Papers

Online Speech Enhancement was mainly reserved for predictive models. A key advantage of these models is that for an incoming signal frame from a stream of data, the model is called only once for enhancement. In contrast, generative Speech Enhancement models often require multiple calls, resulting in a computational complexity that is too high for many online speech enhancement applications. This work presents the Diffusion Buffer, a generative diffusion-based Speech Enhancement model which only requires one neural network call per incoming signal frame from a stream of data and performs enhancement in an online fashion on a consumer-grade GPU. The key idea of the Diffusion Buffer is to align physical time with Diffusion time-steps. The approach progressively denoises frames through physical time, where past frames have more noise removed. Consequently, an enhanced frame is output to the listener with a delay defined by the Diffusion Buffer, and the output frame has a corresponding look-ahead. In this work, we extend upon our previous work by carefully designing a 2D convolutional UNet architecture that specifically aligns with the Diffusion Buffer's look-ahead. We observe that the proposed UNet improves performance, particularly when the algorithmic latency is low. Moreover, we show that using a Data Prediction loss instead of Denoising Score Matching loss enables flexible control over the trade-off between algorithmic latency and quality during inference. The extended Diffusion Buffer equipped with a novel NN and loss function drastically reduces the algorithmic latency from 320 - 960 ms to 32 - 176 ms with an even increased performance. While it has been shown before that offline generative diffusion models outperform predictive approaches in unseen noisy speech data, we confirm that the online Diffusion Buffer also outperforms its predictive counterpart on unseen noisy speech data.

Waveguide-based pinching-antenna systems (PASS) have recently emerged as a promising solution to mitigate severe propagation losses in millimeter-wave and terahertz bands by intelligently and flexibly establishing line-of-sight links. However, their reliance on wire-based feeding confines deployment to areas near the base station (BS), limiting installation flexibility and making them cost-ineffective for serving distant users or regions. To overcome this challenge, this article proposes wireless-fed pinchingantenna systems (Wi-PASS), which employ wireless feeding to energize waveguides. Wi-PASS offer a practical and cost-efficient means to extend coverage beyond the BS vicinity. Several indoor and outdoor use cases demonstrate Wi-PASS advantages over PASS. Numerical results further show that Wi-PASS deliver higher data rates than conventional fixed-antenna systems, confirming the superior feasibility and performance of Wi-PASS. Key future research directions are also discussed to advance Wi-PASS deployment.

Quantized observations are ubiquitous in a wide range of applications across engineering and the social sciences, and algorithms based on the $\ell_1$-norm are well recognized for their robustness to outliers compared with their $\ell_2$-based counterparts. Nevertheless, adaptive identification methods that integrate quantized observations with $\ell_1$-optimization remain largely underexplored. Motivated by this gap, we develop a novel $\ell_1$-based adaptive identification algorithm specifically designed for quantized observations. Without relying on the traditional persistent excitation condition, we establish global convergence of the parameter estimates to their true values and show that the average regret asymptotically vanishes as the data size increases. Finally, we apply our new identification algorithm to a judicial sentencing problem using real-world data, which demonstrates its superior performance and practical significance.

Modulo sampling enables acquisition of signals with unlimited dynamic range by folding the input into a bounded interval prior to sampling, thus eliminating the risk of signal clipping and preserving information without requiring highresolution ADCs. While this enables low-cost hardware, the nonlinear distortion introduced by folding presents recovery challenges, particularly under noise and quantization. We propose a model-based deep unfolding network tailored to this setting, combining the interpretability of classical compress sensing (CS) solvers with the flexibility of learning. A key innovation is a soft-quantization module that encodes the modulo prior by guiding the solution toward discrete multiples of the folding range in a differentiable and learnable way. Our method, modulo soft-quantized unfolded iterative decoder (mSQUID), achieves superior reconstruction performance at low sampling rates under additive Gaussian noise. We further demonstrate its utility in a challenging case where signals with vastly different amplitudes and disjoint frequency bands are acquired simultaneously and quantized. In this scenario, classical sampling often struggles due to weak signal distortion or strong signal clipping, while our approach is able to recover the input signals. Our method also offers significantly reduced runtimes, making it suitable for real-time, resource-limited systems.

