arXiv subjects

Xin Wang

Publications and source records attributed to Xin Wang.

At least 1,009 records · Page 56Linked to original sources

arXiv2024

X-Transfer: A Transfer Learning-Based Framework for GAN-Generated Fake Image Detection

Generative adversarial networks (GANs) have remarkably advanced in diverse domains, especially image generation and editing. However, the misuse of GANs for generating deceptive images, such as face replacement, raises significant security concerns, which have gained widespread attention. Therefore, it is urgent to develop effective detection methods to distinguish between real and fake images. Current research centers around the application of transfer learning. Nevertheless, it encounters challenges such as knowledge forgetting from the original dataset and inadequate performance when dealing with imbalanced data during training. To alleviate this issue, this paper introduces a novel GAN-generated image detection algorithm called X-Transfer, which enhances transfer learning by utilizing two neural networks that employ interleaved parallel gradient transmission. In addition, we combine AUC loss and cross-entropy loss to improve the model's performance. We carry out comprehensive experiments on multiple facial image datasets. The results show that our model outperforms the general transferring approach, and the best metric achieves 99.04%, which is increased by approximately 10%. Furthermore, we demonstrate excellent performance on non-face datasets, validating its generality and broader application prospects.

cs.CV↗

arXiv2024

Active Generation Network of Human Skeleton for Action Recognition

Data generation is a data augmentation technique for enhancing the generalization ability for skeleton-based human action recognition. Most existing data generation methods face challenges to ensure the temporal consistency of the dynamic information for action. In addition, the data generated by these methods lack diversity when only a few training samples are available. To solve those problems, We propose a novel active generative network (AGN), which can adaptively learn various action categories by motion style transfer to generate new actions when the data for a particular action is only a single sample or few samples. The AGN consists of an action generation network and an uncertainty metric network. The former, with ST-GCN as the Backbone, can implicitly learn the morphological features of the target action while preserving the category features of the source action. The latter guides generating actions. Specifically, an action recognition model generates prediction vectors for each action, which is then scored using an uncertainty metric. Finally, UMN provides the uncertainty sampling basis for the generated actions.

cs.CV↗

arXiv2024

CSST Large-scale Structure Analysis Pipeline: II. the CSST Emulator for Slitless Spectroscopy (CESS)

The Chinese Space Station Telescope (CSST) slitless spectroscopic survey will observe objects to a limiting magnitude of ~ 23 mag (5$σ$, point sources) in U, V, and I over 17500 deg$^2$. The spectroscopic observations are expected to be highly efficient and complete for mapping galaxies over 0 < z < 1 with secure redshift measurements at spectral resolutions of R ~ 200, providing unprecedented data sets for cosmological studies. To quantitatively examine the survey potential, we develop a software tool, namely the CSST Emulator for Slitless Spectroscopy (CESS), to quickly generate simulated 1D slitless spectra with limited computing resources. We introduce the architecture of CESS and the detailed process of creating simulated CSST slitless spectra. The extended light distribution of a galaxy induces the self-broadening effect on the 1D slitless spectrum. We quantify the effect using morphological parameters: Sérsic index, effective radius, position angle, and axis ratio. Moreover, we also develop a module for CESS to estimate the overlap contamination rate for CSST grating observations of galaxies in galaxy clusters. Applying CESS to the high-resolution model spectra of a sample of ~ 140 million galaxies with m_z < 21 mag selected from the Dark Energy Spectroscopic Instrument LS DR9 catalogue, we obtain the simulated CSST slitless spectra. We examine the dependence of measurement errors on different types of galaxies due to instrumental and observational effects and quantitatively investigate the redshift completeness for different environments out to z ~ 1. Our results show that the CSST spectroscopy is able to provide secure redshifts for about one-quarter of the sample galaxies.

astro-ph.GA↗

arXiv2024

2.75D: Boosting learning by representing 3D Medical imaging to 2D features for small data

