arXiv subjects

Wei Yu

Publications and source records attributed to Wei Yu.

At least 55 records · Page 3Linked to original sources

arXiv2025

Multimodal Visual Image Based User Association and Beamforming Using Graph Neural Networks

This paper proposes an approach that leverages multimodal data by integrating visual images with radio frequency (RF) pilots to optimize user association and beamforming in a downlink wireless cellular network under a max-min fairness criterion. Traditional methods typically optimize wireless system parameters based on channel state information (CSI). However, obtaining accurate CSI requires extensive pilot transmissions, which lead to increased overhead and latency. Moreover, the optimization of user association and beamforming is a discrete and non-convex optimization problem, which is challenging to solve analytically. In this paper, we propose to incorporate visual camera data in addition to the RF pilots to perform the joint optimization of user association and beamforming. The visual image data helps enhance channel awareness, thereby reducing the dependency on extensive pilot transmissions for system optimization. We employ a learning-based approach based on using first a detection neural network that estimates user locations from images, and subsequently two graph neural networks (GNNs) that extract features for system optimization based on the location information and the received pilots, respectively. Then, a multimodal GNN is constructed to integrate the features for the joint optimization user association and beamforming. Simulation results demonstrate that the proposed method achieves superior performance, while having low computational complexity and being interpretable and generalizable, making it an effective solution as compared to traditional methods based only on RF pilots.

eess.SP↗

arXiv2025

Adaptive Dropout: Unleashing Dropout across Layers for Generalizable Image Super-Resolution

Blind Super-Resolution (blind SR) aims to enhance the model's generalization ability with unknown degradation, yet it still encounters severe overfitting issues. Some previous methods inspired by dropout, which enhances generalization by regularizing features, have shown promising results in blind SR. Nevertheless, these methods focus solely on regularizing features before the final layer and overlook the need for generalization in features at intermediate layers. Without explicit regularization of features at intermediate layers, the blind SR network struggles to obtain well-generalized feature representations. However, the key challenge is that directly applying dropout to intermediate layers leads to a significant performance drop, which we attribute to the inconsistency in training-testing and across layers it introduced. Therefore, we propose Adaptive Dropout, a new regularization method for blind SR models, which mitigates the inconsistency and facilitates application across intermediate layers of networks. Specifically, for training-testing inconsistency, we re-design the form of dropout and integrate the features before and after dropout adaptively. For inconsistency in generalization requirements across different layers, we innovatively design an adaptive training strategy to strengthen feature propagation by layer-wise annealing. Experimental results show that our method outperforms all past regularization methods on both synthetic and real-world benchmark datasets, also highly effective in other image restoration tasks. Code is available at \href{https://github.com/xuhang07/Adpative-Dropout}{https://github.com/xuhang07/Adpative-Dropout}.

cs.CV↗

arXiv2025

Non-Intrusive Load Monitoring Based on Image Load Signatures and Continual Learning

Non-Intrusive Load Monitoring (NILM) identifies the operating status and energy consumption of each electrical device in the circuit by analyzing the electrical signals at the bus, which is of great significance for smart power management. However, the complex and changeable load combinations and application environments lead to the challenges of poor feature robustness and insufficient model generalization of traditional NILM methods. To this end, this paper proposes a new non-intrusive load monitoring method that integrates "image load signature" and continual learning. This method converts multi-dimensional power signals such as current, voltage, and power factor into visual image load feature signatures, and combines deep convolutional neural networks to realize the identification and classification of multiple devices; at the same time, self-supervised pre-training is introduced to improve feature generalization, and continual online learning strategies are used to overcome model forgetting to adapt to the emergence of new loads. This paper conducts a large number of experiments on high-sampling rate load datasets, and compares a variety of existing methods and model variants. The results show that the proposed method has achieved significant improvements in recognition accuracy.

cs.LG↗

arXiv2025

RAW Image Reconstruction from RGB on Smartphones. NTIRE 2025 Challenge Report

Numerous low-level vision tasks operate in the RAW domain due to its linear properties, bit depth, and sensor designs. Despite this, RAW image datasets are scarce and more expensive to collect than the already large and public sRGB datasets. For this reason, many approaches try to generate realistic RAW images using sensor information and sRGB images. This paper covers the second challenge on RAW Reconstruction from sRGB (Reverse ISP). We aim to recover RAW sensor images from smartphones given the corresponding sRGB images without metadata and, by doing this, ``reverse" the ISP transformation. Over 150 participants joined this NTIRE 2025 challenge and submitted efficient models. The proposed methods and benchmark establish the state-of-the-art for generating realistic RAW data.

