Papers

Accurate short-term electricity demand forecasting is critical for reliable power system operation, energy market planning, and infrastructure optimization. This paper benchmarks ten machine learning models for daily electricity demand forecasting across the New England ISO (February 2020 - March 2023). The models span four families: tabular gradient-boosted trees (Random Forest, LightGBM, CatBoost, XGBoost), standalone neural architectures (LSTM, Transformer encoder), and hybrid Transformer+tree variants (Hybrid XGBoost, Hybrid LightGBM, Hybrid CatBoost, Hybrid RF). All models use meteorological data from six cities, calendar and holiday effects, autoregressive demand lags, and COVID-19 epidemiological variables. Hyperparameter optimization uses Optuna (300 trials, multivariate TPE, seed=42) under a leakage-free 70/15/15 chronological split. CatBoost achieves the best test performance: RMSE 8316 MWh, MAPE 1.87%, R-squared 0.917, followed by XGBoost (9066 MWh, R-squared 0.901), Hybrid CatBoost (9068 MWh, R-squared 0.901), and Hybrid XGBoost (9208 MWh, R-squared 0.898). Standalone neural architectures perform substantially worse (Transformer: 21294 MWh; LSTM: 22808 MWh), confirming the Transformer's role as a feature extractor rather than an end-to-end forecaster. An ablation on CatBoost shows that demand lags are the dominant predictor: removal degrades RMSE from 8316 to 11310 MWh (+36%), while weather and calendar features alone achieve an R-squared of 0.864. Removing COVID-19 features improves test RMSE by 1.7% while reducing training RMSE by 17.3%, a signature of temporal validity decay. SHAP analysis confirms this: 3 of 8 COVID features rank higher on the post-acute test set than during pandemic-active training, indicating the model over-applies stale pandemic patterns after behavioral adaptation was complete by August 2022.

Earth observation imagery plays a critical role in environmental monitoring, urban planning, disaster assessment, and climate analysis. While multi-spectral sensors are increasingly available, true-color (RGB) imagery remains widely used due to the power, cost, and deployment constraints of many satellite and aerial platforms. However, existing land-cover segmentation datasets are often limited in geographic coverage, scale, or public accessibility. To bridge this gap, we introduce BELDE (Building a Large-scale Earth-observation Land-cover Dataset for Europe), a publicly available dataset tailored for RGB-based remote sensing semantic segmentation. Constructed from Sentinel-2 true-color images and ESA WorldCover data annotations, BELDE contains 1,088,385 curated image-segmentation map pairs spanning Europe with 7 land-cover classes at 10 m spatial resolution, making it one of the largest publicly available RGB land-cover segmentation datasets for Earth observation. To facilitate cross-region generalization studies, we additionally introduce BELDE-K (16,607 pairs) covering the Republic of Korea and BELDE-CA-NV (88,155 pairs) covering California and Nevada in the United States. We establish baseline results using multiple semantic segmentation architectures and evaluate both in-domain and cross-domain performance. Models trained on BELDE achieve an F1 score of 83.0% on the European test set, while performance decreases to 66.4% on BELDE-CA-NV and 58.3% on BELDE-K, highlighting the challenges posed by out-of-distribution geographic domain shift. By providing a continental-scale RGB segmentation and evaluation benchmark, BELDE supports the development of robust and transferable Earth observation models. The dataset and benchmark resources will be publicly released.

Spectrum sharing forces radars to vary carrier frequency and bandwidth on a pulse-to-pulse basis within a coherent processing interval (CPI). While the resulting range-Doppler distortion is well-studied, the corresponding velocity estimation limit is not. We show that in the resolved-bin slow-time model of coherent agile-CPI processing, the effective Fisher information for radial velocity is the SNR-weighted energy of the carrier-time lever arm that survives projection out of the range and phase nuisance subspace. The carrier sequence thus sets the projection geometry, while the bandwidth sequence enters only through SNR weighting. Two consequences follow. First, the carrier sequence inflates the bound by a closed-form factor governed by the correlation between carrier offset and slow time: randomized or orthogonalized hops are nearly harmless, while ramp-correlated hops can severely degrade velocity information. Second, under matched filtering at equal pulse energy, the velocity Cramer-Rao bound (CRB) is invariant to the bandwidth sequence; a corollary recasts the output-SNR loss of agile-CPI mismatched filtering as a processing cost entering only through a per-pulse mismatch loss. The bound is verified against a brute-force Fisher matrix and Monte-Carlo maximum-likelihood estimation. The result yields a design principle: carrier hopping should be chosen not only for spectral coexistence but also to preserve the velocity-information residual.