In medical-data driven learning, 3D convolutional neural networks (CNNs) have started to show superior performance to 2D CNNs in numerous deep learning tasks, proving the added value of 3D spatial information in feature representation. However, the difficulty in collecting more training samples to converge, more computational resources and longer execution time make this approach less applied. Also, applying transfer learning on 3D CNN is challenging due to a lack of publicly available pre-trained 3D models. To tackle these issues, we proposed a novel 2D strategical representation of volumetric data, namely 2.75D. In this work, the spatial information of 3D images is captured in a single 2D view by a spiral-spinning technique. As a result, 2D CNN networks can also be used to learn volumetric information. Besides, we can fully leverage pre-trained 2D CNNs for downstream vision problems. We also explore a multi-view 2.75D strategy, 2.75D 3 channels (2.75Dx3), to boost the advantage of 2.75D. We evaluated the proposed methods on three public datasets with different modalities or organs (Lung CT, Breast MRI, and Prostate MRI), against their 2D, 2.5D, and 3D counterparts in classification tasks. Results show that the proposed methods significantly outperform other counterparts when all methods were trained from scratch on the lung dataset. Such performance gain is more pronounced with transfer learning or in the case of limited training data. Our methods also achieved comparable performance on other datasets. In addition, our methods achieved a substantial reduction in time consumption of training and inference compared with the 2.5D or 3D method.

eess.IV↗

arXiv2024

Efficient Data Reduction Strategies for Big Data and High-Dimensional LASSO Regressions

The IBOSS approach proposed by Wang et al. (2019) selects the most informative subset of n points. It assumes that the ordinary least squares method is used and requires that the number of variables, p, is not large. However, in many practical problems, p is very large and penalty-based model fitting methods such as LASSO is used. We study the big data problems, in which both n and p are large. In the first part, we focus on reduction in data points. We develop theoretical results showing that the IBOSS type of approach can be applicable to penalty-based regressions such as LASSO. In the second part, we consider the situations where p is extremely large. We propose a two-step approach that involves first reducing the number of variables and then reducing the number of data points. Two separate algorithms are developed, whose performances are studied through extensive simulation studies. Compared to existing methods including well-known split-and-conquer approach, the proposed methods enjoy advantages in terms of estimation accuracy, prediction accuracy, and computation time.

stat.ME↗

arXiv2024

Enhancing Video Super-Resolution via Implicit Resampling-based Alignment

In video super-resolution, it is common to use a frame-wise alignment to support the propagation of information over time. The role of alignment is well-studied for low-level enhancement in video, but existing works overlook a critical step -- resampling. We show through extensive experiments that for alignment to be effective, the resampling should preserve the reference frequency spectrum while minimizing spatial distortions. However, most existing works simply use a default choice of bilinear interpolation for resampling even though bilinear interpolation has a smoothing effect and hinders super-resolution. From these observations, we propose an implicit resampling-based alignment. The sampling positions are encoded by a sinusoidal positional encoding, while the value is estimated with a coordinate network and a window-based cross-attention. We show that bilinear interpolation inherently attenuates high-frequency information while an MLP-based coordinate network can approximate more frequencies. Experiments on synthetic and real-world datasets show that alignment with our proposed implicit resampling enhances the performance of state-of-the-art frameworks with minimal impact on both compute and parameters.

cs.CV↗

arXiv2024

Efficient Image Super-Resolution via Symmetric Visual Attention Network

An important development direction in the Single-Image Super-Resolution (SISR) algorithms is to improve the efficiency of the algorithms. Recently, efficient Super-Resolution (SR) research focuses on reducing model complexity and improving efficiency through improved deep small kernel convolution, leading to a small receptive field. The large receptive field obtained by large kernel convolution can significantly improve image quality, but the computational cost is too high. To improve the reconstruction details of efficient super-resolution reconstruction, we propose a Symmetric Visual Attention Network (SVAN) by applying large receptive fields. The SVAN decomposes a large kernel convolution into three different combinations of convolution operations and combines them with an attention mechanism to form a Symmetric Large Kernel Attention Block (SLKAB), which forms a symmetric attention block with a bottleneck structure by the size of the receptive field in the convolution combination to extract depth features effectively as the basic component of the SVAN. Our network gets a large receptive field while minimizing the number of parameters and improving the perceptual ability of the model. The experimental results show that the proposed SVAN can obtain high-quality super-resolution reconstruction results using only about 30% of the parameters of existing SOTA methods.

cs.CV↗

arXiv2024

To deform or not: treatment-aware longitudinal registration for breast DCE-MRI during neoadjuvant chemotherapy via unsupervised keypoints detection

