Papers

Multi-agent learning faces a fundamental tension: leveraging distributed collaboration without sacrificing the personalization needed for diverse agents. This tension intensifies when aiming for full personalization while adapting to unknown heterogeneity levels – gaining collaborative speedup when agents are similar, without performance degradation when they are different. Embracing the challenge, we propose personalized collaborative learning (PCL), a novel framework for heterogeneous agents to collaboratively learn personalized solutions with seamless adaptivity. Through carefully designed bias correction and importance correction mechanisms, our method AffPCL robustly handles both environment and objective heterogeneity. We prove that AffPCL reduces sample complexity over independent learning by a factor of $\max\{n^{-1}, \delta\}$, where $n$ is the number of agents and $\delta\in[0,1]$ measures their heterogeneity. This affinity-based acceleration automatically interpolates between the linear speedup of federated learning in homogeneous settings and the baseline of independent learning, without requiring prior knowledge of the system. Our analysis further reveals that an agent may obtain linear speedup even by collaborating with arbitrarily dissimilar agents, unveiling new insights into personalization and collaboration in the high heterogeneity regime.

Intelligent robotic disassembly of end-of-life (EOL) products has been a long-standing challenge in robotics. While machine learning techniques have shown promise, the lack of specialized hardware limits their application in real-world scenarios. We introduce DeGrip, a customized gripper designed for the disassembly of EOL computer desktops. DeGrip provides three degrees of freedom (DOF), enabling arbitrary configurations within the disassembly environment when mounted on a robotic manipulator. It employs a cable-driven transmission mechanism that reduces its overall size and enables operation in confined spaces. The wrist is designed to decouple the actuation of wrist and jaw joints. We also developed an EOL desktop disassembly environment in Isaac Sim to evaluate the effectiveness of DeGrip. The tasks were designed to demonstrate its ability to operate in confined spaces and disassemble components in arbitrary configurations. The evaluation results confirm the capability of DeGrip for EOL desktop disassembly.

While sequential task assignment for a single agent has been widely studied, such problems in a multi-agent setting, where the agents have heterogeneous task preferences or capabilities, remain less well-characterized. We study a multi-agent task assignment problem where a central planner assigns recurring tasks to multiple members of a team over a finite time horizon. For any given task, the members have heterogeneous capabilities in terms of task completion times, task resource consumption (which can model variables such as energy or attention), and preferences in terms of the rewards they collect upon task completion. We assume that the reward, execution time, and resource consumption for each member to complete any task are stochastic with unknown distributions. The goal of the planner is to maximize the total expected reward that the team receives over the problem horizon while ensuring that the resource consumption required for any assigned task is within the capability of the agent. We propose and analyze a bandit algorithm for this problem. Since the bandit algorithm relies on solving an optimal task assignment problem repeatedly, we analyze the achievable regret in two cases: when we can solve the optimal task assignment exactly and when we can solve it only approximately.

We study a nonstationary bandit problem where rewards depend on both actions and latent states, the latter governed by unknown linear dynamics. Crucially, the state dynamics also depend on the actions, resulting in tension between short-term and long-term rewards. We propose an explore-then-commit algorithm for a finite horizon $T$. During the exploration phase, random Rademacher actions enable estimation of the Markov parameters of the linear dynamics, which characterize the action-reward relationship. In the commit phase, the algorithm uses the estimated parameters to design an optimized action sequence for long-term reward. Our proposed algorithm achieves $\tilde{\mathcal{O}}(T^{2/3})$ regret. Our analysis handles two key challenges: learning from temporally correlated rewards, and designing action sequences with optimal long-term reward. We address the first challenge by providing near-optimal sample complexity and error bounds for system identification using bilinear rewards. We address the second challenge by proving an equivalence with indefinite quadratic optimization over a hypercube, a known NP-hard problem. We provide a sub-optimality guarantee for this problem, enabling our regret upper bound. Lastly, we propose a semidefinite relaxation with Goemans-Williamson rounding as a practical approach.

