Papers

The ability to synthesize realistic X-ray images has catalyzed the development of AI models for X-ray image-guided procedures, which otherwise suffer from a lack of available annotated data. Prior work has demonstrated the effectiveness of mechanistic simulation of digitally reconstructed radiographs (DRRs) as a training data source for a myriad of tasks, including segmentation and anatomical landmark detection, with comparable or superior performance to real data training. However, mechanistic DRR synthesis still relies on the availability of annotated high-resolution anatomical models. Deriving these from CT images of real patients or specimens imposes an undesirable bottleneck on data quantity and variability. In this work, we explore two methods for synthesizing training data: (1) a 3D conditional latent diffusion model that generates CT volumes to use as inputs for mechanistic DRR generation without real, 3D anatomical models, and (2) a view-conditioned 2D diffusion model that produces synthetic X-rays. In controlled experiments, we demonstrate that synthetic 2D diffusion-based X-rays can be used to train an anatomical landmark detection model that generalized to real X-ray images with performance rivaling that of a model trained on real X-ray images. Thus, we provide preliminary evidence that synthetic, 2D diffusion-based training data can substitute for real X-ray data, identifying a promising avenue towards generating large, diverse datasets for training robust AI models in interventional X-ray imaging.

Benchmarking is critical for the systematic evaluation of machine learning systems. While several open-source datasets are available for Hindi, existing benchmarks remain limited in terms of modality, geographic diversity, demographic representation, and transcription robustness. We introduce an inclusive, multimodal Hindi benchmark dataset collected from 102 districts across India. The dataset consists of spontaneous speech elicited using image prompts and recorded under real-world acoustic conditions across diverse demographic groups. Each audio segment is associated with the image prompt that elicited it and is annotated with three independent transcriptions, enabling multi-reference evaluation that accounts for permissible orthographic and lexical variations. We show that single-reference evaluation overstates errors, while multi-reference evaluation provides a more robust, inclusive, and realistic assessment of automatic speech recognition (ASR) systems. The pairing of each utterance with its eliciting image further enables image retrieval evaluation using both speech and text queries. We evaluate multiple multimodal embedding models for image retrieval and report their performance on the combined speech–image and text–image retrieval tasks. The results reveal a performance gap between text-based and audio-based retrieval, with text-based retrieval consistently achieving higher performance.

Geometric misalignments between the nominal and true acquisition parameters in tomography degrade reconstructions. We propose a framework that jointly reconstructs the volume and calibrates the acquisition geometry for arbitrary source–detector configurations. The core of our framework is an x-ray transform operator whose gradients with respect to the acquisition geometry can be efficiently computed with a ray-tracing method of structure and computational complexity similar to those of the forward operator. We represent the volume in a B-spline basis to provide a continuously differentiable model. This results in a better-behaved optimization landscape compared to voxel-based representations. We validate our framework with CT, micro-CT, nano-CT, and positron emission tomography data under a variety of geometric misalignments.

The Manakov equation governs the propagation of signals in dual-polarization systems. Its solution is usually approximated by regular perturbation on the nonlinear Kerr parameter. In this paper, we propose a novel regular perturbation on the group-velocity dispersion parameter for the Manakov equation.

This work examines sexualised AI-generated English-speaking voices offered by a popular commercial platform. New technologies may enable sexual empowerment and greater diversity in gender expression, yet toxic masculinity, heteronormativity, and the abuse of women and LGBTQ+ people remain pervasive online. Drawing on a Feminist HCI perspective, we examine how commercial voice AI systems reproduce and circulate particular performances of gender. We conducted a listening experiment with a diverse group of listeners, combining quantitative adjective selection, qualitative free-text responses, and acoustic analysis. Participants evaluated male- and female-coded voices presented with either sexualised scripts or neutral text. Results reveal a narrow range of gender expression, largely binary and heteronormative. Female-coded voices are more frequently described using sexualised and submissive terms, while male-coded voices are more often associated with dominance and positive traits.

Voice reconstruction using Text-to-Speech (TTS) offers a communication method for people with speech disorders, which aims to retain their speaker identity while improving intelligibility. Previous work generally relies on Mean Opinion Score (MOS) to evaluate naturalness and speaker similarity, but this has limited sensitivity and reliability. We propose an evaluation framework with subjective and objective components. Subjectively, we evaluate perceived intelligibility and speaker identity using Best Worst Scaling (BWS) with situational framing. Objectively, we demonstrate that standard measures fail to predict reconstruction success for highly unintelligible speakers, so we introduce a novel dual-reference distributional measure to assess the trade-off between intelligibility and speaker identity. By evaluating the output of 17 zero-shot TTS systems for 193 speakers, we show that our framework provides a reliable and task-aligned approach for assessing voice reconstruction.

We present an ML-based vessel detection and localization system, trained with weak supervision from imperfect AIS labels, that achieves a 97.8% detection rate at 1.98% false-trigger rate, successfully identifies dark-vessel events from unlabeled data.

In electromyogram (EMG)-based motion recognition, it is impractical to predefine all motions that may be required during deployment, necessitating class-incremental learning that sequentially adds new motion classes. The primary challenges in class-incremental learning are catastrophic forgetting, where previously acquired knowledge is overwritten when learning new classes, and the memory cost of retaining past data to counteract it. In particular, for EMG-based motion recognition intended for edge devices with limited computational resources, it is essential to suppress catastrophic forgetting and maintain low memory cost. In this paper, we conducted a comparative evaluation of eight class-incremental learning methods spanning generative and discriminative approaches, including both deep and non-deep learning methods, for EMG signal classification. Using four datasets, we evaluated each method in terms of classification accuracy, backward transfer, and memory cost. The results demonstrated that deep learning-based methods suffered significant accuracy degradation from catastrophic forgetting as the number of tasks increased, whereas generative models maintained stable accuracy with low memory cost. Among generative models, the scale mixture classification model (SMCM), which captures EMG signal variability, achieved the most favorable accuracy-memory trade-off while effectively suppressing catastrophic forgetting across all datasets.