Clinicians compare breast DCE-MRI after neoadjuvant chemotherapy (NAC) with pre-treatment scans to evaluate the response to NAC. Clinical evidence supports that accurate longitudinal deformable registration without deforming treated tumor regions is key to quantifying tumor changes. We propose a conditional pyramid registration network based on unsupervised keypoint detection and selective volume-preserving to quantify changes over time. In this approach, we extract the structural and the abnormal keypoints from DCE-MRI, apply the structural keypoints for the registration algorithm to restrict large deformation, and employ volume-preserving loss based on abnormal keypoints to keep the volume of the tumor unchanged after registration. We use a clinical dataset with 1630 MRI scans from 314 patients treated with NAC. The results demonstrate that our method registers with better performance and better volume preservation of the tumors. Furthermore, a local-global-combining biomarker based on the proposed method achieves high accuracy in pathological complete response (pCR) prediction, indicating that predictive information exists outside tumor regions. The biomarkers could potentially be used to avoid unnecessary surgeries for certain patients. It may be valuable for clinicians and/or computer systems to conduct follow-up tumor segmentation and response prediction on images registered by our method. Our code is available on \url{https://github.com/fiy2W/Treatment-aware-Longitudinal-Registration}.

eess.IV↗

arXiv2024

Interferometric Geometric Phases of $\mathcal{PT}$-symmetric Quantum Mechanics

We present a generalization of the geometric phase to pure and thermal states in $\mathcal{PT}$-symmetric quantum mechanics (PTQM) based on the approach of the interferometric geometric phase (IGP). The formalism first introduces the parallel-transport conditions of quantum states and reveals two geometric phases, $θ^1$ and $θ^2$, for pure states in PTQM according to the states under parallel-transport. Due to the non-Hermitian Hamiltonian in PTQM, $θ^1$ is complex and $θ^2$ is its real part. The imaginary part of $θ^1$ plays an important role when we generalize the IGP to thermal states in PTQM. The generalized IGP modifies the thermal distribution of a thermal state, thereby introducing effective temperatures. At certain critical points, the generalized IGP exhibits discrete jumps at finite temperatures, signaling a geometric phase transition. We demonstrate the finite-temperature geometric phase transition in PTQM by a two-level system and visualize its results.

quant-ph↗

arXiv2024

Ion temperature gradient mode mitigation by energetic particles, mediated by forced-driven zonal flows

In this work, we use the global electromagnetic and electrostatic gyro kinetic approaches to investigate the effects of zonal flows forced-driven by Alfvén modes due to their excitation by energetic particles (EPs), on the dynamics of ITG (Ion temperature gradient) instabilities. The equilibrium of the 92416 JET tokamak shot is considered. The linear and nonlinear Alfvén modes dynamics, as well as the zonal flow dynamics, are investigated and their respective radial structures and saturation levels are reported. ITG dynamics in the presence of the zonal flows excited by these Alfvén modes are also investigated. We find that, the zonal flows forced-driven by Alfvén modes can significantly impact the ITG dynamics. A zonal flow amplitude scan reveals the existence of an inverse relation between the zonal flow amplitude and the ITG growth rate. These results show that, forced-driven zonal flows can be an important indirect part of turbulence mitigation due to the injection of energetic particles.

physics.plasm-ph↗

arXiv2024

Spoofing attack augmentation: can differently-trained attack models improve generalisation?

A reliable deepfake detector or spoofing countermeasure (CM) should be robust in the face of unpredictable spoofing attacks. To encourage the learning of more generaliseable artefacts, rather than those specific only to known attacks, CMs are usually exposed to a broad variety of different attacks during training. Even so, the performance of deep-learning-based CM solutions are known to vary, sometimes substantially, when they are retrained with different initialisations, hyper-parameters or training data partitions. We show in this paper that the potency of spoofing attacks, also deep-learning-based, can similarly vary according to training conditions, sometimes resulting in substantial degradations to detection performance. Nevertheless, while a RawNet2 CM model is vulnerable when only modest adjustments are made to the attack algorithm, those based upon graph attention networks and self-supervised learning are reassuringly robust. The focus upon training data generated with different attack algorithms might not be sufficient on its own to ensure generaliability; some form of spoofing attack augmentation at the algorithm level can be complementary.

cs.CR↗

arXiv2024

Lyman Continuum Emission from AGN at 2.3$\lesssim$z$\lesssim$3.7 in the UVCANDELS Fields