Integrated sensing and communications (ISAC), radar, and beamforming require real-time, high-resolution estimation algorithms to determine delay-Doppler values of specular paths within the wireless propagation channel. Our contribution is the measurement-based performance comparison of the delay-Doppler estimation between three different algorithms, comprising maximum likelihood (ML), convolutional neural network (CNN), and constant false alarm rate (CFAR) approaches. We apply these algorithms to publicly available channel data which includes two spherical targets with analytically describable delay-Doppler parameters. The comparison of the three algorithms features the target detection rate, root mean squared errors (RMSEs) of the delay-Doppler estimates, and a runtime analysis. Notably, all three algorithms demonstrate similar parameter estimation capabilities in bi-static scenarios, achieving target detection probabilities of up to 80%. Conversely, forward and backward scattering conditions pose a problem to the estimation due to strong line-of-sight (LoS) contribution, reducing the corresponding detection probability down to 0%.

We present a fast, differentiable, GPU-accelerated optimization method for ray path tracing in environments containing planar reflectors and straight diffraction edges. Based on Fermat's principle, our approach reformulates the path-finding problem as the minimization of total path length, enabling efficient parallel execution on modern GPU architectures. Unlike existing methods that require separate algorithms for reflections and diffractions, our unified formulation maintains consistent problem dimensions across all interaction sequences, making it particularly suitable for vectorized computation. Through implicit differentiation, we achieve efficient gradient computation without differentiating through solver iterations, significantly outperforming traditional automatic differentiation approaches. Numerical simulations demonstrate convergence rates comparable to specialized Newton methods while providing superior scalability for large-scale applications. The method integrates seamlessly with differentiable programming libraries such as JAX and DrJIT, enabling new possibilities in inverse design and optimization for wireless propagation modeling. The source code is openly available at https://github.com/jeertmans/fpt-jax.

Effective human-AI collaboration on complex reasoning tasks requires that users understand and interact with the model's process, not just receive an output. However, the monolithic text from methods like Chain-of-Thought (CoT) prevents this, as current interfaces lack real-time verbalization and robust user barge-in. We present AsyncVoice Agent, a system whose asynchronous architecture decouples a streaming LLM backend from a conversational voice frontend. This design allows narration and inference to run in parallel, empowering users to interrupt, query, and steer the model's reasoning process at any time. Objective benchmarks show this approach reduces interaction latency by more than 600x compared to monolithic baselines while ensuring high fidelity and competitive task accuracy. By enabling a two-way dialogue with a model's thought process, AsyncVoice Agent offers a new paradigm for building more effective, steerable, and trustworthy human-AI systems for high-stakes tasks.

Bistatic radar measurements offer unique spatial diversity and enhanced target characterization capabilities, rendering them increasingly vital for contemporary sensing application research. The reliability of such measurements is contingent upon precise system and antenna calibration. The prevailing technique is the substitution method, which involves the use of known reference objects. We propose an over-the-air calibration algorithm for spherical bistatic measurement systems. Our method is both significantly simpler and twice as fast as existing algorithms. The application of our technique to reflectivity measurements of a metal sphere from 76 to 81 GHz demonstrates a dynamic range enhancement of up to 40 dB when compared with uncalibrated data. A comparison with simulation data demonstrates a high degree of agreement between measurement and simulation.

Structured reporting (SR) and artificial intelligence (AI) may transform how radiologists interact with imaging studies. This prospective study (July to December 2024) evaluated the impact of three reporting modes: free-text (FT), structured reporting (SR), and AI-assisted structured reporting (AI-SR), on image analysis behavior, diagnostic accuracy, efficiency, and user experience. Four novice and four non-novice readers (radiologists and medical students) each analyzed 35 bedside chest radiographs per session using a customized viewer and an eye-tracking system. Outcomes included diagnostic accuracy (compared with expert consensus using Cohen's $\kappa$), reporting time per radiograph, eye-tracking metrics, and questionnaire-based user experience. Statistical analysis used generalized linear mixed models with Bonferroni post-hoc tests with a significance level of ($P \le .01$). Diagnostic accuracy was similar in FT ($\kappa = 0.58$) and SR ($\kappa = 0.60$) but higher in AI-SR ($\kappa = 0.71$, $P < .001$). Reporting times decreased from $88 \pm 38$ s (FT) to $37 \pm 18$ s (SR) and $25 \pm 9$ s (AI-SR) ($P < .001$). Saccade counts for the radiograph field ($205 \pm 135$ (FT), $123 \pm 88$ (SR), $97 \pm 58$ (AI-SR)) and total fixation duration for the report field ($11 \pm 5$ s (FT), $5 \pm 3$ s (SR), $4 \pm 1$ s (AI-SR)) were lower with SR and AI-SR ($P < .001$ each). Novice readers shifted gaze towards the radiograph in SR, while non-novice readers maintained their focus on the radiograph. AI-SR was the preferred mode. In conclusion, SR improves efficiency by guiding visual attention toward the image, and AI-prefilled SR further enhances diagnostic accuracy and user satisfaction.