Despite achieving impressive results on standard benchmarks, large foundational models still struggle against code-switching test cases. When data scarcity cannot be used as the usual justification for poor performance, the reason may lie in the infrequent occurrence of code-switched moments, where the embedding of the second language appears subtly. Instead of expecting the models to learn this infrequency on their own, it might be beneficial to provide the training process with labels. Evaluating model performance on code-switching data requires careful localization of code-switching points where recognition errors are most consequential, so that the analysis emphasizes mistakes occurring at those moments. Building on this observation, we leverage the difference between the embedded and the main language to highlight those code-switching points and thereby emphasize learning at those locations. This simple yet effective differentiable surrogate mitigates context bias during generation – the central challenge in code-switching – thereby improving the model's robustness. Our experiments with Arabic and Chinese-English showed that the models are able to predict the switching places more correctly, reflected by the reduced substitution error.

Large speech foundation models achieve strong performance across many domains, but they often require adaptation to handle local needs such as code-switching, where speakers mix languages within the same utterance. Direct fine-tuning of these models risks overfitting to the target domain and overwriting the broad capabilities of the base model. To address this challenge, we explore Bayesian factorized adapters for speech foundation models, which place priors near zero to achieve sparser adaptation matrices and thereby retain general performance while adapting to specific domains. We apply our approach to the Whisper model and evaluate on different multilingual code-switching scenarios. Our results show only minimal adaptation loss while significantly reducing catastrophic forgetting of the base model. Compared to LoRA, our method achieves a backward gain of 54% with only a 4% drop on the new domain. These findings highlight the effectiveness of Bayesian adaptation for fine-tuning speech foundation models without sacrificing generalization.

In this technical note, we prove that the ODEFTC algorithm constitutes the first optimal distributed state estimator for continuous-time linear time-varying systems subject to stochastic disturbances. Particularly, we formally show that it is able to asymptotically recover the performance, in terms of error covariance of the estimates at each node, of the centralized Kalman-Bucy filter, which is known to be the optimal filter for the considered class of systems. Moreover, we provide a simple sufficient value for the consensus gain to guarantee the stability of the distributed estimator.

Simulation studies are conducted at different levels of details for assessing the performance of Media Access Control (MAC) protocols in Wireless Sensor Networks (WSN). In the present-day scenario where hundreds of MAC protocols have been proposed, it is important to assess the quality of performance evaluation being conducted for each of the proposed protocols. It therefore becomes crucial to compare the results of high-level theoretical simulations with the detailed implementation results before any network protocol could be deployed for a real-world scenario. In this work, we present a comparison of high-level theoretical and detailed implementation results for Adaptive and Dynamic Polling-MAC (ADP-MAC). MATLAB has been used for conducting initial theoretical simulations and TinyOS has been used to develop the detailed implementation of protocol for Mica2 platform. Performance evaluation of ADP-MAC using the two levels of simulation has been conducted based on energy and delay. In the high-level implementation, energy consumption was found to be decreasing whereas delay was found to be increasing for increasing channel polling intervals. On the other hand, when detailed implementation was developed, it was observed that both energy consumption and delay revealed an increasing trend with the increasing polling intervals. Therefore, it has been shown that the trends for high- and low-level simulations for ADP-MAC are significantly different, due to the lack of realistic assumptions in the higher-level study.

Managing delay is one of the core requirements of industrial automation applications due to the high risk associated for equipment and human lives. Using efficient Media Access Control (MAC) schemes guarantees the timely transmission of critical data, particularly in the industrial environments where heterogeneous data is inherently expected. This paper compares the performance of Fragmentation based MAC (FROG-MAC) against Fuzzy Priority Scheduling based MAC (FPS-MAC), both of which have been designed to optimize the performance of heterogenous wireless networks. Contiki has been used as a simulation platform and a single hop star topology has been assumed to resemble the industrial environment. It has been shown that FROG-MAC has the potential to outperform FPS-MAC in terms of energy efficiency and delay both, due to its inherent feature of interrupting ongoing lower priority transmission on the channel.

Networked Control Systems (NCSs) are integral in critical infrastructures such as power grids, transportation networks, and production systems. Ensuring the resilient operation of these large-scale NCSs against cyber-attacks is crucial for societal well-being. Over the past two decades, extensive research has been focused on developing metrics to quantify the vulnerabilities of NCSs against attacks. Once the vulnerabilities are quantified, mitigation strategies can be employed to enhance system resilience. This article provides a comprehensive overview of methods developed for assessing NCS vulnerabilities and the corresponding mitigation strategies. Furthermore, we emphasize the importance of probabilistic risk metrics to model vulnerabilities under adversaries with imperfect process knowledge. The article concludes by outlining promising directions for future research.