eess.IV↗

arXiv2025

Quadratic Transform for Fractional Programming in Signal Processing and Machine Learning

Fractional programming (FP) is a branch of mathematical optimization that deals with the optimization of ratios. It is an invaluable tool for signal processing and machine learning, because many key metrics in these fields are fractionally structured, e.g., the signal-to-interference-plus-noise ratio (SINR) in wireless communications, the Cramér-Rao bound (CRB) in radar sensing, the normalized cut in graph clustering, and the margin in support vector machine (SVM). This article provides a comprehensive review of both the theory and applications of a recently developed FP technique known as the quadratic transform, which can be applied to a wide variety of FP problems, including both the minimization and the maximization of the sum of functions of ratios as well as matrix-ratio problems.

cs.IT↗

arXiv2025

Pursuing Temporal-Consistent Video Virtual Try-On via Dynamic Pose Interaction

Video virtual try-on aims to seamlessly dress a subject in a video with a specific garment. The primary challenge involves preserving the visual authenticity of the garment while dynamically adapting to the pose and physique of the subject. While existing methods have predominantly focused on image-based virtual try-on, extending these techniques directly to videos often results in temporal inconsistencies. Most current video virtual try-on approaches alleviate this challenge by incorporating temporal modules, yet still overlook the critical spatiotemporal pose interactions between human and garment. Effective pose interactions in videos should not only consider spatial alignment between human and garment poses in each frame but also account for the temporal dynamics of human poses throughout the entire video. With such motivation, we propose a new framework, namely Dynamic Pose Interaction Diffusion Models (DPIDM), to leverage diffusion models to delve into dynamic pose interactions for video virtual try-on. Technically, DPIDM introduces a skeleton-based pose adapter to integrate synchronized human and garment poses into the denoising network. A hierarchical attention module is then exquisitely designed to model intra-frame human-garment pose interactions and long-term human pose dynamics across frames through pose-aware spatial and temporal attention mechanisms. Moreover, DPIDM capitalizes on a temporal regularized attention loss between consecutive frames to enhance temporal consistency. Extensive experiments conducted on VITON-HD, VVT and ViViD datasets demonstrate the superiority of our DPIDM against the baseline methods. Notably, DPIDM achieves VFID score of 0.506 on VVT dataset, leading to 60.5% improvement over the state-of-the-art GPD-VVTO approach.

cs.CV↗

arXiv2025

Coded Downlink Massive Random Access and a Finite de Finetti Theorem

This paper considers a massive connectivity setting in which a base-station (BS) aims to communicate sources $(X_1,\cdots,X_k)$ to a randomly activated subset of $k$ users, among a large pool of $n$ users, via a common message in the downlink. Although the identities of the $k$ active users are assumed to be known at the BS, each active user only knows whether itself is active and does not know the identities of the other active users. A naive coding strategy is to transmit the sources alongside the identities of the users for which the source information is intended. This requires $H(X_1,\cdots,X_k) + k\log(n)$ bits, because the cost of specifying the identity of one out of $n$ users is $\log(n)$ bits. For large $n$, this overhead can be significant. This paper shows that it is possible to develop coding techniques that eliminate the dependency of the overhead on $n$, if the source distribution follows certain symmetry. Specifically, if the source distribution is independently and identically distributed (i.i.d.) then the overhead can be reduced to at most $O(\log(k))$ bits, and in case of uniform i.i.d. sources, the overhead can be further reduced to $O(1)$ bits. For sources that follow a more general exchangeable distribution, the overhead is at most $O(k)$ bits, and in case of finite-alphabet exchangeable sources, the overhead can be further reduced to $O(\log(k))$ bits. The downlink massive random access problem is closely connected to the study of finite exchangeable sequences. The proposed coding strategy allows bounds on the Kullback-Leibler (KL) divergence between finite exchangeable distributions and i.i.d. mixture distributions to be developed and gives a new KL divergence version of the finite de Finetti theorem, which is scaling optimal.

cs.IT↗

arXiv2025

Rate-Distortion-Perception Tradeoff Based on the Conditional-Distribution Perception Measure