We present the results of our search for Lyman continuum (LyC) emitting AGN at redshifts 2.3$\lesssim$z$\lesssim$4.9 from HST WFC3 F275W observations in the UVCANDELS fields. We also include LyC emission from AGN using HST WFC3 F225W, F275W, and F336W found in the ERS and HDUV data. We performed exhaustive queries of the Vizier database to locate AGN with high quality spectroscopic redshifts. In total, we found 51 AGN that met our criteria within the UVCANDELS and ERS footprints. Of these 51, we find 12 AGN had $\geq$4$σ$ detected LyC flux in the WFC3/UVIS images. Using space- and ground-based data from X-ray to radio, we fit the multi-wavelength photometric data of each AGN to a CIGALE SED and correlate various SED parameters to the LyC flux. KS-tests of the SED parameter distributions for the LyC-detected and non-detected AGN showed they are likely not distinct samples. However, we find that X-ray luminosity, star-formation onset age, and disk luminosity show strong correlations relative to their emitted LyC flux. We also find strong correlation of the LyC flux to several dust parameters, i.e., polar and toroidal dust emission, 6 $μm$ luminosity, and anti-correlation with metallicity and $A_{FUV}$. We simulate the LyC escape fraction ($f_{esc}$) using the CIGALE and IGM transmission models for the LyC-detected AGN and find an average $f_{esc}$$\simeq$18%, weighted by uncertainties. We stack the LyC flux of subsamples of AGN according to the wavelength continuum region in which they are detected and find no significant distinctions in their LyC emission, although our $sub-mm\ detected$ F336W sample shows the brightest stacked LyC flux. These findings indicate that LyC-production and -escape in AGN is more complicated than the simple assumption of thermal emission and a 100% escape fraction. Further testing of AGN models with larger samples than presented here is needed.

astro-ph.GA↗

arXiv2024

Quantifying the unextendibility of entanglement

Entanglement is a striking feature of quantum mechanics, and it has a key property called unextendibility. In this paper, we present a framework for quantifying and investigating the unextendibility of general bipartite quantum states. First, we define the unextendible entanglement, a family of entanglement measures based on the concept of a state-dependent set of free states. The intuition behind these measures is that the more entangled a bipartite state is, the less entangled each of its individual systems is with a third party. Second, we demonstrate that the unextendible entanglement is an entanglement monotone under two-extendible quantum operations, including local operations and one-way classical communication as a special case. Normalization and faithfulness are two other desirable properties of unextendible entanglement, which we establish here. We further show that the unextendible entanglement provides efficiently computable benchmarks for the rate of exact entanglement or secret key distillation, as well as the overhead of probabilistic entanglement or secret key distillation.

quant-ph↗

arXiv2024

High-Efficiency Resonant Beam Charging and Communication

With the development of Internet of Things (IoT), demands of power and data for IoT devices increase drastically. In order to resolve the supply-demand contradiction, simultaneous wireless information and power transfer (SWIPT) has been envisioned as an enabling technology by providing high-power energy transfer and high-rate data delivering concurrently. In this paper, we introduce a high-efficiency resonant beam (RB) charging and communication scheme. The scheme utilizes the semiconductor materials as gain medium, which has a better energy absorption capacity compared with the traditional solid-state one. Moreover, the telescope internal modulator (TIM) are adopted in the scheme which can concentrate beams to match the gain size, reducing the transmission loss. To evaluate the scheme SWIPT performance, we establish an analytical model and study the influence factors of its beam transmission, energy conversion, output power, and spectral efficiency. Numerical results shows that the proposed RB system can realize 16 W electric power output with 11 % end-to-end conversion efficiency, and support 18 bit/s/Hz spectral efficiency for communication.

cs.ET↗

arXiv2024

Collaborative Watermarking for Adversarial Speech Synthesis

Advances in neural speech synthesis have brought us technology that is not only close to human naturalness, but is also capable of instant voice cloning with little data, and is highly accessible with pre-trained models available. Naturally, the potential flood of generated content raises the need for synthetic speech detection and watermarking. Recently, considerable research effort in synthetic speech detection has been related to the Automatic Speaker Verification and Spoofing Countermeasure Challenge (ASVspoof), which focuses on passive countermeasures. This paper takes a complementary view to generated speech detection: a synthesis system should make an active effort to watermark the generated speech in a way that aids detection by another machine, but remains transparent to a human listener. We propose a collaborative training scheme for synthetic speech watermarking and show that a HiFi-GAN neural vocoder collaborating with the ASVspoof 2021 baseline countermeasure models consistently improves detection performance over conventional classifier training. Furthermore, we demonstrate how collaborative training can be paired with augmentation strategies for added robustness against noise and time-stretching. Finally, listening tests demonstrate that collaborative training has little adverse effect on perceptual quality of vocoded speech.