In large scale short video platforms, CDN resource selection plays a critical role in maintaining Quality of Experience (QoE) while controlling escalating traffic costs. To better understand this phenomenon, we conduct in the wild network measurements during video playback in a production short video system. The results reveal that CDNs delivering higher average QoE often come at greater financial cost, yet their connection quality fluctuates even within a single video underscoring a fundamental and dynamic trade off between QoE and cost. However, the problem of sustaining high QoE under cost constraints remains insufficiently investigated in the context of CDN selection for short video streaming. To address this, we propose PIRA, a dynamic resource selection algorithm that optimizes QoE and cost in real time during video playback. PIRA formally integrating QoE and cost by a mathematical model, and introduce a intra video control theoretic CDN resource selection approach which can balance QoE and cost under network dynamics. To reduce the computation overheads, PIRA employs state space pruning and adaptive parameter adjustment to efficiently solve the high dimensional optimization problem. In large scale production experiments involving 450,000 users over two weeks, PIRA outperforms the production baseline, achieving a 2.1% reduction in start up delay, 15.2% shorter rebuffering time, and 10% lower average unit traffic cost, demonstrating its effectiveness in balancing user experience and financial cost at scale.

Deep learning-based semantic communication has largely relied on analog or semi-digital transmission, which limits compatibility with modern digital communication infrastructures. Recent studies have employed vector quantization (VQ) to enable discrete semantic transmission, yet existing methods neglect channel state information during codebook optimization, leading to suboptimal robustness. To bridge this gap, we propose a channel-aware vector quantization (CAVQ) algorithm within a joint source-channel coding (JSCC) framework, termed VQJSCC, established on a discrete memoryless channel. In this framework, semantic features are discretized and directly mapped to modulation constellation symbols, while CAVQ integrates channel transition probabilities into the quantization process, aligning easily confused symbols with semantically similar codewords. A multi-codebook alignment mechanism is further introduced to handle mismatches between codebook order and modulation order by decomposing the transmission stream into multiple independently optimized subchannels. Experimental results demonstrate that VQJSCC effectively mitigates the digital cliff effect, achieves superior reconstruction quality across various modulation schemes, and outperforms state-of-the-art digital semantic communication baselines in both robustness and efficiency.

Robust speaker verification under noisy conditions remains an open challenge. Conventional deep learning methods learn a robust unified speaker representation space against diverse background noise and achieve significant improvement. In contrast, this paper presents a noise-conditioned mixture-ofexperts framework that decomposes the feature space into specialized noise-aware subspaces for speaker verification. Specifically, we propose a noise-conditioned expert routing mechanism, a universal model based expert specialization strategy, and an SNR-decaying curriculum learning protocol, collectively improving model robustness and generalization under diverse noise conditions. The proposed method can automatically route inputs to expert networks based on noise information derived from the inputs, where each expert targets distinct noise characteristics while preserving speaker identity information. Comprehensive experiments demonstrate consistent superiority over baselines

Learning robust speaker representations under noisy conditions presents significant challenges, which requires careful handling of both discriminative and noise-invariant properties. In this work, we proposed an anchor-based stage-wise learning strategy for robust speaker representation learning. Specifically, our approach begins by training a base model to establish discriminative speaker boundaries, and then extract anchor embeddings from this model as stable references. Finally, a copy of the base model is fine-tuned on noisy inputs, regularized by enforcing proximity to their corresponding fixed anchor embeddings to preserve speaker identity under distortion. Experimental results suggest that this strategy offers advantages over conventional joint optimization, particularly in maintaining discrimination while improving noise robustness. The proposed method demonstrates consistent improvements across various noise conditions, potentially due to its ability to handle boundary stabilization and variation suppression separately.

This paper introduces FedSVD, a novel unsupervised federated learning framework for real-time anomaly detection in IoT networks. By leveraging Singular Value Decomposition (SVD) and optimization on the Grassmann manifolds, FedSVD enables accurate detection of both known and unknown intrusions without relying on labeled data or centralized data sharing. Tailored for deployment on low-power devices like the NVIDIA Jetson AGX Orin, the proposed method significantly reduces communication overhead and computational cost. Experimental results show that FedSVD achieves performance comparable to deep learning baselines while reducing inference latency by over 10x, making it suitable for latency-sensitive IoT applications.