Solving the nonlinear AC optimal power flow (AC OPF) problem remains a major computational bottleneck for real-time grid operations. In this paper, we propose a residual learning paradigm that uses fast DC optimal power flow (DC OPF) solutions as a baseline, and learns only the nonlinear corrections required to provide the full AC-OPF solution. The method utilizes a topology-aware Graph Neural Network with local attention and two-level DC feature integration, trained using a physics-informed loss that enforces AC power-flow feasibility and operational limits. Evaluations on OPFData for 57-, 118-, and 2000-bus systems show around 25% lower MSE, up to 3X reduction in feasibility error, and up to 13X runtime speedup compared to conventional AC OPF solvers. The model maintains accuracy under N-1 contingencies and scales efficiently to large networks. These results demonstrate that residual learning is a practical and scalable bridge between linear approximations and AC-feasible OPF, enabling near real-time operational decision making.

Accurate voltage estimation in distribution networks is critical for real-time monitoring and increasing the reliability of the grid. As DER penetration and distribution level voltage variability increase, robust distribution system state estimation (DSSE) has become more essential to maintain safe and efficient operations. Traditional DSSE techniques, however, struggle with sparse measurements and the scale of modern feeders, limiting their scalability to large networks. This paper presents a hierarchical graph neural network for substation-level voltage estimation that exploits both electrical topology and physical features, while remaining robust to the low observability levels common to real-world distribution networks. Leveraging the public SMART-DS datasets, the model is trained and evaluated on thousands of buses across multiple substations and DER penetration scenarios. Comprehensive experiments demonstrate that the proposed method achieves up to 2 times lower RMSE than alternative data-driven models, and maintains high accuracy with as little as 1% measurement coverage. The results highlight the potential of GNNs to enable scalable, reproducible, and data-driven voltage monitoring for distribution systems.

Microgrids are emerging as key enablers of resilient, sustainable, and intelligent power systems, but they continue to face challenges in dynamic disturbance handling, protection coordination, and uncertainty. Recent efforts have explored Brain Emotional Learning (BEL) controllers as bio-inspired solutions for microgrid control. Building on this growing trajectory, this article introduces a new paradigm for Neuro-Microgrids, inspired by the brain's sensorimotor gating mechanisms, specifically the Prepulse Inhibition (PPI) and Prepulse Facilitation (PPF). Sensorimotor gating offers a biological model for selectively suppressing or amplifying responses depending on contextual relevance. By mapping these principles onto the hierarchical control architecture of microgrids, we propose a Sensorimotor Gating-Inspired Neuro-Microgrid (SG-NMG) framework. In this architecture, PPI-like control decisions correspond to protective damping in primary and secondary management of microgrids, whereas PPF-like decisions correspond to adaptive amplification of corrective control actions. The framework is presented through analytical workflow design, neuro-circuitry analogies, and integration with machine learning methods. Finally, open challenges and research directions are outlined, including the mathematical modeling of gating, digital twin validation, and cross-disciplinary collaboration between neuroscience and industrial power systems. The resulting paradigm highlights sensorimotor gating as a promising framework for designing self-protective, adaptive, and resilient microgrids.

This work addresses the radio resource management (RRM) design in downlink full-duplex integrated sensing and communications (ISAC) systems, jointly optimizing timeslot allocation and beam selection under imperfect self-interference cancellation. Timeslot allocation governs the distribution of discrete channel uses between sensing and communication tasks, while beam selection determines transmit and receive directions along with adaptive beamwidths. The joint design leads to a semi-infinite, nonconvex mixed-integer nonlinear program (MINLP), which is difficult to solve. To overcome this, we develop a tailored reformulation strategy that transforms the problem into a tractable mixed-integer linear program (MILP), enabling globally optimal solutions. Our approach provides insights into the coordinated optimization of timeslot allocation and beam selection, enhancing the efficiency of full-duplex ISAC systems while ensuring resilience against residual self-interference.