This paper studies the rate-distortion-perception (RDP) tradeoff for a memoryless source model in the asymptotic limit of large block-lengths. The perception measure is based on a divergence between the distributions of the source and reconstruction sequences \emph{conditioned} on the encoder output, first proposed by Mentzer et al. We consider the case when there is no shared randomness between the encoder and the decoder and derive a single-letter characterization of the RDP function for the case of discrete memoryless sources. This is in contrast to the marginal-distribution metric case (introduced by Blau and Michaeli), whose RDP characterization remains open when there is no shared randomness. The achievability scheme is based on lossy source coding with a posterior reference map. For the case of continuous valued sources under the squared error distortion measure and the squared quadratic Wasserstein perception measure, we also derive a single-letter characterization and show that the decoder can be restricted to a noise-adding mechanism. Interestingly, the RDP function characterized for the case of zero perception loss coincides with that of the marginal metric, and further zero perception loss can be achieved with a 3-dB penalty in minimum distortion. Finally we specialize to the case of Gaussian sources, and derive the RDP function for Gaussian vector case and propose a reverse water-filling type solution. We also partially characterize the RDP function for a mixture of Gaussian vector sources.

cs.IT↗

arXiv2025

PID-GM: PID Control with Gain Mapping

Proportional-Integral-Differential (PID) control is widely used in industrial control systems. However, up to now there are at least two open problems related with PID control. One is to have a comprehensive understanding of its robustness with respect to model uncertainties and disturbances. The other is to build intuitive, explicit and mathematically provable guidelines for PID gain tuning. In this paper, we introduce a simple nonlinear mapping to determine PID gains from three auxiliary parameters. By the mapping, PID control is shown to be equivalent to a new PD control (serving as a nominal control) plus an uncertainty and disturbance compensator (to recover the nominal performance). Then PID control can be understood, designed and tuned in a Two-Degree-of-Freedom (2-DoF) control framework. We discuss some basic properties of the mapping, including the existence, uniqueness and invertibility. Taking as an example the PID control applied to a general uncertain second-order plant, we prove by the singular perturbation theory that the closed-loop steady-state and transient performance depends explicitly on one auxiliary parameter which can be viewed as the virtual singular perturbation parameter (SPP) of PID control. All the three PID gains are monotonically decreasing functions of the SPP, indicating that the smaller the SPP is, the higher the PID gains are, and the better the robustness of PID control is. Simulation and experimental examples are provided to demonstrate the properties of the mapping as well as the effectiveness of the mapping based PID gain turning.

eess.SY↗

arXiv2025

NTIRE 2025 Challenge on Event-Based Image Deblurring: Methods and Results

This paper presents an overview of NTIRE 2025 the First Challenge on Event-Based Image Deblurring, detailing the proposed methodologies and corresponding results. The primary goal of the challenge is to design an event-based method that achieves high-quality image deblurring, with performance quantitatively assessed using Peak Signal-to-Noise Ratio (PSNR). Notably, there are no restrictions on computational complexity or model size. The task focuses on leveraging both events and images as inputs for single-image deblurring. A total of 199 participants registered, among whom 15 teams successfully submitted valid results, offering valuable insights into the current state of event-based image deblurring. We anticipate that this challenge will drive further advancements in event-based vision research.

cs.CV↗

arXiv2025

Learning Beamforming Codebooks for Active Sensing with Reconfigurable Intelligent Surface

This paper explores the design of beamforming codebooks for the base station (BS) and for the reconfigurable intelligent surfaces (RISs) in an active sensing scheme for uplink localization, in which the mobile user transmits a sequence of pilots to the BS through reflection at the RISs, and the BS and the RISs are adaptively configured by carefully choosing BS beamforming codeword and RIS codewords from their respective codebooks in a sequential manner to progressively focus onto the user. Most existing codebook designs for RIS are not tailored for active sensing, by which we mean the choice of the next codeword should depend on the measurements made so far, and the sequence of codewords should dynamically focus reflection toward the user. Moreover, most existing codeword selection methods rely on exhaustive search in beam training to identify the codeword with the highest signal-to-noise ratio (SNR), thus incurring substantial pilot overhead as the size of the codebook scales. This paper proposes a learning-based approach for codebook construction and for codeword selection for active sensing. The proposed learning approach aims to locate a target in the service area by recursively selecting a sequence of BS beamforming codewords and RIS codewords from the respective codebooks as more measurements become available without exhaustive beam training. The codebook design and the codeword selection fuse key ideas from the vector quantized variational autoencoder (VQ-VAE) and the long short-term memory (LSTM) network to learn respectively the discrete function space of the codebook and the temporal dependencies between measurements.