Trajectory design in cislunar space under a high-fidelity ephemeris model (HFEM) is pursued through a nonlinear optimization perspective anchored on the transition of solutions from lower fidelity models, namely the Circular Restricted Three-Body Problem (CR3BP). The optimization problem is posed in the likeness of a multiple-shooting approach, aiming for segment-to-segment continuity while tracking proximity to the original CR3BP structures. The analysis of various formulations leads to the selection of an unconstrained least-squares problem for further investigation. The nonlinear optimization problem is convexified and the use of the Levenberg–Marquardt algorithm, as an alternative to the minimum-norm update equation found in most literature, is investigated for its control over the update step and inherent robustness. Additional techniques, such as adaptive weighting, are employed to further consolidate the behavior of the proposed algorithm in challenging scenarios. Numerical trials evaluate the adequacy of the methodology presented and compare it to the minimum-norm baseline over various application cases, including the generation of quasi-periodic trajectories and orbital transfers between them. The proposed technique is found to be a suitable alternative to the minimum-norm scheme, generally retaining better proximity to the original CR3BP trajectories and providing benefits in numerical robustness and stability. Moreover, the ease of including proximity objectives in a relaxed manner is shown to facilitate control over the shape of the final converged solution.

Short video streaming has become a dominant paradigm in digital media, characterized by rapid swiping interactions and diverse media content. A key technical challenge is designing an effective preloading strategy that dynamically selects and prioritizes download tasks from an evolving playlist, balancing Quality of Experience (QoE) and bandwidth efficiency under practical commercial constraints. However, real world analysis reveals critical limitations of existing approaches: (1) insufficient adaptation of download task sizes to dynamic conditions, and (2) watch time prediction models that are difficult to deploy reliably at scale. In this paper, we propose DeLoad, a novel preloading framework that addresses these issues by introducing dynamic task sizing and a practical, multi dimensional watch time estimation method. Additionally, a Deep Reinforcement Learning (DRL) enhanced agent is trained to optimize the download range decisions adaptively. Extensive evaluations conducted on an offline testing platform, leveraging massive real world network data, demonstrate that DeLoad achieves significant improvements in QoE metrics (34.4% to 87.4% gain). Furthermore, after deployment on a large scale commercial short video platform, DeLoad has increased overall user watch time by 0.09% while simultaneously reducing rebuffering events and 3.76% bandwidth consumption.

Language-audio joint representation learning frameworks typically depend on deterministic embeddings, assuming a one-to-one correspondence between audio and text. In real-world settings, however, the language-audio relationship is inherently many-to-many: one audio segment can be described by multiple captions and vice versa. To address this, we propose Probabilistic Language-Audio Pre-training (ProLAP), which models multiplicity as the spread of probability distributions in a joint language-audio embedding space. To train the intra-modal hierarchical relationship effectively, we also introduce two objectives: (i) hierarchical inclusion loss to promote semantic hierarchical understanding of inputs and (ii) mask repulsive loss to improve the efficiency of learning when optimizing the hierarchical inclusion loss. With this training strategy, our model can learn the hierarchical structure inherent in the data even from small datasets, in contrast to prior probabilistic approaches that rely on large-scale datasets. In our experiments, ProLAP outperforms existing deterministic approaches on audio-text retrieval tasks. Moreover, through experiments on the audio traversal task introduced in this paper, we demonstrate that ProLAP captures the plausible semantic hierarchy.

The increasing of digital radio frequency memory based electronic countermeasures poses a significant threat to the survivability and effectiveness of radar systems. These jammers can generate a multitude of deceptive false targets, overwhelming the radar's processing capabilities and masking targets. Consequently, the ability to robustly discriminate between true targets and complex jamming signals, especially in low signal-to-noise ratio (SNR) environments, is of importance. This paper introduces the attention-based dual-tree wavelet scattering prototypical network (AWSPNet), a deep learning framework designed for simultaneous radar target recognition and jamming suppression. The core of AWSPNet is the encoder that leverages the dual-tree complex wavelet transform to extract features that are inherently robust to noise and signal translations. These features are further refined by an attention mechanism and a pre-trained backbone network. To address the challenge of limited labeled data and enhance generalization, we employ a supervised contrastive learning strategy during the training phase. The classification is performed by a prototypical network, which is particularly effective in few-shot learning scenarios, enabling rapid adaptation to new signal types. We demonstrate the efficacy of our approach through extensive experiments. The results show that AWSPNet achieves 90.45% accuracy at -6 dB SNR. Furthermore, we provide a physical interpretation of the network's inner workings through t-SNE visualizations, which analyze the feature separability at different stages of the model. Finally, by integrating AWSPNet with a time-domain sliding window approach, we present a complete algorithm capable of not only identifying but also effectively suppressing various types of jamming, thereby validating its potential for practical application in complex electromagnetic environments.