High-altitude platform stations (HAPS), originally designed for communication services, can also provide structured signals of opportunity (SoOP) to augment the global navigation satellite system (GNSS). However, dense urban environments introduce severe blockage and non-line-of-sight (NLOS) conditions that undermine GNSS accuracy and render geometric placement metrics insufficient. To address this, we propose a metaheuristic framework for jointly optimizing the number and placement of HAPS under practical constraints by integrating high-fidelity 3D city models, ray-tracing, and multi-objective optimization to handle the discrete and highly non-convex design space. Three metaheuristic solutions based on distinct search principles are developed to efficiently explore the solution space, all demonstrating rapid convergence and consistently outperforming a greedy baseline, particularly in the low-to-moderate HAPS regime. For representative dense urban scenarios, we show that four HAPS are sufficient to satisfy an 18-m average 3D positioning error bound (PEB) threshold, while configurations with two to five HAPS achieve over 50% reduction in mean and root mean square (RMS) PEB and up to 94% and 87% reduction in standard deviation and coefficient of variation (CV), respectively, compared to the satellite-only case. Diminishing returns are observed beyond six HAPS due to geometric redundancy, emphasizing the importance of optimized placement. The framework further demonstrates strong robustness and generalizability across diverse urban environments with varying building morphology and propagation conditions, establishing it as an effective and scalable solution for HAPS-assisted localization in realistic urban settings.

Optimal frame-level quantization parameter (QP) allocation remains a persistent challenge in modern video encoders. The fixed-QP scheme widely adopted in practical systems is inherently content-agnostic, while classical Lagrangian rate-distortion optimization (RDO) methods often suffer from inaccurate multiplier settings. In this paper, we explore the use of large language models (LLMs) to automatically design RDO heuristics for frame-level QP adaptation. We construct a closed-loop evolutionary framework in which the LLM iteratively proposes RDO heuristics as algorithmic ideas with executable code, and these candidates are evaluated directly through encoding with the Fraunhofer Versatile Video Encoder (VVenC), where each heuristic acts as a scoring function that compares different QP choices based on the encoding statistics of past frames and current candidates. Experimental results across multiple test sets show that the evolved heuristic achieves promising rate-distortion improvements over both the fixed-QP scheme and the Lagrangian baseline. Further analysis reveals that the LLM can autonomously discover an adaptive heuristic that penalizes QP fluctuations via entropy-based terms, providing new insights into the design of RDO algorithms

We study the problem of system identification in heterogeneous settings, where different systems may follow distinct underlying dynamics. Existing clustered system identification approaches often rely on iterative training-based cluster assignment, which can be sensitive to learning uncertainty and model initialization. In contrast, we propose a one-shot, training-free clustering method that identifies similar systems using the structure of their locally observed data. Specifically, each system estimates a local state covariance matrix, and cluster identities are inferred by measuring the alignment between the leading covariance eigenspaces of different systems. We provide a mathematical interpretation of the proposed similarity score and develop a finite-sample analysis that characterizes how covariance estimation error induces eigenspace perturbations in terms of the underlying system dynamics. We then bound the probability of pairwise false eigenspace alignment and establish sufficient conditions for uniform inter-cluster eigenspace separation with high probability. Numerical experiments demonstrate that the proposed eigenspace-based clustering method effectively identifies systems with shared dynamics, leading to lower personalized model-estimation error compared with training-based clustering and non-clustered baselines.