eess.SP↗

arXiv2025

Rate-Distortion-Perception Tradeoff for Gaussian Vector Sources

This paper studies the rate-distortion-perception (RDP) tradeoff for a Gaussian vector source coding problem where the goal is to compress the multi-component source subject to distortion and perception constraints. Specifically, the RDP setting with either the Kullback-Leibler (KL) divergence or Wasserstein-2 metric as the perception loss function is examined, and it is shown that for Gaussian vector sources, jointly Gaussian reconstructions are optimal. We further demonstrate that the optimal tradeoff can be expressed as an optimization problem, which can be explicitly solved. An interesting property of the optimal solution is as follows. Without the perception constraint, the traditional reverse water-filling solution for characterizing the rate-distortion (RD) tradeoff of a Gaussian vector source states that the optimal rate allocated to each component depends on a constant, called the water level. If the variance of a specific component is below the water level, it is assigned a zero compression rate. However, with active distortion and perception constraints, we show that the optimal rates allocated to the different components are always positive. Moreover, the water levels that determine the optimal rate allocation for different components are unequal. We further treat the special case of perceptually perfect reconstruction and study its RDP function in the high-distortion and low-distortion regimes to obtain insight to the structure of the optimal solution.

cs.IT↗

arXiv2025

Spectral analysis of the X-ray flares in the 2023 outburst of the new black binary transient Swift J1727.8--1613 observed with Insight-HXMT

The new black hole transient Swift J1727.8--1613 exhibited a series of X-ray flares during its 2023 outburst extensively observed with Insight-HXMT. We analyze the spectra of the flaring period using a series of models consisting of a multi-color disk and several different non-thermal components, and several consistent conclusions are obtained among these models. First, Swift J1727.8--1613 was in the transition process from the hard intermediate state (HIMS) to the very high state (VHS) during the first flaring period (MJD 60197--60204), and afterwards it exhibited typical VHS parameter characteristics, such as high temperature of the disk inner radius and a steep power-law spectrum with a photon index of 2.6. Second, the flares in the VHS are characterized by a rapid increase in the flux of accretion disk, accompanied by a simultaneous rapid expansion of the inner radius, which could be apparent if the accretion disk hardening factor varies significantly. The strong power-law component during the VHS is likely produced by synchrotron self-Compton process in the relativistic jets, in agreement with the observed weak reflection component and lack of correlation with the disk component.

astro-ph.HE↗

arXiv2025

First frequency phase transfer from the 3 mm to the 1 mm band on an Earth-sized baseline

Frequency Phase Transfer (FPT) is a technique designed to increase coherence and sensitivity in radio interferometry by making use of the non-dispersive nature of the troposphere to calibrate high-frequency data using solutions derived at a lower frequency. While the Korean VLBI Network has pioneered the use of simultaneous multi-band systems for routine FPT up to an observing frequency of 130 GHz, this technique remains largely untested in the (sub)millimeter regime. A recent effort has been made to outfit dual-band systems at (sub)millimeter observatories participating in the Event Horizon Telescope (EHT) and to test the feasibility and performance of FPT up to the observing frequencies of the EHT. We present the results of simultaneous dual-frequency observations conducted in January 2024 on an Earth-sized baseline between the IRAM 30-m in Spain and the JCMT and SMA in Hawai`i. We performed simultaneous observations at 86 and 215 GHz on the bright sources J0958+6533 and OJ287, with strong detections obtained at both frequencies. We observe a strong correlation between the interferometric phases at the two frequencies, matching the trend expected for atmospheric fluctuations and demonstrating for the first time the viability of FPT for VLBI at a wavelength of $\sim$1 millimeter. We show that the application of FPT systematically increases the 215 GHz coherence on all averaging timescales. In addition, the use of the co-located JCMT and SMA as a single dual-frequency station demonstrates the feasibility of paired-antenna FPT for VLBI for the first time, with implications for future array capabilities (e.g., ALMA sub-arraying and ngVLA calibration strategies).