Permanent Magnet Synchronous Motors (PMSMs) are widely employed in high-performance drive systems due to their high efficiency, power density, and precise dynamic behavior. However, nonlinearities, load disturbances, and parameter uncertainties present persistent challenges to control. Sliding-Mode Control (SMC) remains one of the most reliable strategies for high-performance PMSM drives. Yet, the rapid proliferation of adaptive, fractional-order, and intelligent variants has fragmented recent literature. This paper presents a comprehensive review and taxonomy of SMC-based PMSM speed-control methods published between 2020 and 2025. More than 200 studies are systematically analyzed and classified according to control order, surface design, disturbance-observer integration, optimization approach, and intelligent augmentation. Trends in publication activity, dominant hybrid structures, and application domains are quantitatively summarized. The review reveals a clear evolution from conventional discontinuous SMC toward adaptive, higher-order, and data-driven frameworks that mitigate chattering while preserving robustness. Persistent research gaps are identified in hardware validation, energy-efficiency assessment, and real-time tuning strategies. The taxonomy and critical synthesis provided herein establish a coherent reference for researchers and form the conceptual foundation for the companion paper (Part II), which delivers a unified benchmark and comparative simulation study of representative SMC designs.

Motivated by the application of using model predictive control (MPC) for motion planning of autonomous mobile robots, a form of output tracking MPC for non-holonomic systems and with non-convex constraints is studied. Although the advantages of using MPC for motion planning have been demonstrated in several papers, in most of the available fundamental literature on output tracking MPC it is assumed, often implicitly, that the model is holonomic and generally the state or output constraints must be convex. Thus, in application-oriented publications, empirical results dominate and the topic of proving completeness, in particular under which assumptions the target is always reached, has received comparatively little attention. To address this gap, we present a novel MPC formulation that guarantees convergence to the desired target under realistic assumptions, which can be verified in relevant real-world scenarios.

Conventional acoustic beamformers typically assume short-time stationarity and process frequency bins independently, ignoring inter-frequency correlations. This is suboptimal for almost-periodic noise sources such as engines, fans, and musical instruments: these signals are better modeled as (almost) cyclostationary (ACS) processes with statistically correlated spectral components. This paper introduces the cyclic minimum power distortionless response (cMPDR) beamformer, which extends the conventional MPDR to jointly exploit spatial and spectral correlations. Building on frequency-shifted (FRESH) filtering, it suppresses noise components that are coherent across harmonically related frequencies, reducing residual noise beyond what spatial filtering alone can achieve. To address inharmonicity, where partials deviate from exact integer multiples of a fundamental frequency, we estimate resonant frequencies from a periodogram and derive frequency shifts from their pairwise spacing. Theoretical analysis yields closed-form expressions for residual noise and proves that output power decreases monotonically with the number of cyclic components. Experiments on synthetic harmonic noise and real UAV motor recordings confirm these findings: in low-SNR scenarios, the cMPDR achieves up to 5 dB improvement in SI-SDR over the MPDR, yields consistent STOI gains, and remains effective with a single microphone. When spectral correlation is absent, the method reduces to conventional MPDR and does not degrade performance. These results suggest that cyclic processing is a viable direction for acoustic noise reduction that deserves further investigation. Code and audio samples are available at https://github.com/Screeen/cMPDR.