This paper presents a safety-critical control framework to maintain bounded lateral motions for vehicles braking on asymmetric surfaces. We synthesize a brake controller that assists drivers and guarantees safety against excessive lateral motions (i.e., prevents the vehicle from spinning out) while minimizing the stopping distance. We address this safety-critical control problem in the presence of input constraints, since braking forces are limited by the available friction on the road. We use backup control barrier functions for safe control design. As this approach requires the construction of a backup set and a backup controller, we propose a novel, systematic method to creating valid backup set-backup controller pairs based on feedback linearization and continuous-time Lyapunov equations. We use simple examples to demonstrate our proposed safety-critical control method. Finally, we implement our approach on a four-wheel vehicle model for braking on asymmetric surfaces and present simulation results.

This paper presents machine learning method for tuning of cavity duplexer with a large amount of adjustment screws. After testing we declined conventional reinforcement learning approach and reformulated our task in the supervised learning setup. The suggested neural network architecture includes 1d ResNet-like backbone and processing of some additional information about S-parameters, like the shape of curve and peaks positions and amplitudes. This neural network with external control algorithm is capable to reach almost the tuned state of the duplexer within 4-5 rotations per screw.

In this paper, we proposed a hybrid simultaneous wireless information and power transfer (SWIPT)-enabled multiple-input multiple-output (MIMO) architecture, where the base station (BS) uses a conventional RF transmitter for downlink transmission and a Rydberg atomic quantum receiver (RAQR) for receiving uplink signal from Internet of Things (IoT) devices. To fully exploit this integration, we jointly design the transmission scheme and the power-splitting strategy to maximize the sum rate, which leads to a non-convex problem. To address this challenge, we first derive closed-form lower bounds on the uplink achievable rates for maximum ratio combining (MRC) and zero-forcing (ZF), as well as on the downlink rate and harvested energy for maximum ratio transmission (MRT) and ZF precoding. Building upon these bounds, we propose an iterative algorithm relying on the best monomial approximation and geometric programming (GP) to solve the non-convex problem. Finally, simulations validate the tightness of our derived lower bounds and demonstrate the superiority of the proposed algorithm over benchmark schemes. Importantly, by integrating RAQR with SWIPT-enabled MIMO, the BS can reliably detect weak uplink signals from IoT devices powered only by harvested energy, enabling battery-free communication.

Light field images capture multi-view scene information and play a crucial role in 3D scene reconstruction. However, their high-dimensional nature results in enormous data volumes, posing a significant challenge for efficient compression in practical storage and transmission scenarios. Although neural representation-based methods have shown promise in light field image compression, most approaches rely on direct coordinate-to-pixel mapping through implicit neural representation (INR), often neglecting the explicit modeling of scene structure. Moreover, they typically lack end-to-end rate-distortion optimization, limiting their compression efficiency. To address these limitations, we propose SANR, a Scene-Aware Neural Representation framework for light field image compression with end-to-end rate-distortion optimization. For scene awareness, SANR introduces a hierarchical scene modeling block that leverages multi-scale latent codes to capture intrinsic scene structures, thereby reducing the information gap between INR input coordinates and the target light field image. From a compression perspective, SANR is the first to incorporate entropy-constrained quantization-aware training (QAT) into neural representation-based light field image compression, enabling end-to-end rate-distortion optimization. Extensive experiment results demonstrate that SANR significantly outperforms state-of-the-art techniques regarding rate-distortion performance with a 65.62% BD-rate saving against HEVC.

This paper investigates the performance advantages of Rydberg atomic quantum (RAQ)-based multiple-input multiple-output (MIMO) satellites for enhancing direct ground-to-space uplink access.We analytically evaluate the impact of Rydberg atoms on channel estimation by deriving closed-form expressions for the mean-square error (MSE) and normalized mean-square error (NMSE). Based on the estimated channels, we further derive lower bounds on the achievable data rates for maximum ratio combining (MRC) and zero-forcing (ZF) detection schemes. Rigorous analysis demonstrates that RAQ-MIMO outperforms conventional radio-frequency (RF) MIMO under both Rayleigh and satellite channel conditions. Specifically, compared with conventional MIMO, RAQR achieves a “squaring" gain under Rayleigh fading, especially in long-distance transmission scenarios with stringent power constraints. In contrast, under line-of-sight (LoS)-dominated satellite channels, this gain saturates as channel-estimation benefits diminish, with the remaining improvement primarily arising from the normalized noise background. Monte Carlo simulations validate the analytical results and show that the performance gains of RAQ-MIMO satellites translate into smaller antenna apertures, lower transmit power, and longer communication ranges, thereby paving the way for next-generation satellite networks.