Camera-first autonomous-driving models predict future ego waypoints from images, ego-state features, and route commands, but waypoint supervision alone does not explicitly supervise actor-level representations of nearby road users. We study this as supervised representation learning for open-loop waypoint prediction. The deployable model uses multi-view RGB, ego state, and route command at inference. During training, simulator-derived sidecar labels supervise actor grounding, privileged hindsight actor relevance relative to the logged ego trajectory, and selected-actor short-horizon motion; these labels are never inference inputs. We evaluate route-disjoint splits with matched architecture, optimizer, validation criterion, checkpoint selection, and three seeds. A plain waypoint-only RGB baseline obtains 1.815$\pm$0.02 m final displacement error (FDE), and the matched no-teacher non-sidecar RGB control obtains 1.716$\pm$0.02 m. Road-user sidecar supervision (RU-sidecar) reduces FDE to 1.223$\pm$0.01 m, a 32.6% reduction over the plain baseline and 28.7% over the matched no-teacher non-sidecar RGB control. It improves over the plain baseline on 1445/1494 routes and over the matched no-teacher non-sidecar RGB control on 1417/1494 routes. Actor-conditioned slices show gains in all nonempty subsets, including 29.1% reduction for samples with at least four valid sidecar actors and 30.0% when a vulnerable road user is present. Optional simulator-state teacher alignment reaches 1.186$\pm$0.15 m FDE, but higher seed variability makes it secondary. Non-deployable simulator-state diagnostics remain stronger, indicating a privileged-to-camera gap. The evidence is limited to open-loop simulation diagnostics.

Visual anomaly detection is a fundamental task in industrial automation. While existing approaches have achieved notable progress in identifying structural defects, the detection of logical anomalies remains relatively underexplored. In practice, structural and logical anomalies frequently co-occur in industrial workflows. Therefore, a solution capable of detecting both structural and logical anomalies is crucial for advancing comprehensive anomaly detection research. To address this limitation, we propose a unified framework, termed UniSLAD, which jointly addresses logical and structural anomalies without additional training, enabling a practical solution for dynamic industrial environments. First, we introduce a dual-feature extractor that synergistically integrates a Convolutional Neural Network (CNN) backbone for local texture perception with a Transformer backbone for global contextual reasoning, yielding richer and more comprehensive representations. Building on this foundation, we design dual-granularity feature representation modules. At the patch level, memory banks enhanced by the Mahalanobis Transform (MT) preserve representative features and support more discriminative anomaly scoring. At the image level, distribution maps are aggregated using Lower-Upper Mean (LUM) and Power Mean Pooling (PMP), yielding a more robust global representation than conventional average pooling. Extensive experiments on the two industrial benchmarks demonstrate that UniSLAD achieves competitive performance in comprehensive anomaly detection, achieving 99.4% and 93.1%, respectively. Furthermore, ablation studies verify the individual contributions and effectiveness of each proposed component.

Deep learning for 3D medical image segmentation requires extensive manual annotations, a major bottleneck in volumetric medical imaging. Active learning aims to reduce this burden by selecting informative samples for annotation, but most methods assume that an initial labeled set is already available. This leaves the cold-start problem largely unresolved: how to select the first volumes from a fully unlabeled pool before any task-specific model is trained. We propose CSCS, a Curriculum-Stratified Cold-Start framework that adapts initial sample selection to the structure of the unlabeled dataset. CSCS combines two self-supervised, label-free signals: local typicality, measuring representativeness in the embedding space, and reconstruction-based uncertainty, used as a proxy for sample difficulty. These signals are combined through a weighted geometric score, where the weighting is determined by a closed-form pacing rule based on the effective annotation budget and the Difficulty-Coverage Ratio, a pool-level statistic measuring the alignment between difficulty and representativeness. We evaluate CSCS on four 3D medical image segmentation benchmarks: BraTS, FeTA, Spleen, and an in-house fetal MRI dataset. Using nnU-Net as downstream segmentation model, CSCS shows consistently competitive performance across datasets and annotation budgets, with the strongest gains in low-to-mid annotation regimes. These results suggest that dataset-aware cold-start initialization can improve the robustness of active learning for 3D medical image segmentation by adapting sample selection to the geometry of the unlabeled pool.