Quantification of protoporphyrin IX (PpIX) fluorescence in human brain tumours has the potential to significantly improve patient outcomes in neuro-oncology, but represents a formidable imaging challenge. Protoporphyrin is a biological molecule which interacts with the tissue micro-environment to form two photochemical states in glioma. Each exhibits markedly different quantum efficiencies, with distinct but overlapping emission spectra that also overlap with tissue autofluorescence. Fluorescence emission is known to be distorted by the intrinsic optical properties of tissue, coupled with marked intra-tumoural heterogeneity as a hallmark of glioma tumours. Existing quantitative fluorescence systems are developed and validated using simplified phantoms that do not simultaneously mimic the complex interactions between fluorophores and tissue optical properties or micro-environment. Consequently, existing systems risk introducing systematic errors into PpIX quantification when used in tissue. In this work, we introduce a novel pipeline for quantification of PpIX in glioma, which robustly differentiates both emission states from background autofluorescence without reliance on a priori spectral information, and accounts for variations in their quantum efficiency. Unmixed PpIX emission forms are then corrected for wavelength-dependent optical distortions and weighted for accurate quantification. Significantly, this pipeline is developed and validated using novel tissue-mimicking phantoms replicating the optical properties of glioma tissues and photochemical variability of PpIX fluorescence in glioma. Our workflow achieves strong correlation with ground-truth PpIX concentrations (R2 = 0.918+-0.002), demonstrating its potential for robust, quantitative PpIX fluorescence imaging in clinical settings.

Validation of autonomous driving systems requires a trade-off between test fidelity, cost, and scalability. While miniaturized hardware-in-the-loop (HIL) platforms have emerged as a promising solution, a systematic framework supporting rigorous quantitative analysis is generally lacking, limiting their value as scientific evaluation tools. To address this challenge, we propose MMRHP, a miniature mixed-reality HIL platform that elevates miniaturized testing from functional demonstration to rigorous, reproducible quantitative analysis. The core contributions are threefold. First, we propose a systematic three-phase testing process oriented toward the Safety of the Intended Functionality(SOTIF)standard, providing actionable guidance for identifying the performance limits and triggering conditions of otherwise correctly functioning systems. Second, we design and implement a HIL platform centered around a unified spatiotemporal measurement core to support this process, ensuring consistent and traceable quantification of physical motion and system timing. Finally, we demonstrate the effectiveness of this solution through comprehensive experiments. The platform itself was first validated, achieving a spatial accuracy of 10.27 mm RMSE and a stable closed-loop latency baseline of approximately 45 ms. Subsequently, an in-depth Autoware case study leveraged this validated platform to quantify its performance baseline and identify a critical performance cliff at an injected latency of 40 ms. This work shows that a structured process, combined with a platform offering a unified spatio-temporal benchmark, enables reproducible, interpretable, and quantitative closed-loop evaluation of autonomous driving systems.

This article addresses collaborative 3D map reconstruction using multiple drones. Achieving high-quality reconstruction requires capturing images of keypoints within the target scene from diverse viewing angles, and coverage control offers an effective framework to meet this requirement. Meanwhile, recent advances in real-time 3D reconstruction algorithms make it possible to render an evolving map during flight, enabling immediate feedback to guide drone motion. Building on this, we present Coverage-Recon, a novel coordinated image sampling algorithm that integrates online map feedback to improve reconstruction quality on-the-fly. In Coverage-Recon, the coordinated motion of drones is governed by a Quadratic Programming (QP)-based angle-aware coverage controller, which ensures multi-viewpoint image capture while enforcing safety constraints. The captured images are processed in real time by the NeuralRecon algorithm to generate an evolving 3D mesh. Mesh changes across the scene are interpreted as indicators of reconstruction uncertainty and serve as feedback to update the importance index of the coverage control as the map evolves. The effectiveness of Coverage-Recon is validated through simulation and experiments, demonstrating both qualitatively and quantitatively that incorporating online map feedback yields more complete and accurate 3D reconstructions than conventional methods. Project page: https://htnk-lab.github.io/coverage-recon/

As wireless communication systems evolve, automatic modulation recognition (AMR) plays a key role in improving spectrum efficiency, especially in cognitive radio systems. Traditional AMR methods face challenges in complex, noisy environments, particularly in low signal-to-noise ratio (SNR) conditions. This paper introduces MCANet (Multimodal Collaborative Attention Network), a multimodal deep learning framework designed to address these challenges. MCANet employs refined feature extraction and global modeling to support its fusion strategy.Experimental results across multiple benchmark datasets show that MCANet outperforms mainstream AMR models, offering better robustness in low-SNR conditions.