In this paper, we propose a novel and low-complexity atomic multiple-input multiple-output (MIMO) receiver architecture assisted by a reconfigurable intelligent surface (RIS). By introducing RIS and utilizing pulse amplitude modulation (PAM), the phase of the transmitted signal is effectively aligned with that of the local oscillator (LO), thereby mitigating phase ambiguity and substantially reducing both signal detection complexity and overall receiver complexity.To tackle the resulting non-convex optimization problem, we reformulate it into a tractable form by minimizing the Frobenius norm of an equivalent matrix, which is efficiently solved using an Adam-based gradient descent algorithm.

The Sixth Generation (6G) radio technology is expected to include local 6G networks as a special use case, extending the capabilities of `generic' 6G networks towards more demanding performance requirements. Reconfigurable intelligent surfaces (RISs) offer a novel paradigm for next-generation wireless communications, especially in the context of local 6G networks, enabling advanced signal propagation control through intelligent phase-shift configurations. However, in practical deployments, their performance can be adversely affected by electromagnetic interference (EMI) from external sources and inter-RIS reflections (IRR) caused by signal reflections between multiple colocated RIS units. This paper presents a comprehensive analysis of the joint impact of EMI and IRR in a multi-RIS multi-cell system deployed within an indoor factory environment. A detailed evaluation study is first carried out to investigate their impact on system performance. System-level simulations demonstrate that the joint impact of EMI and IRR degrades system performance more significantly than their individual effects, particularly as RIS dimensions and transmit power increase. To address these adverse effects, an alternate optimization algorithm using the Riemannian conjugate gradient method is then proposed. The novel algorithm optimizes the phase shifts of the RIS elements considering the spatial correlation among their associated channels, and is found to provide up to several orders of magnitude gains in terms of the system sum rate and the outage probability.

Dynamic line rating (DLR) enables greater utilization of existing transmission lines by leveraging real-time weather data. However, the elevated temperature operation (ETO) of conductors under DLR, particularly in the presence of uncertainty, is often overlooked, despite its long-term impact on conductor health. This paper addresses ETO under DLR and wind power uncertainty by 1) quantifying risk-based depreciation costs associated with ETO, 2) characterizing correlation-driven ETO risk from wind power and DLR forecast errors, and 3) proposing a Conductor Health-Aware Unit Commitment (CHA-UC) that internalizes these costs in operational decisions. CHA-UC incorporates a robust linear approximation of conductor temperature and integrates expected depreciation costs due to hourly ETO into the objective function. Case studies on the Texas 123-bus backbone test system demonstrate that the proposed CHA-UC model reduces the total cost by 0.75% and renewable curtailment by 82% compared to static line rating (SLR) and outperforms quantile regression forest-based methods, while conventional DLR operation without risk consideration resulted in higher costs due to excessive ETO. Further analysis shows that CHA-UC achieves safer line utilization by shifting generator commitments and endogenously adapting to uncertainty correlation, relaxing flows under risk-hedging conditions and tightening flows under risk-amplifying conditions.

Camera movement classification (CMC) models trained on contemporary, high-quality footage often degrade when applied to archival film, where noise, missing frames, and low contrast obscure motion cues. We bridge this gap by assembling a unified benchmark that consolidates two modern corpora into four canonical classes and restructures the HISTORIAN collection into five balanced categories. Building on this benchmark, we introduce DGME-T, a lightweight extension to the Video Swin Transformer that injects directional grid motion encoding, derived from optical flow, via a learnable and normalised late-fusion layer. DGME-T raises the backbone's top-1 accuracy from 81.78% to 86.14% and its macro F1 from 82.08% to 87.81% on modern clips, while still improving the demanding World-War-II footage from 83.43% to 84.62% accuracy and from 81.72% to 82.63% macro F1. A cross-domain study further shows that an intermediate fine-tuning stage on modern data increases historical performance by more than five percentage points. These results demonstrate that structured motion priors and transformer representations are complementary and that even a small, carefully calibrated motion head can substantially enhance robustness in degraded film analysis. Related resources are available at https://github.com/linty5/DGME-T.