astro-ph.IM↗

arXiv2025

On Self-Adaptive Perception Loss Function for Sequential Lossy Compression

We consider causal, low-latency, sequential lossy compression, with mean squared-error (MSE) as the distortion loss, and a perception loss function (PLF) to enhance the realism of reconstructions. As the main contribution, we propose and analyze a new PLF that considers the joint distribution between the current source frame and the previous reconstructions. We establish the theoretical rate-distortion-perception function for first-order Markov sources and analyze the Gaussian model in detail. From a qualitative perspective, the proposed metric can simultaneously avoid the error-permanence phenomenon and also better exploit the temporal correlation between high-quality reconstructions. The proposed metric is referred to as self-adaptive perception loss function (PLF-SA), as its behavior adapts to the quality of reconstructed frames. We provide a detailed comparison of the proposed perception loss function with previous approaches through both information theoretic analysis as well as experiments involving moving MNIST and UVG datasets.

cs.LG↗

arXiv2025

Correlated spectro-polarimetric study along the Z track in XTE J1701-462 puts constraints on its coronal geometry

Context. In September 2022, the transient neutron star low-mass X-ray binary XTE J1701-462 went into a new outburst. Aims. The objective of this work is to examine the evolution of the accretion geometry of XTE J1701-462 by studying the spectro-polarimetric properties along the Z track of this source. The simultaneous observations archived by the Insight-Hard X-ray Modulation Telescope (HXMT) and the Imaging X-ray Polarimetry Explorer (IXPE) give us the opportunity. Methods. We present a comprehensive X-ray spectro-polarimetric analysis of XTE J1701-462, using simultaneous observations from IXPE, Insight-HXMT and NuSTAR. For IXPE observations, two methods are employed to measure the polarization: a model-independent measurement with PCUBE and a model-dependent polarization-spectral analysis with XSPEC. The corresponding spectra from Insight-HXMT and NuSTAR are studied with two configurations that correspond to a slab-like corona and a spherical shell-like corona, respectively. Results. Significant polarization characteristics are detected in XTE J1701-462. The polarization degree shows a decreasing trend along the Z track, reducing from (4.84 $\pm$ 0.37)% to (3.76 $\pm$ 0.43)% on the horizontal branch and jumping to less than 1% on the normal branch. The simultaneous spectral analysis from Insight-HXMT and NuSTAR suggests that the evolution of the PD is closely linked to changes in the flux of the Comptonized component and its covering factor along the Z track, supporting a shrinking corona.

astro-ph.HE↗

arXiv2025

MTGA: Multi-View Temporal Granularity Aligned Aggregation for Event-Based Lip-Reading

Lip-reading is to utilize the visual information of the speaker's lip movements to recognize words and sentences. Existing event-based lip-reading solutions integrate different frame rate branches to learn spatio-temporal features of varying granularities. However, aggregating events into event frames inevitably leads to the loss of fine-grained temporal information within frames. To remedy this drawback, we propose a novel framework termed Multi-view Temporal Granularity aligned Aggregation (MTGA). Specifically, we first present a novel event representation method, namely time-segmented voxel graph list, where the most significant local voxels are temporally connected into a graph list. Then we design a spatio-temporal fusion module based on temporal granularity alignment, where the global spatial features extracted from event frames, together with the local relative spatial and temporal features contained in voxel graph list are effectively aligned and integrated. Finally, we design a temporal aggregation module that incorporates positional encoding, which enables the capture of local absolute spatial and global temporal information. Experiments demonstrate that our method outperforms both the event-based and video-based lip-reading counterparts.

cs.CV↗

arXiv2025

What Kind of Visual Tokens Do We Need? Training-free Visual Token Pruning for Multi-modal Large Language Models from the Perspective of Graph

Recent Multimodal Large Language Models(MLLMs) often use a large number of visual tokens to compensate their visual shortcoming, leading to excessive computation and obvious visual redundancy. In this paper, we investigate what kind of visual tokens are needed for MLLMs, and reveal that both foreground and background tokens are critical for MLLMs given the varying difficulties of examples. Based on this observation, we propose a graph-based method towards training-free visual token pruning, termed G-Prune.In particular, G-Prune regards visual tokens as nodes, and construct their connections based on their semantic similarities. Afterwards, the information flow is propagated via weighted links, and the most important tokens after iterations are kept for MLLMs, which can be front or background.To validate G-Prune, we apply it to a recent MLLM called LLaVA-NeXT, and conduct extensive experiments on a set of benchmarks.The experiment results show that G-Prune can greatly reduce computation overhead while retaining high performance on both coarse- and fine-grained tasks. For instance, G-Prune can reduce 63.57\% FLOPs of LLaVA-NeXT on VQA2.0 and TextVQA with only 0.95\% and 2.34\% accuracy drops, respectively.

cs.CV↗