eess.AS↗

arXiv2024

Early Results from GLASS-JWST XXIII: The transmission of Lyman-alpha from UV-faint z ~ 3-6 galaxies

Lyman-alpha (Ly$α$) emission from galaxies can be used to trace neutral hydrogen in the epoch of reionization, however, there is a degeneracy between the attenuation of Ly$α$ in the intergalactic medium (IGM) and the line profile emitted from the galaxy. Large shifts of Ly$α$ redward of systemic due to scattering in the interstellar medium can boost Ly$α$ transmission in the IGM during reionization. The relationship between Ly$α$ velocity offset from systemic and other galaxy properties is not well-established at high-redshift or low luminosities, due to the difficulty of observing emission lines which trace systemic redshift. Rest-frame optical spectroscopy with JWST/NIRSpec has opened a new window into understanding of Ly$α$ at z>3. We present a sample of 12 UV-faint galaxies ($-20 \lesssim$ MUV $\lesssim -16$) at $3 \lesssim z \lesssim 6$, with Ly$α$ velocity offsets, $Δv_{\mathrm{Ly}α}$, measured from VLT/MUSE and JWST/NIRSpec from the GLASS-JWST Early Release Program. We find median $Δv_{\mathrm{Ly}α}$ of 205 km s$^{-1}$ and standard deviation 75 km s$^{-1}$, compared to 320 and 170km s$^{-1}$ for MUV < -20 galaxies in the literature. Our new sample demonstrates the previously observed trend of decreasing Ly$α$ velocity offset with decreasing UV luminosity and optical line velocity dispersion, extends to MUV $\gtrsim$ -20, consistent with a picture where the Ly$α$ profile is shaped by gas close to the systemic redshift. Our results imply that during reionization Ly$α$ from UV-faint galaxies will be preferentially attenuated, but that detecting Ly$α$ with low $Δv_{\mathrm{Ly}α}$ can be an indicator of large ionized bubbles.

astro-ph.GA↗

arXiv2023

Can large-scale vocoded spoofed data improve speech spoofing countermeasure with a self-supervised front end?

A speech spoofing countermeasure (CM) that discriminates between unseen spoofed and bona fide data requires diverse training data. While many datasets use spoofed data generated by speech synthesis systems, it was recently found that data vocoded by neural vocoders were also effective as the spoofed training data. Since many neural vocoders are fast in building and generation, this study used multiple neural vocoders and created more than 9,000 hours of vocoded data on the basis of the VoxCeleb2 corpus. This study investigates how this large-scale vocoded data can improve spoofing countermeasures that use data-hungry self-supervised learning (SSL) models. Experiments demonstrated that the overall CM performance on multiple test sets improved when using features extracted by an SSL model continually trained on the vocoded data. Further improvement was observed when using a new SSL distilled from the two SSLs before and after the continual training. The CM with the distilled SSL outperformed the previous best model on challenging unseen test sets, including the ASVspoof 2019 logical access, WaveFake, and In-the-Wild.

eess.AS↗

arXiv2023

Grounding-Prompter: Prompting LLM with Multimodal Information for Temporal Sentence Grounding in Long Videos

Temporal Sentence Grounding (TSG), which aims to localize moments from videos based on the given natural language queries, has attracted widespread attention. Existing works are mainly designed for short videos, failing to handle TSG in long videos, which poses two challenges: i) complicated contexts in long videos require temporal reasoning over longer moment sequences, and ii) multiple modalities including textual speech with rich information require special designs for content understanding in long videos. To tackle these challenges, in this work we propose a Grounding-Prompter method, which is capable of conducting TSG in long videos through prompting LLM with multimodal information. In detail, we first transform the TSG task and its multimodal inputs including speech and visual, into compressed task textualization. Furthermore, to enhance temporal reasoning under complicated contexts, a Boundary-Perceptive Prompting strategy is proposed, which contains three folds: i) we design a novel Multiscale Denoising Chain-of-Thought (CoT) to combine global and local semantics with noise filtering step by step, ii) we set up validity principles capable of constraining LLM to generate reasonable predictions following specific formats, and iii) we introduce one-shot In-Context-Learning (ICL) to boost reasoning through imitation, enhancing LLM in TSG task understanding. Experiments demonstrate the state-of-the-art performance of our Grounding-Prompter method, revealing the benefits of prompting LLM with multimodal information for TSG in long videos.

cs.CV↗