Solving 3D medical inverse problems typically requires training dedicated supervised models for each specific task and measurement setting. To break this dependency, we present TF-PRDiT: a training-free conditional sampling framework that converts a frozen voxel-level 3D Diffusion Transformer prior into a versatile inverse medical problem solver. Building on the posterior-sampling view of diffusion inverse solvers, TF-PRDiT enforces measurement consistency during sampling via a task-specific forward operator rather than updating model weights, enabling a single pretrained prior to be reused across diverse conditional settings. Our method combines a predictor-corrector sampler with likelihood-based guidance on the denoised prediction, providing stable data-fidelity correction while preserving the underlying 3D anatomical prior. We highlight our framework's capability on the challenging task of X-ray-to-CT reconstruction by integrating a differentiable DRR projector to allow gradients to propagate directly from projection space back to voxels without any retraining. Experiments on LIDC-IDRI demonstrate that TF-PRDiT achieves strong reconstruction quality and uniquely scales to an arbitrary number of input X-rays (1-12) under a unified model, with performance improving consistently as additional views are provided. Beyond X-ray-to-CT, we show that simply swapping the forward operator extends the same frozen model to 3D super-resolution, volumetric infilling, and deblurring without any task-specific retraining, demonstrating that a single 3D diffusion prior can serve as a universal solver for volumetric medical inverse problems.

Industrial automation is being transformed by digitalization and the increasing use of cyber-physical systems. Modern production environments require greater adaptability, faster reconfiguration, and more intuitive human-machine interaction. However, traditional rule-based systems rely on fixed logic and cannot autonomously adapt to changing conditions. Consequently, current automation systems lack a systematic approach for integrating adaptive and generalizable reasoning capabilities for interpreting, planning, and executing user tasks across dynamic environments and heterogeneous components. This dissertation proposes a three-layer framework that integrates large language models (LLMs), digital twins, and automation systems into an autonomous system. Autonomy is defined as a design property assigned to system components and enabled through LLM-based reasoning to achieve adaptive, goal-oriented behavior. The Task-Process-Service-Resource (TPSR) model is introduced to transform user tasks into executable processes. Four LLM roles are identified: process orchestration, service matching, digital resource generation, and agent-as-a-service. Five peer-reviewed studies develop and refine these concepts using the design science research methodology. Case studies and prototypes demonstrate adaptive task planning, event-driven control, simulation-based parameterization, and digital model generation. Results show high task executability, command correctness, and content-generation accuracy while reducing manual effort. The framework enables the integration of LLM-based reasoning into industrial automation systems and improves adaptability and usability. Limitations include dependence on accurate digital representations, the computational demands of LLMs, and the need for human intervention in safety-critical situations.

Acoustic-to-articulatory inversion (AAI) remains challenging under domain shifts where changes in speaker attributes and cross-language conditions often degrade performance. We conduct a systematic evaluation under such shifts and establish baseline benchmarks on FROST-EMA, a Finnish-Russian bilingual EMA corpus. FROST-EMA addresses the English bias and limited speaker diversity of existing resources. We benchmark (i) articulatory targets (raw EMA coordinates vs tract variables), (ii) acoustic front-ends (MFCC vs SSL features), and (iii) inversion back-ends (BiLSTM vs a lightweight attention-based sequence model). We further define evaluation protocols for cross-gender transfer (within language) and cross-language transfer (within gender). The results indicate that cross-gender mismatch introduces moderate Pearson correlation declines (approximately 0.05 to 0.10) relative to the in-domain baseline, whereas cross-language mismatch causes larger drops (approximately 0.10 to 0.20).

Classifier guidance is a way to control diffusion generation by using a noise-conditioned classifier to steer the sampling process toward a target class. One drawback of classifier guidance is that it requires two separately trained models: a classifier and a diffusion model. We therefore study a more compact alternative in which a conventionally trained speech classifier is repurposed as the backbone for diffusion generation. Starting from a frozen noise-conditioned classifier in log-Mel space, we attach a lightweight subnetwork that reuses intermediate classifier representations and train only this subnetwork under a Denoising Score Matching objective. Our work shows that a pretrained classifier can be repurposed for conditional generation, providing an appealing bridge between discriminative modeling and conditional speech synthesis resulting in high speech quality within a single-backbone model, with reduced memory footprint and computational cost.

