robust to both domain and spoof type variations. Extensive experiments on four FAS datasets demonstrate the superiority of our method in improving the generalization ability of FAS models in various unseen scenarios.

Pointwise superconvergence of block finite elements for the three-dimensional variable coefficient elliptic equation

<p>This study investigated the point-wise superconvergence of block finite elements for the variable coefficient elliptic equation in a regular family of rectangular partitions of the domain in three-dimensional space. Initially, the estimates for the three-dimensional discrete Greens function and discrete derivative Greens function were presented. Subsequently, employing an interpolation operator of projection type, two essential weak estimates were derived, which were crucial for superconvergence analysis. Ultimately, by combining the aforementioned estimates, we achieved superconvergence estimates for the derivatives and function values of the finite element approximation in the point-wise sense of the $ L^\infty $-norm. A numerical example illustrated the theoretical results.</p>

StAC-DA: Structure aware cross-modality domain adaptation framework with image and feature-level adaptation for medical image segmentation.

Convolutional neural networks (CNNs) have achieved state-of-the-art results in various medical image segmentation tasks. However, CNNs often assume that the source and target dataset follow the same probability distribution and when this assumption is not satisfied their performance degrades significantly. This poses a limitation in medical image analysis, where including information from different imaging modalities can bring large clinical benefits. In this work, we present an unsupervised Structure Aware Cross-modality Domain Adaptation (StAC-DA) framework for medical image segmentation. StAC-DA implements an image- and feature-level adaptation in a sequential two-step approach. The first step performs an image-level alignment, where images from the source domain are translated to the target domain in pixel space by implementing a CycleGAN-based model. The latter model includes a structure-aware network that preserves the shape of the anatomical structure during translation. The second step consists of a feature-level alignment. A U-Net network with deep supervision is trained with the transformed source domain images and target domain images in an adversarial manner to produce probable segmentations for the target domain. The framework is evaluated on bidirectional cardiac substructure segmentation. StAC-DA outperforms leading unsupervised domain adaptation approaches, being ranked first in the segmentation of the ascending aorta when adapting from Magnetic Resonance Imaging (MRI) to Computed Tomography (CT) domain and from CT to MRI domain. The presented framework overcomes the limitations posed by differing distributions in training and testing datasets. Moreover, the experimental results highlight its potential to improve the accuracy of medical image segmentation across diverse imaging modalities.

The properties of generalized John domains in metric spaces

<abstract><p>In this paper, we studied the properties of generalized John domains in metric space. We prove that a domain $ D $ is a $ \varphi $-John domain if, and only if, $ D\backslash P $ is a $ \varphi' $-John domain, where $ P $ is a subset of $ D $ containing finitely many points of $ D $. Meanwhile, we also showed that the union of $ \varphi $-John domains is a $ \varphi'' $-John domain in metric space.</p></abstract>

Multi-Source Data Integration for Segmentation of Unannotated MRI Images.

Automatic semantic segmentation of magnetic resonance imaging (MRI) images using deep neural networks greatly assists in evaluating and planning treatments for various clinical applications. However, training these models is conditioned on the availability of abundant annotated data. Even if we annotate enough data, MRI images display considerable variability due to factors such as differences among patients, MRI scanners, and imaging protocols. This variability necessitates retraining neural networks for each specific application domain, which, in turn requires manual annotation by expert radiologists for all new domains. To relax the need for persistent data annotation, we develop a method for unsupervised federated domain adaptation using multiple annotated source domains. Our approach enables the transfer of knowledge from several annotated source domains for use in an unannotated target domain. Initially, we ensure that the target domain data shares similar representations with each source domain in a latent embedding space by minimizing the pair-wise distances between the distributions for the target and the source domains. We then employ an ensemble approach to leverage the knowledge obtained from all domains to build an integrated outcome. We perform experiments on two datasets to demonstrate our method is effective. Our implementation code is publicly available: https://github.com/navapatn/Unsupervised -Federated-Domain-Adaptation-for-Image-Segmentation new.