Controlling speaking style in text-to-speech (TTS) systems has become a growing focus in both academia and industry. While many existing approaches rely on reference audio to guide style generation, such methods are often impractical due to privacy concerns and limited accessibility. More recently, large language models (LLMs) have been used to control speaking style through natural language prompts; however, their high computational cost, lack of interpretability, and sensitivity to prompt phrasing limit their applicability in real-time and resource-constrained environments. In this work, we propose ParaStyleTTS, a lightweight and interpretable TTS framework that enables expressive style control from text prompts alone. ParaStyleTTS features a novel two-level style adaptation architecture that separates prosodic and paralinguistic speech style modeling. It allows fine-grained and robust control over factors such as emotion, gender, and age. Unlike LLM-based methods, ParaStyleTTS maintains consistent style realization across varied prompt formulations and is well-suited for real-world applications, including on-device and low-resource deployment. Experimental results show that ParaStyleTTS generates high-quality speech with performance comparable to state-of-the-art LLM-based systems while being 30x faster, using 8x fewer parameters, and requiring 2.5x less CUDA memory. Moreover, ParaStyleTTS exhibits superior robustness and controllability over paralinguistic speaking styles, providing a practical and efficient solution for style-controllable text-to-speech generation. Demo can be found at https://parastyletts.github.io/ParaStyleTTS_Demo/. Code can be found at https://github.com/haoweilou/ParaStyleTTS.

Distributionally robust optimization (DRO) is an effective framework for controlling real-world systems with various uncertainties, typically modeled using distributional uncertainty balls. However, DRO problems often involve infinitely many inequality constraints, rendering exact solutions computationally expensive. In this study, we propose a discrete DRO (DDRO) method that significantly simplifies the problem by reducing it to a single trivial constraint. Specifically, the proposed method utilizes two types of distributional uncertainty balls to reformulate the DDRO problem into a single-layer smooth convex program, significantly improving tractability. Furthermore, we provide practical guidance for selecting the appropriate ball sizes. The original DDRO problem is further reformulated into two optimization problems: one minimizing the mean and standard deviation, and the other minimizing the conditional value at risk (CVaR). These formulations account for the choice of ball sizes, thereby enhancing the practical applicability of the method. The proposed method was applied to a distributionally robust patrol-agent design problem, identifying a Pareto front in which the mean and standard deviation of the mean hitting time varied by up to 3% and 14%, respectively, while achieving a CVaR reduction of up to 13%.

Distributed resource allocation (DRA) is fundamental to modern networked systems, spanning applications from economic dispatch in smart grids to CPU scheduling in data centers. Conventional DRA approaches require reliable communication, yet real-world networks frequently suffer from link failures, packet drops, and communication delays due to environmental conditions, network congestion, and security threats. We introduce a novel resilient DRA algorithm that addresses these critical challenges, and our main contributions are as follows: (1) guaranteed constraint feasibility at all times, ensuring resource-demand balance even during algorithm termination or network disruption; (2) robust convergence despite sector-bound nonlinearities at nodes/links, accommodating practical constraints like quantization and saturation; and (3) optimal performance under merely uniformly-connected networks, eliminating the need for continuous connectivity. Unlike existing approaches that require persistent network connectivity and provide only asymptotic feasibility, our graph-theoretic solution leverages network percolation theory to maintain performance during intermittent disconnections. This makes it particularly valuable for mobile multi-agent systems where nodes frequently move out of communication range. Theoretical analysis and simulations demonstrate that our algorithm converges to optimal solutions despite heterogeneous time delays and substantial link failures, significantly advancing the reliability of distributed resource allocation in practical network environments.

This paper proposes Warshall-type algorithms for determining connectedness and clustering in an undirected matrix-weighted graphs. While a path between two vertices guarantees their connectedness in a scalar-weighted graph, the existence of one or more paths between them does not necessarily guarantee that they belong to the same cluster in a matrix-weighted graph. First, a sufficient condition for pairwise connectedness is established via aggregating path kernels between them. Second, we introduce three block matrix logic operators that enables the connectedness condition to be compactly represented and manipulated with positive semidefinite matrices. The proposed Warshall algorithm simultaneously determines connectivity between every pair of vertices in the graph and provides an approximated graph partition. Third, a distributed version of the Warshall algorithm is developed. Proofs of correctness, together with computational complexity analysis and numerical examples, are provided to establish the validity of the proposed algorithms.