Pinching-antenna systems (PASs) can overcome signal blockage by repositioning dielectric radiating elements, called pinching antennas (PAs), along meter-scale waveguides to create line-of-sight links. Since each waveguide is driven by a single radio-frequency (RF) chain, non-orthogonal multiple access (NOMA) is well suited for PAS-based multi-user communications. This paper studies a PAS-enabled multi-user downlink NOMA system with multiple waveguides, each equipped with multiple PAs. The PA positions and base-station transmit precoding are jointly optimized to maximize the minimum user rate. The resulting problem is highly non-smooth and non-convex because of the rapidly oscillating coherent sums caused by inter-PA interference. To tackle this challenge, we propose a two-stage structured optimization framework. In the first stage, coarse PA-position and power-allocation optimization is performed using an interior-point algorithm while neglecting the PA channel phases, which gives solutions near the true optima. In the second stage, PA positions and transmit precoding are fine-tuned while accounting for the PA channel phase shifts. This stage first applies phase zeroing, where each PA is locally repositioned to align the corresponding channel phase toward zero and promote constructive coherent combining. It then uses an alternating procedure that iteratively performs forward-backward PA position refinement and successive-convex-approximation-based complex transmit precoding optimization until convergence, thereby reducing residual phase mismatch. Simulation results show that the proposed framework significantly outperforms heuristic optimization benchmarks with much lower computational time. They also demonstrate large gains over a comparable multiple-input multiple-output downlink NOMA system and reveal the impact of the number of PAs, users, and transmit power on system performance.

Real-time process monitoring requires methods that extract actionable information from high-dimensional time-series data. In this work, we present a new approach for process monitoring that combines tools of topological data analysis (TDA) and machine learning. In the proposed approach, we represent multivariate time-series data as manifolds and use topological descriptors to summarize the structure of such data; we then use a neural ordinary differential equation to learn the dynamic evolution of the topological structure of the system. Using real data from an industrial process, we show that this trajectory-based event detection approach is effective at detecting diverse types of events. We contrast this approach against reconstruction-based approaches such as principal component analysis and autoencoders and against a trajectory-based approach that uses Koopman autoencoders.

Integrated Sensing and Communication (ISAC) refers to the capability for the network to provide communications services whilst also being able to sense the environment in a scalable manner. One of the key functions of ISAC is the accurate localization of passive and mobile sensing targets. This paper introduces a novel hybrid TRP-UE sensing mechanism that improves network-based sensing performance. Evaluation results are provided using 3GPP-compliant ISAC channel models. The results demonstrate the significant benefit in complimenting TRP-based sensing with UE-assisted sensing in challenging propagation environments such as indoor factory.

Large Language Model (LLM)-based agents are increasingly being used to automate multi-step engineering work flows by interacting with software tools, interpreting intermediate results, and autonomously planning subsequent actions. Power system dynamic studies represent a particularly promising yet largely unexplored application domain for these agents. Unlike static computational tasks, dynamic studies often require more time on model parameter calibration, engineering judgment, and decision making under constrained action spaces. This paper introduces PowerAgentBench-Dyn, a benchmark designed to evaluate Agentic AI systems on power system dynamic-analysis tasks. The benchmark targets problems that cannot be reduced to a single optimization or coding task, but instead require a type of reasoning, tool usage, and iterative experimentation routinely performed by experienced power system engineers. The proposed framework includes two initial benchmark tasks. The first, the Dynamic Model Quality Review Benchmark, evaluates agents' ability to validate and diagnose dynamic models based on model-quality compliance criteria specified by system operators. The second, the Dynamic Security Risk Screening Benchmark, assesses agents' capability to leverage semantic memory and a limited simulation budget to identify, rank, and analyze the most critical short-circuit contingencies from an unseen fault dataset, as well as propose and evaluate possible mitigation measures. For each task, we define the simulation environment, observation and action spaces, and evaluation metrics. The benchmark is reproducible in a metric-based sense: released cases and simulator settings define a deterministic evaluator, while stochastic agent behavior is assessed over repeated runs using success rates and other metrics. The benchmark supports the development of future Agentic AI for power system operation and planning.