The N-terminal domain of Type IV-A1 CRISPR-associated DinG is vulnerable to proteolysis.

CasDinG is an ATP-dependent 5'-3' DNA helicase essential for bacterial Type IV-A1 CRISPR associated immunity. CasDinG contains an essential N-terminal domain predicted to bind DNA. To better understand the role of the N-terminal domain, we attempted to co-crystallize CasDinG with DNA substrates. We successfully crystallized CasDinG in a tightly packed, crystal conformation with previously unobserved unit cell dimensions. However, the structure lacked electron density for a bound DNA substrate and the CasDinG N-terminal domain. Additionally, the tight crystal packing disallowed space for the N-terminal domain, indicating that the N-terminal domain was proteolyzed before crystallization. Follow up experiments revealed that the N-terminal domain of CasDinG is proteolyzed after a few days at room temperature, but is protected from proteolysis at 4°C. These data provide a distinct x-ray crystal structure of CasDinG and indicate the essential N-terminal domain of CasDinG is prone to proteolysis.

Exploring Video Denoising in Thermal Infrared Imaging: Physics-Inspired Noise Generator, Dataset, and Model.

We endeavor on a rarely explored task named thermal infrared video denoising. Perception in the thermal infrared significantly enhances the capabilities of machine vision. Nonetheless, noise in imaging systems is one of the factors that hampers the large-scale application of equipment. Existing thermal infrared denoising methods, primarily focusing on the image level, inadequately utilize time-domain information and insufficiently conduct investigation of system-level mixed noise, presenting the inferior ability in the video-recorded era; while video denoising methods, commonly applied to RGB cameras, exhibit uncertain effectiveness owing to substantial dissimilarities in the noise models and modalities between RGB and thermal infrared images. In sight of this, we initially revisit the imaging mechanism, while concurrently introducing a physics-inspired noise generator based on the sources and characteristics of system noise. Subsequently, a thermal infrared video denoising dataset consisting of 518 real-world videos is constructed. Lastly, we propose a denoising model called multi-domain infrared video denoising network, capable of concentrating features from the time, space, and frequency domains to restore high-fidelity videos. Extensive experiments demonstrate that the proposed method achieves state-of-the-art denoising quality and can be successfully applied to commercial cameras and downstream vision tasks, providing a new avenue for clear videography in the thermal infrared world. The dataset and code will be available.

The Language of Infographics: Toward Understanding Conceptual Metaphor Use in Scientific Storytelling.

We apply an approach from cognitive linguistics by mapping Conceptual Metaphor Theory (CMT) to the visualization domain to address patterns of visual conceptual metaphors that are often used in science infographics. Metaphors play an essential part in visual communication and are frequently employed to explain complex concepts. However, their use is often based on intuition, rather than following a formal process. At present, we lack tools and language for understanding and describing metaphor use in visualization to the extent where taxonomy and grammar could guide the creation of visual components, e.g., infographics. Our classification of the visual conceptual mappings within scientific representations is based on the breakdown of visual components in existing scientific infographics. We demonstrate the development of this mapping through a detailed analysis of data collected from four domains (biomedicine, climate, space, and anthropology) that represent a diverse range of visual conceptual metaphors used in the visual communication of science. This work allows us to identify patterns of visual conceptual metaphor use within the domains, resolve ambiguities about why specific conceptual metaphors are used, and develop a better overall understanding of visual metaphor use in scientific infographics. Our analysis shows that ontological and orientational conceptual metaphors are the most widely applied to translate complex scientific concepts. To support our findings we developed a visual exploratory tool based on the collected database that places the individual infographics on a spatio-temporal scale and illustrates the breakdown of visual conceptual metaphors.