Accurate localization of the seizure onset zone (SOZ) is essential for successful epilepsy surgery, yet the reliability of commonly used interictal biomarkers is limited by temporal variability and behavioral state. This study aims to investigate the circadian and sleep-dependent dynamics of epileptic biomarkers and to identify conditions that maximize seizure localization precision. Longterm intracranial EEG recordings from nine patients with drug-resistant focal epilepsy were retrospectively analyzed using automated signal processing and machine learning techniques. Interictal spikes, spike sequences, high-frequency oscillations (HFOs), and pathological HFOs were automatically detected, while sleep and wake states were classified using the alpha-delta power ratio. Biomarker rates, spatial distributions, and localization accuracy were quantitatively evaluated using Euclidean distance relative to the clinically defined SOZ. The results show that all biomarkers exhibit significantly higher rates during sleep, with pronounced early-morning peaks. Importantly, spike sequences and pathological HFOs demonstrated superior spatial precision compared to conventional spikes or HFOs alone. Mean distances to the SOZ were substantially lower for pathological HFOs and spike sequences, with statistically significant differences among biomarkers (ANOVA, p < 0.001). These findings demonstrate that sleep-state analysis, particularly using propagated spike sequences and pathological HFOs, substantially improves SOZ localization accuracy. The proposed framework provides practical guidance for sleep-focused presurgical EEG analysis and supports the development of automated and clinically efficient seizure localization systems.

This paper presents a practical alternative to programmable-logic-controller-centric automation by implementing an event-driven architecture built with industrial Internet of Things tools. A layered design on a local edge server (i) abstracts actuators, (ii) enforces mutual exclusion of shared physical resources through an interlock with priority queueing, (iii) composes deterministic singular operations, and (iv) orchestrates complete workflows as state machines in Node-RED, with communication over MQTT. The device layer uses low-cost ESP32-based gateways to interface sensors and actuators, while all automation logic is offloaded to the server side. As part of a larger project involving the first scientifically-documented integration of Industry 4.0 technologies in a maple syrup boiling center, this work demonstrates the deployment of the proposed system as a case-study. Evaluation over an entire production season shows median message time of flight around one tenth of a second, command issuance-to-motion latencies of about two to three seconds, and command completion near six seconds dominated by actuator mechanics; operation runtimes span tens of seconds to minutes. These results indicate that network and orchestration overheads are negligible relative to process dynamics, enabling modular, distributed control without compromising determinism or fault isolation. The approach reduces material and integration effort, supports portable containerized deployment, and naturally enables an edge/cloud split in which persistence and analytics are offloaded while automation remains at the edge.

This paper proposes a pitch control strategy to mitigate the underwater acoustic footprint of offshore wind turbines, a measure that will soon become necessary to minimize impacts on marine life, which rely on sound for communication, navigation, and survival. First, we quantify the underwater acoustic signature of blade-generated aerodynamic noise from three reference turbines, the NREL 5 MW, DTU 10 MW, and IEA 22 MW, using coupling blade element momentum and coupled air-water acoustic propagation modeling. Second, we propose and implement an open-loop individual pitch control (IPC) strategy that modulates the pitch of the blade at the blade passing frequency to attenuate the overall sound pressure level (OSPL) and the amplitude modulation (AM) of the transmitted noise. Third, we benchmark IPC performance against conventional pitch schemes. The results indicate that up to 5 dB reductions in OSPL and a decrease in AM depth 20% can be achieved with a pitch variation of $\Delta\theta\approx 5^\circ$, with small losses (5-10%) in energy capture. These findings highlight a previously underappreciated noise pathway and demonstrate that targeted blade-pitch modulation can mitigate its impact.

Beyond-diagonal reconfigurable intelligent surface (BD-RIS) has recently been introduced to enable advanced control over electromagnetic waves to further increase the benefits of traditional RIS in enhancing signal quality and improving spectral and energy efficiency for next-generation wireless networks. A significant issue in designing and deploying BD-RIS is the tradeoff between its performance and circuit complexity. While existing studies have explored optimal architectures to minimize circuit complexity in ideal BD-RIS, architecture discovery for non-ideal BD-RIS remains uninvestigated. Consequently, how non-idealities and circuit complexity jointly affect the performance of BD-RIS remains unclear, making it difficult to achieve the performance-circuit complexity tradeoff in the presence of non-idealities. Essentially, architecture discovery for non-ideal BD-RIS faces challenges from both the computational complexity of global architecture search and the difficulty in achieving global optima. To tackle these challenges, we propose a learning-based two-tier architecture discovery framework (LTTADF) consisting of an architecture generator and a performance optimizer to jointly discover optimal architectures for non-ideal BD-RIS given specific circuit complexities, which can effectively explore over a large architecture space while avoiding getting trapped in poor local optima and thus achieving near-optimal solutions for the performance optimization. Numerical results provide valuable insights for deploying non-ideal BD-RIS considering the performance-circuit complexity tradeoff.