The performance of high-tech systems is often dictated by a few performance objectives shared among the many closed-loop controlled subsystems operating in the machine, such as synchronization, coordination, and alignment, which necessitates control methods that explicitly address them to achieve optimal performance. The aim of this paper is to introduce a framework that improves system performance through system-level controllers designed to be implemented as add-ons to the existing subsystem control structure. The developed method parametrizes all stabilizing system-level add-on controllers using the Youla framework, enabling a convex formulation of the sparse $\mathcal{H}_2$ synthesis problem. The result is a sparse add-on controller that achieves the optimal trade-off between combined performance and interconnection complexity, as demonstrated through numerical simulations.

Stuttering detection and classification using deep learning methods has the potential to improve the process of stuttering severity assessment. Most stuttering classification datasets provide clip-level labels, making them unsuitable for fine-grained frame-level classification needed to determine the duration of individual stuttering dysfluencies. To overcome this challenge, we present a multiple instance neural network architecture based on fine-tuned wav2vec 2.0, WavLM and Whisper encoders. We apply instance- and embedding-based multiple instance learning approaches to train models on a clip-level dataset for both clip-level and frame-level stuttering classification tasks. Our results show a 23% improvement in frame-level F1 score and between 2% and 9% in clip-level F1 score, demonstrating the ability of our models to utilize clip-level data for frame-level segmentation.

We study a data-driven control problem in the presence of arbitrary data poisoning attacks. We assume that a subset of offline output data is stored in unprotected locations and may be poisoned by an adversary. We first establish fundamental limitations for data-driven control arising from such poisoned data: poisoning attacks are not detected/identified from the dataset alone; unprotected data are non-informative for controller design with worst-case guarantees; and hard constraints on unprotected outputs are not certifiable. Motivated by these limitations and the data-enabled predictive control (DeePC) technique, we propose Secure DeePC, a data-driven control algorithm that is resilient against poisoning attacks. It first runs output-truncated DeePC using only the protected dataset until the online input becomes persistently exciting. It then uses online measurements to reconstruct the partial offline dataset, and finally returns to full-output DeePC. Secure DeePC achieves MPC-equivalent performance in finite time almost surely under certain conditions. Simulation results illustrate the efficacy of the proposed framework against poisoning attacks.

Recent flow-matching text-to-speech (TTS) models, such as F5-TTS, rely on a reference transcript at inference time, obtained from an external ASR system. This dependency makes zero-shot TTS brittle for accented or dysarthric speakers, precisely the scenarios where it is most needed. Moreover, we find that text-based reference conditioning can propagate atypical acoustic patterns from atypical speech into synthesis, even when ground-truth transcripts are available. To address this, we propose RTFree-F5, which replaces the reference transcript with continuous self-supervised speech representations mapped into F5-TTS's text-conditioning space via a lightweight adapter, while reusing the pretrained checkpoint. On dysarthric speech, RTFree-F5 reduces WER from 24.6% to 10.4%, surpassing even the ground-truth reference transcript baselines, while improving naturalness and remaining competitive on standard benchmarks without requiring any reference transcript.

This paper studies a free-space optical (FSO) link assisted by an optical reconfigurable intelligent surface (ORIS) and enhanced by a hybrid automatic repeat request (HARQ) scheme. The ORIS creates a virtual line-of-sight path around obstacles, while HARQ recovers frames corrupted by turbulence, pointing jitter, and geometric loss through retransmission and combining. We first derive a tractable statistical model for the end-to-end transmitter-ORIS-receiver (Tx-ORIS-Rx) reflected channel by jointly accounting for atmospheric turbulence, ORIS-induced pointing errors, and geometric attenuation. Building on these results, we obtain closed-form outage probability (OP) expressions for HARQ with Chase combining (HARQ-CC) and analytical outage upper bounds for HARQ with incremental redundancy (HARQ-IR), valid for an arbitrary maximum number of transmission rounds. We further conduct a high signal-to-noise ratio (SNR) analysis that provides a thorough characterization of the outage behavior and reveals the diversity order of both schemes. In addition, we characterize the delay behavior of the truncated HARQ process through the mean number of transmission rounds and the conditional mean number of rounds given successful decoding. Finally, numerical and Monte Carlo results validate the proposed analysis and show that HARQ substantially improves ORIS-assisted FSO reliability, with HARQ-IR achieving lower outage and delay than HARQ-CC, even for a small number of retransmission rounds.

