Federated Learning (FL) is a promising paradigm for realizing edge intelligence. However, its practical deployment remains vulnerable to model perturbations, such as communication noise and quantization errors, which can significantly degrade model accuracy. Moreover, this degradation is further amplified in the presence of data heterogeneity. To address the above issues, we propose a novel FL framework, termed Sharpness-aware Minimization-based Robust Federated Learning (SMRFL), which provides intrinsic model robustness against perturbations by leveraging an intriguing geometrical property of the loss landscape: models around a flat minimum tend to exhibit similarly low loss values and thus higher robustness. Specifically, SMRFL encourages model convergence toward flat minima by solving a min-max optimization problem that minimizes the worst-case loss within a neighborhood of the model. To further mitigate the impact of data heterogeneity, we proposed Locally Aligned-SMRFL (LA-SMRFL), which incorporates a consensus constraint into the min-max optimization and solves it via inexact Alternating Direction Method of Multipliers (ADMM). We provide theoretical analysis for the convergence of SMRFL and LA-SMRFL in the non-convex FL setting, and derive robustness bounds via the certified radius. Extensive experiments on two real-world datasets under three perturbation scenarios demonstrate that SMRFL and LASMRFL substantially outperform baseline methods in terms of robustness.

Semantic communication is a novel communication paradigm focusing on the transmission of meaningful, task-oriented information. Recent results have shown that graphical structures represent the most robust and structurally faithful formalism for modeling the semantic knowledge within a wide range of source signals. However, previous solutions focus primarily on the pairwise relational graphs, which inherently lack the capacity to encapsulate complex, higher-order interactions fundamental to specific semantic contexts. In contrast, hypergraphs provide a more flexible mathematical framework that can more accurately map the multidimensional dependencies found in intricate data sources. In this paper, we investigate hypergraph-based semantic representation for the semantic communication system. We propose a novel hypergraph information bottleneck-based implicit semantic communication framework (HIB-SC), in which the semantic encoder is developed and optimized to extract the minimally sufficient representation of the hypergraph-based semantic information source, that maximizes the mutual information between the encoded representation and the implicit high-order semantic relations that are intended for the receiver. We theoretically prove that the proposed framework is able to extract the most informative subgraphs or motifs that are significantly more robust against adversarial attacks with improved generalization performance. Extensive experiments verify that the proposed HIB-SC achieves superior semantic compression efficiency, higher accuracy in implicit semantic inference, and enhanced resilience against noise, compared to the state-of-the-art solutions.

We introduce WearVQA, the first benchmark specifically designed to evaluate the visual questionanswering (VQA) capabilities of multi-modal AI assistant on wearable devices like smart glasses. Unlikeprior benchmarks that focus on high-quality, third-person imagery, WearVQA reflects the unique chal-lenges of ego-centric interaction—where visual inputs may be occluded, poorly lit, unzoomed, or blurry,and questions are grounded in realistic wearable use cases. The benchmark comprises 2,500 carefullycurated image-question-answer triplets, spanning 7 diverse image domains including both text-centricand general scenes, 10 cognitive task types ranging from basic recognition to various forms of reasoning,and 6 common wearables-specific image quality issues. All questions are designed to be answerable usingonly the visual input and common senses. WearVQA is paired with a rigorous LLM-as-a-judge evaluationframework with 96% labeling accuracy. Open-source and proprietary multi-modal LLMs achieved a QAaccuracy as low as 24–52% on WearVQA, with substantial drops on lower-quality images and reasoning-heavy tasks. These observations position WearVQA as a comprehensive and challenging benchmark forguiding technicial advancement towards robust, real-world multi-modal wearables AI systems.

In low-light environments, images often suffer from quality degradation issues such as low contrast, insufficient brightness, and color distortion due to inadequate light reaching the camera sensor. Most existing methods overlook the issue of color distortion caused by insufficient illumination when enhancing low-light images. In this work, we propose GWRetinex-Net, a novel deep learning network model based on Retinex theory and the gray world assumption. It consists of an image decomposition network, a reflection component enhancement network, and an illumination component enhancement network. During the training process of the image decomposition network, the gray world assumption is innovatively introduced to constrain the illposed decomposition problem caused by the absence of ground truth for reflection and illumination components, ensuring that the reflection components obtained after decomposition retain accurate color information. Based on the decomposition results, the reflection component enhancement network is responsible for mitigating degradation in the reflection components of low-light images, while the illumination component enhancement network focuses on adjusting the illumination distribution in the illumination components of low-light images. Comprehensive qualitative and quantitative experiments demonstrate that our GWRetinex-Net significantly outperforms comparative methods on multiple public datasets. Compared with the best-performing comparative methods, the images enhanced by the proposed method achieve an average improvement of 6.55% in SSIM and 8.87% in PSNR, along with an average decrease of 14.68% in MAE and 6.10% in NIQE. Additionally, object detection experiment in low-light environments further reveals the potential application value of GWRetinex-Net.

Semantic-aware communication is a novel paradigm that draws inspiration from human communication focusing on the delivery of the meaning of messages. It has attracted significant interest recently due to its potential to improve the efficiency and reliability of communication and enhance users’ quality-of-experience (QoE). Most existing works focus on transmitting and delivering the explicit semantic meaning that can be directly identified from the source signal. This paper investigates the implicit semantic-aware communication in which the hidden information, e.g., hidden relations, concepts and implicit reasoning mechanisms of users, that cannot be directly observed from the source signal must be recognized and interpreted by the intended users. To this end, a novel implicit semantic-aware communication (iSAC) architecture is proposed for representing, communicating, and interpreting the implicit semantic meaning between source and destination users. A graph-inspired structure is first developed to represent the complete semantics, including both explicit and implicit, of a message. A projection-based semantic encoder is then proposed to convert the high-dimensional graphical representation of explicit semantics into a low-dimensional semantic constellation space for efficient physical channel transmission. To enable the destination user to learn and imitate the implicit semantic reasoning process of source user, a generative adversarial imitation learning-based solution, called G-RML, is proposed. Different from existing communication solutions, the source user in G-RML does not focus only on sending as much of the useful messages as possible; but, instead, it tries to guide the destination user to learn a reasoning mechanism to map any observed explicit semantics to the corresponding implicit semantics that are most relevant to the semantic meaning. By applying G-RML, we prove that the destination user can accurately imitate the reasoning process of the source user and automatically generate a set of implicit reasoning paths following the same probability distribution as the expert paths. Compared to the existing solutions, our proposed G-RML requires much less communication and computational resources and scales well to the scenarios involving the communication of rich semantic meanings consisting of a large number of concepts and relations. Numerical results show that the proposed solution achieves up to 92% accuracy of implicit meaning interpretation.

The distributed detection fusion is investigated for conditionally dependent sensor networks with channel errors. When the joint probability density functions of the sensor observations are dependent and high dimensional, it is known to be a challenging problem. This paper deals with this problem under Monte Carlo framework. The Bayesian cost function is approximated by Monte Carlo importance sampling. Necessary conditions for optimal sensor rules and optimal fusion rule are derived in the sense of minimizing the approximated Bayesian cost function, respectively. A Gauss-Seidel/person-by-person optimization algorithm is developed to search the optimal sensor rules. It is proved that the discretized algorithm is finitely convergent. Since the error rate of Monte Carlo integration is regardless of dimensionality, the complexity of the new algorithm is much less than that of the previous algorithm based on Riemann sum approximation. The proposed method allows us to design the sensor networks with a higher dimensional joint probability density function of the sensor observations. The typical examples with dependent observations and channel errors are examined. The results of numerical examples demonstrate the effectiveness of the new algorithm.