In light of recent breakthroughs in text-to-image (T2I) generation, particularly with diffusion transformers (DiT), subject-driven technologies are increasingly being employed for high-fidelity customized production that preserves subject identity from reference inputs, enabling thrilling design workflows and engaging entertainment. Existing alternatives typically require either per-subject optimization via trainable text embeddings or training specialized encoders for subject feature extraction on large-scale datasets. Such dependencies on training procedures fundamentally constrain their practical applications. More importantly, current methodologies fail to fully leverage the inherent zero-shot potential of modern diffusion transformers (e.g., the Flux series) for authentic subject-driven synthesis. To bridge this gap, we propose FreeCus, a genuinely training-free framework that activates DiT's capabilities through three key innovations: 1) We introduce a pivotal attention sharing mechanism that captures the subject's layout integrity while preserving crucial editing flexibility. 2) Through a straightforward analysis of DiT's dynamic shifting, we propose an upgraded variant that significantly improves fine-grained feature extraction. 3) We further integrate advanced Multimodal Large Language Models (MLLMs) to enrich cross-modal semantic representations. Extensive experiments reflect that our method successfully unlocks DiT's zero-shot ability for consistent subject synthesis across diverse contexts, achieving state-of-the-art or comparable results compared to approaches that require additional training. Notably, our framework demonstrates seamless compatibility with existing inpainting pipelines and control modules, facilitating more compelling experiences. Our code is available at: https://github.com/Monalissaa/FreeCus.

Recent thrilling progress in large-scale text-to-image (T2I) models has unlocked unprecedented synthesis quality of AI-generated content (AIGC) including image generation, 3D and video composition. Further, personalized techniques enable appealing customized production of a novel concept given only several images as reference. However, an intriguing problem persists: Is it possible to capture multiple, novel concepts from one single reference image? In this paper, we identify that existing approaches fail to preserve visual consistency with the reference image and eliminate cross-influence from concepts. To alleviate this, we propose an attention calibration mechanism to improve the concept-level understanding of the T2I model. Specifically, we first introduce new learnable modifiers bound with classes to capture attributes of multiple concepts. Then, the classes are separated and strengthened following the activation of the cross-attention operation, ensuring comprehensive and self-contained concepts. Additionally, we suppress the attention activation of different classes to mitigate mutual influence among concepts. Together, our proposed method, dubbed DisenDiff, can learn disentangled multiple concepts from one single image and produce novel customized images with learned concepts. We demonstrate that our method outperforms the current state of the art in both qualitative and quantitative evaluations. More importantly, our proposed techniques are compatible with LoRA and inpainting pipelines, enabling more interactive experiences.

Training GANs under limited data often leads to discriminator overfitting and memorization issues, causing divergent training. Existing approaches mitigate the overfitting by employing data augmentations, model regularization, or attention mechanisms. However, they ignore the frequency bias of GANs and take poor consideration towards frequency information, especially high-frequency signals that contain rich details. To fully utilize the frequency information of limited data, this paper proposes FreGAN, which raises the model's frequency awareness and draws more attention to synthesising high-frequency signals, facilitating high-quality generation. In addition to exploiting both real and generated images' frequency information, we also involve the frequency signals of real images as a self-supervised constraint, which alleviates the GAN disequilibrium and encourages the generator to synthesis adequate rather than arbitrary frequency signals. Extensive results demonstrate the superiority and effectiveness of our FreGAN in ameliorating generation quality in the low-data regime (especially when training data is less than 100). Besides, FreGAN can be seamlessly applied to existing regularization and attention mechanism models to further boost the performance.

Virtual Coupling (VC) is a hot topic in railways. A virtually coupled train set operates as a single train and consists of multiple unit trains (UTs). Each UT keeps a desired gap distance with its predecessor by following a spacing policy. Constant time headway (CTH) spacing policy is widely adopted in the existing studies about VC. Arrival time difference (ATD), i.e., the difference between the time instants of two successive UTs arriving at a platform, is large when using CTH. It is not expected because the passengers loading of all UTs should operate at the same time. This paper firstly analyzes the causes of ATD mathematically and proves ATD is inevitable when adopting CTH. Then, the change tendency of ATD with respect to each parameter is analyzed. The above analysis indicates that CTH is not suitable for the arrival process. Thus, a quadratic spacing policy is designed to reduce ATD. Experimental results show that ATD can hardly be reduced by only optimizing the parameters of CTH, but the quadratic spacing policy greatly reduces it.