This paper investigates the techno-economic viability of shared electric vehicle (EV) fleets for demand charge reduction under practical logistical and operational constraints. Unlike idealized models that overlook transit overheads, we propose a high-fidelity fleet management framework that explicitly accounts for the spatio-temporal coupling of energy consumption, labor costs for EV drivers, and battery degradation. We formulate the dispatch problem as a mixed-integer linear program (MILP) that jointly minimizes demand charges and total cost of ownership. To address the computational complexity arising from path-dependent constraints, we develop a marginal-value-based heuristic algorithm that achieves near-optimal performance with high computational efficiency. Using real-world data from San Francisco, our analysis reveals that a modest number of EVs can achieve significant demand charge savings, sufficient to recover the ownership and operational expenses. Our results also show how tariff structures, fleet size, and cost components influence overall profitability.

Existing mean opinion score (MOS) prediction models typically predict utterance-level naturalness MOS and can be insensitive to localized pitch-accent errors. We propose Pitch-Accent-focused Speech Quality Assessment (PASQA), which explicitly targets pitch-accent correctness. To train our model, we construct a controlled Japanese accent-error dataset by changing accent patterns using an accent-controllable text-to-speech system, and compute a pseudo accent-quality score from the accent-error rate. PASQA builds on self-supervised representations and employs mora-conditioned fusion, ranking loss, an auxiliary accent-error localization task, and speaker-invariant training. Experiments show that conventional models fail to preserve the ordering by accent-error severity, whereas PASQA achieves high ordering accuracy on both seen and unseen speakers. Further, PASQA shows stronger agreement with human accent-correctness judgments. The code is available at https://github.com/lycorp-jp/PASQA.

This paper presents a novel neural nonlinear control framework for a multi-drone slung payload system with variable-length cables and a rigid-body payload. The equations of motion are formulated into a decoupled structure, where the payload and cable length dynamics are governed by independent control channels, facilitating modularized controller design on reduced-order subsystems. A neural control contraction metric (CCM) controller and a neural feedback controller are jointly trained to enforce contraction conditions for the payload subsystem. Separately, a cable length control law is derived that exploits the variable-length degree of freedom for obstacle avoidance. Numerical simulations demonstrate trajectory tracking of a rigid-body payload and gate traversal capabilities of the overall system under the proposed control framework.

Generating high-resolution 3D CT volumes with fine details remains challenging due to substantial computational demands and optimization difficulties inherent to existing generative models. In this paper, we propose the Pixel-Level Residual Diffusion Transformer (PRDiT), a scalable generative framework that synthesizes high-quality 3D medical volumes directly at voxel-level. PRDiT introduces a two-stage training architecture comprising 1) a local denoiser in the form of an MLP-based blind estimator operating on overlapping 3D patches to separate low-frequency structures efficiently, and 2) a global residual diffusion transformer employing memory-efficient attention to model and refine high-frequency residuals across entire volumes. This coarse-to-fine modeling strategy simplifies optimization, enhances training stability, and effectively preserves subtle structures without the limitations of an autoencoder bottleneck. Extensive experiments conducted on the LIDC-IDRI and RAD-ChestCT datasets demonstrate that PRDiT consistently outperforms state-of-the-art models, such as HA-GAN, 3D LDM and WDM-3D, achieving significantly lower 3D FID, MMD and Wasserstein distance scores.