AI/ML for safety-critical software: the case of the space domain

AI/ML for safety-critical software: the case of the space domain

Video matting tampering detection based on time and space domain traces

Video matting tampering detection based on time and space domain traces

Near‑Field Status Analysis Method for High‑Power Laser Facility Based on Space Domain Evaluation and Distribution Statistics

Near‑Field Status Analysis Method for High‑Power Laser Facility Based on Space Domain Evaluation and Distribution Statistics

Application and Development of 3D Printing Technology in Clothing Design

Abstract As traditional hand knitting is limited, it can only make simple three-dimensional stacking and flat knitting texture of the garment. This paper applies 3D printing technology to improve this drawback. Establishing research variables for clothing design by combining the geometric characteristics of the STL model after layering with the characteristics of 3D printing technology to construct an optimization model for 3D printing technology. The contradiction between molding quality and molding efficiency is a typical multi-factor and multi-objective optimization problem. To solve this kind of optimization problem, a genetic algorithm is used to search for the optimal solution of the objective function in the feasible domain space, aiming to optimize the parameters of 3D printing technology. An in-depth analysis of the application practice of 3D printing technology in apparel design is presented, which further points out the implementation architecture of apparel design incorporating 3D printing technology, and the performance test of apparel incorporating 3D printing technology is carried out. The data show that the average values of ST (thickness difference), B1 (warp bending stiffness), B2 (weft bending stiffness), E51 (warp elongation), and E52 (weft elongation) are 8.386mm, 5.22 μNm, 0.16 μNm, 0.421%, 0.252%, which indicates that the garments designed based on 3D printing technology have good tensile properties and hand feel appearance, and are more in line with the comfort and aesthetic demands of consumers for garment design. It shows that the garments designed using 3D printing technology have good stretchability and a hand-feel appearance, which is more in line with the comfort and aesthetic demands of consumers nowadays.

The world of “worlds” in the dimensions of semiosis With the example of the linguacultural space of Kyiv

The relevance of the article is due to the constant dynamics of the lexicon as a manifestation of the essential property of the language and, in particular, of the phenomenon of the development of semantics of the name world (світ), which became one of the most frequently used notional units of the Ukrainian language in the 21st century. The purpose of the study is to identify the peculiar features of the semantization process of the lexeme world (світ) using the example of the linguacultural space of Kyiv taking into account various dimensions of semiosis. The study is also aimed at establishing the factors that influenced the nominative activity of the unit and determining its linguo-pragmatic vector. The subject matter of the study is the nominative-referential and pragmatic potential of the name world (світ), recorded in mass media texts and Kyiv ergonymicon. To achieve the goal, the following methods and techniques were used: component analysis of the word, distributive analysis, analysis of dictionary definitions, the descriptive method, results of associative experiments, the constructive method, contextual analysis, content analysis, and techniques of quantitative analysis. The novelty of the study is determined by the complex multifaceted approach to elucidating the semantization process of the name world (світ), carried out on the basis of semiotics and linguacultural studies. As a result of the study of the semantics of numerous common and proper nominals with the world component, it was concluded that the reference of these nominations was affected by the actualization of the nuclear semes ‘limit’, ‘sphere’, ‘totality’, ‘completeness of volume’ (‘межа’, ‘сфера’, ‘сукупність’, ‘повнота обсягу’), and in addition, by the potential evaluative seme. Emphasis is placed on the general tendency towards the nomination of various semiotic spheres by units with spatial semantics. One of the main factors of the nominative activity of the name has been determined as the sound-symbolic and iconic nature of the sign, which ensures its euphony and aesthetic appeal, which, in its turn, is especially significant for ‘commercial’ names and advertising texts. Attention is paid to defining the specifics of comprehension and interpretation of knowledge and ideas about the world and their arrangement in mental space. On the comparison basis of the concept WORLD (СВІТ) as a Ukrainian linguacultural dominant with the universal domain SPACE (ПРОСТІР), the axiological cognitive features of ‘comprehensiveness’, ‘infinity’, ‘matter’, ‘cognition’ (‘всеосяжність’, ‘безмежність’, ‘матерія’, ‘пізнання’), which determine the linguo-pragmatic vector of nominations, have been allocated. Prospects for further research are seen in the study of the functional and pragmatic properties of the name world (світ) in different speech registers.