In this paper, we propose a novel distributed network decomposition algorithm with the aid of the factor graph model and the max-product algorithm, which aims to achieve minimum cut weight. Its effectiveness is testified for general graph partition as well as distributed inference in wireless networks. Our algorithm is fully distributed, simple in computation, and readily extensible, thus providing a potentially powerful, data-independent clustering scheme for a wide range of data processing and networking applications.

Existing works on distributed consensus explore linear iterations based on reversible Markov chains, which contribute to the slow convergence of the algorithms. It has been observed that by overcoming the diffusive behavior of reversible chains, certain nonreversible chains lifted from reversible ones mix substantially faster than the original chains. In this paper, the idea of Markov chain lifting is studied to accelerate the convergence of distributed consensus, and two general pseudoalgorithms are presented. These pseudoalgorithms are then instantiated through a class of location-aided distributed averaging (LADA) algorithms for wireless networks, where nodes' coarse location information is used to construct nonreversible chains that facilitate distributed computing and cooperative processing. Our first LADA algorithm is designed for grid networks; for ak×kgrid network, it achieves an ε-averaging time ofO(klog(ε-1)). Based on this algorithm, in a wireless network with transmission ranger, an ε-averaging time ofO(r-1log(ε-1)) can be attained through a centralized algorithm. Subsequently, a distributed LADA algorithm is presented, achieving the same scaling law in averaging time as the centralized scheme in wireless networks for allrsatisfying the connectivity requirement; the constructed chain also attains the optimal scaling law in terms of an important mixing metric, the fill time, in its class. Finally, a cluster-based LADA algorithm is proposed, which, requiring no central coordination, provides the additional benefit of reduced message complexity compared with the distributed LADA algorithm.

In this paper, a variational message passing framework is proposed for Markov random fields, which is computationally more efficient and admits wider applicability compared to the belief propagation algorithm. Based on this framework, structured variational methods are explored to take advantage of both the simplicity of variational approximation (for inter-cluster processing) and the accuracy of exact inference (for intra-cluster processing). Its performance is elaborated on a Gaussian Markov random field, through both theoretical analysis and simulation results.

In this paper, the asymptotic performance of a recently proposed distributed inference framework, structured variational methods, is investigated. We first distinguish the intra- and inter-cluster inference algorithms as vertex and edge processes respectively. Their difference is illustrated, and convergence rate is derived for the intra-cluster inference procedure which is based on an edge process. Then, viewed as a mixed vertex-edge process, the overall performance of structured variational methods is characterized via the coupling approach. Tradeoff between complexity and performance of this algorithm is also addressed, which provides insights for network design and analysis.

Recent research reveals that information security and information-hiding capabilities can be enhanced by proper exploitation of space-time techniques. Meanwhile, intrinsic properties of ultra wideband (UWB) signals make it an outstanding candidate for secure applications. In this paper, we propose a unitary space-time coding scheme for impulse radio UWB systems. Its transmission secrecy, including low probability of intercept (LPI), low probability of detection (LPD) and anti-jamming performance, is analyzed. Theoretical and simulation results show its superiority in wireless secure communications over other concurrent schemes

Energy efficiency is one of the most critical concerns for wireless sensor networks. While cooperative transmission strategies have the potential to significantly improve the system performance, they also incur additional energy cost and system overhead. In this paper, energy efficiency of relevant transmission strategies is studied both for wideband asymptotes and realistic system settings. Based on this analysis, general guidelines are presented for optimal transmission strategy selection in some typical scenarios, aiming at minimum energy consumption with a target BER. The proposed selection rules, especially those based on system-level metrics, are easy to implement for sensor applications. The framework provided here may also be readily extended to other scenarios or applications.