User-defined keyword spotting (UD-KWS) enables zero-shot wake-word detection from text, but existing systems learn speaker-invariant representations that cannot reject impostors uttering the correct keyword. We address this dual zero-shot setting – unseen keywords and unseen speakers – with ZP-KWS, a lightweight framework combining a phoneme-supervised audio encoder with a GE2E-pretrained compact speaker encoder (about 0.9M parameters). Multiplicative late fusion at inference grants each branch independent veto power, supporting modes from conventional detection to strict speaker-gated activation without retraining. On LibriPhrase, Google Speech Commands, and Qualcomm datasets, ZP-KWS reduces target-only FRR at 1% FAR by up to 60% relative to the strongest baseline while maintaining competitive keyword detection, all within a 1.55M parameter budget for edge deployment.

This paper explores the use of generative models to synthesize high-quality, site-specific multiple-input multiple-output (MIMO) channel data, addressing the high cost of the extensive measurement campaigns required to acquire real-world data for AI-native wireless networks. Two location-conditioned generative paradigms are compared: a conditional denoising diffusion implicit model (cDDIM), and a conditional flow matching model (cFMM). Both these models generate MIMO channel matrices conditioned on user coordinates, to preserve the spatial structure of the deployment site. The approaches are evaluated across three dimensions: statistical fidelity (including beam consistency and effective rank), generation efficiency, and utility in downstream tasks such as channel-state information compression and beam alignment. Results across diverse propagation scenarios (28 GHz and 3.5 GHz, both line-of-sight and non-line-of-sight) demonstrate that both models accurately capture site-specific characteristics, even when trained on scarce ground-truth data. Notably, cFMM achieves a quality comparable to cDDIM with roughly an order of magnitude less inference time. Augmenting scarce site-specific datasets with these synthetic channels yields hefty performance gains in downstream physical layer tasks compared to using scarce data alone or stochastic channels.

Burst suppression (BS) is a clinically relevant electroencephalographic (EEG) pattern used to monitor sedation depth and brain activity in critically ill patients, particularly during induced coma in Intensive Care Units (ICUs). Automatic burst detection remains challenging because BS patterns vary substantially between patients and annotated datasets are scarce. Recently, EEG Foundation Models (FMs) have shown promise across several downstream EEG applications, but their usefulness for BS detection remains unexplored. We present the first study to evaluate EEG FMs for burst detection in reduced-montage ICU EEG without patient-specific calibration. We compare REVE-base, LUNA-large and LuMamba-Tiny with an adaptive thresholding baseline and a task-specific EEGNet baseline. Additionally, we complement conventional EEG window-based classification with event-based burst detection evaluation. This helps assessing clinically whether burst episodes are correctly detected, reducing the impact of expected annotation variability. The best model, REVE-base, achieved the highest event-based F1-score ($0.868 \pm 0.167$) and reduced burst-per-minute error by 52.1% and 36.2% compared to EEGNet and adaptive thresholding respectively, supporting FMs for scalable EEG monitoring in ICU. Ablation experiments showed that full fine-tuning was the most effective adaptation strategy with respect to frozen-backbone training, two-step fine-tuning, and LoRA-based adaptation, improving event-based F1-score over frozen-backbone training by up to $+0.102$ for LUNA-large. With reduced labeled datasets, pretrained REVE-base outperformed random initialization by $+0.723$ event-based F1 points at 25% of the cohort, demonstrating the benefit of pretraining FM representations when adapted to burst detection with limited labeled data.

Large-scale power outages are typically caused by cascading failures. These unfold dynamically through complex interactions between network dynamics and individual component failures. In contrast, the study of cascading failures in physics has focused on analyzing line overloads in the quasi-static regime. We introduce a new model that integrates the dynamics of node and line failures with a paradigmatic oscillator model for power grid synchronization. This enables us to investigate the collective cascading behavior of coupled failures for the first time. We study the impact of nodal robustness, the ability of nodes to tolerate transient disturbances, and inertia, the ability of nodes to resist frequency deviations, on cascade sizes. We discover two novel mechanisms driving system fragility: i) While low inertia is widely considered a major challenge for power grids, we find that high inertia can amplify cascade sizes unless accompanied by appropriate adjustments of other dynamical properties. ii) Further, we find that an increase in the robustness of individual nodes can paradoxically lead to larger cascades. This latter phenomenon constitutes a novel type of Braess's paradox. Understanding such counterintuitive collective effects may become central for achieving resilient future power grids.