European countries are ambitious in both the net-zero transition and offshore energy resource development. The Irish and UK governments announced their commitments to offshore wind capacities - 37 and 125 GW, respectively, in 2050, more than two times higher than their projected power demands. While other continental countries, such as Germany, are calling for cleaner fuel resources. Exporting surplus offshore green hydrogen and bridging supply and demand could be pivotal in carbon emission mitigation for Europe. Yet, the potentials of these Island countries, are usually underestimated. This paper developed a bottom-up method to investigate the role of offshore hydrogen from Ireland and the UK in the decarbonisation of the entire Europe. We evaluate the future hydrogen/ammonia trading and the contributions of each country in carbon emission mitigation, considering their relative cost-competitiveness in offshore hydrogen production, domestic hourly power and gas system operation, and international shipping costs. Results indicate that the offshore green hydrogen could reduce 175.16 Mt/year of carbon dioxide emissions in Europe. The UK will be the largest hydrogen supplier from 2030 to 2040, while surpassed by Ireland in 2050, with 161 TWh of hydrogen exports to France and Spain. The offshore green hydrogen can contribute to 175.16 Mt of annual carbon dioxide emission reductions in total. This general flow of hydrogen from the West to the East not only facilitates Europe's net-zero progress, but also reshapes the energy supply structure and helps to ensure energy security across the European continent.

Many authors have examined various boundary behaviors of injective harmonic mappings in the open unit disk. Building on Laugesen's work, Bshouty and others explored the boundary behavior of harmonic mappings under different conditions. In this paper, we extend their work and find out the angular limits of the arguments and logarithms of analytic functions under various conditions. We also examined the dilatation possesses only a finite set of zeros within any stolz angle if the first derivative of harmonic function $f$ at the boundary is positive infinity.

Freehand 3D ultrasound (US) imaging using conventional 2D probes offers flexibility and accessibility for diverse clinical applications but faces challenges in accurate probe pose estimation. Traditional methods depend on costly tracking systems, while neural network-based methods struggle with image noise and error accumulation, compromising reconstruction precision. We propose a cost-effective and versatile solution that leverages lightweight cameras and visual servoing in simulated environments for precise 3D US imaging. These cameras capture visual feedback from a textured planar workspace. To counter occlusions and lighting issues, we introduce an image restoration method that reconstructs occluded regions by matching surrounding texture patterns. For pose estimation, we develop a simulation-in-the-loop approach, which replicates the system setup in simulation and iteratively minimizes pose errors between simulated and real-world observations. A visual servoing controller refines the alignment of camera views, improving translational estimation by optimizing image alignment. Validations on a soft vascular phantom, a 3D-printed conical model, and a human arm demonstrate the robustness and accuracy of our approach, with Hausdorff distances to the reference reconstructions of 0.359 mm, 1.171 mm, and 0.858 mm, respectively. These results confirm the method's potential for reliable freehand 3D US reconstruction.

Phase-based features related to vocal source characteristics can be incorporated into magnitude-based speaker recognition systems to improve the system performance. However, traditional feature-level fusion methods typically ignore the unique contributions of speaker semantics in the magnitude and phase domains. To address this issue, this paper proposed a feature-level fusion framework using the co-attention mechanism for speaker recognition. The framework consists of two separate sub-networks for the magnitude and phase domains respectively. Then, the intermediate high-level outputs of both domains are fused by the co-attention mechanism before a pooling layer. A correlation matrix from the co-attention module is supposed to re-assign the weights for dynamically scaling contributions in the magnitude and phase domains according to different pronunciations. Experiments on VoxCeleb showed that the proposed feature-level fusion strategy using the co-attention mechanism gave the Top-1 accuracy of 97.20%, outperforming the state-of-the-art system with 0.82% absolutely, and obtained EER reduction of 0.45% compared to single feature system using FBank.