Differentiable audio processors are habitually designed and optimised in machine-learning frameworks, but deploying them as real-time audio effects still often requires non-automatic implementation in a dedicated digital signal processing language. The translation is error-prone, demands an onerous verification process, and detaches research prototypes from usable production tools. That being so, we present ADAC, a compiler that lowers a trained model to a framework-agnostic intermediate representation and emits efficient FAUST code whose impulse response matches the source model to within floating-point arithmetic noise, direct paths included. The optimisation loop is made audible by replacing the model in a running plugin after each gradient step. The exported processor carries a small set of macro-controls that leave its stability intact. A stability certificate computed from the shipped parameters is checked before the plugin is built. At the demonstration, a feedback delay network is trained and exported to a working plugin.

Pixel antennas enable highly radiation pattern reconfigurability to enhance wireless systems, but its antenna coding design, that is optimizing the states of switches embedded in pixel antennas, remains an NP-hard challenge. Conventional approaches for antenna coding design typically rely on heuristic search algorithms, which suffer from high computational complexity. To overcome this issue, we propose a novel efficient data-free optimization algorithm called physics-informed neural optimizer (PINO) for antenna coding design. By integrating a deep convolutional neural network prior and a Gumbel-Sigmoid continuous relaxation into a differentiable physics engine, the proposed algorithm transforms the binary optimization problem into a continuous differentiable problem, which enables the antenna coding optimization problem to be efficiently solved via gradient descent. Simulation results demonstrate that the proposed algorithm outperforms the heuristic search based algorithms, reducing computational time while achieving higher average channel gain.

As expressive text-to-speech (TTS) and voice conversion (VC) systems increasingly generate non-verbal vocalizations (NVVs) to enhance naturalness, reliable speaker verification (SV) becomes essential to objectively assess identity consistency across both verbal and non-verbal segments. Yet current SV systems generalize poorly to NVVs, and fine-tuning on NVV data causes catastrophic forgetting of speech performance. We present the first systematic study across 10 NVV types and propose a framework combining frozen Data2Vec self-supervised features with ECAPA-TDNN, enhanced by a Mixture of Experts (MoE) module with learned domain-aware routing. A conditional distillation loss on speech inputs via a pretrained teacher retains speech-to-speech accuracy, while a contrastive loss bridges the speech-NVV domain gap. Our method reduces speech-NVV EER from 38.93% to 22.66% over a pretrained baseline, and improves speech EER from 13.17% to 9.24% via distillation.

Leveraging their flexible and efficient deployment capabilities, unmanned aerial vehicle (UAV) swarms have been widely applied in various mission scenarios. However, the open communication environment also exposes them to the threat of Byzantine attacks. Most existing studies assume independent decision-making by each UAV, neglecting that local conformity amplifies false information propagation. This paper constructs an evolutionary game model for UAV swarm under malicious attacks based on graph evolutionary game theory, revealing how local conformity rules govern the spread of deceptive strategies. Using death-birth updating rules, we derive the macroscopic dynamic equation for the fraction of deceptive strategies and the analytical solutions to its evolutionary stable states. Sim ulations reveal observation errors weaken malicious induction, while higher proportions of malicious nodes and greater attack intensity drastically amplify attack impacts. Moreover, the model exhibits strong topological robustness across regular, scale-free and random networks.

In this paper, we present a fully discrete approach for the accurate and numerically efficient dynamical modeling and state estimation of continuum robots. The model is based on geometrically exact beams in a minimal, strain-based formulation and derived in the framework of Lie group variational integrators, allowing to preserve important geometric properties that we exploit to achieve high accuracy and numerical efficiency. We then propose a disturbance observer based on an extended Kalman filter formulation that reliably estimates system states as well as model uncertainties and external disturbances. Experiments on a real system validate the accuracy and efficiency of the proposed model and observer.

We introduce a semi-parametric framework for nonlinear system identification, which decouples discrepancy functions from physics-based components. Orthogonal Gaussian process regression balances sparse parameter selection (the white box) with discrepancy learning (the black box) to produce interpretable models from incomplete physics.

We demonstrate the first KAN-based DPD model for 5G analog RoF fronthaul link, achieving a 24.2% lower EVM than multi-layer perceptron and 29.6% lower than Volterra-based GMP at equivalent Bit Operations. To attain an EVM below 2%, KAN requires ~52% fewer BOPs than a perceptron.

Segmenting the temporomandibular joint (TMJ) disc from MRI is essential for accurate diagnosis of internal derangement, yet it remains unreliable in practice due to its small size, low contrast, and morphological variability. Existing methods, primarily adapted from general segmentation architectures, often produce fragmented or anatomically inconsistent masks, leading to unstable measurements of disc position and shape for downstream diagnosis. To address these challenges, we propose TISC, a TMJ disc segmentation framework that integrates semantic anchoring with clinical metadata-guided boundary refinement. The framework first establishes robust disc localization in the foundation model feature space via a Prototypical Semantic Anchoring (PSA) module that aggregates adjacent-slice MedDINOv3 features and derives a prototype-driven similarity map. It then performs targeted boundary refinement through a Clinical-Metadata Point Refinement (C-MPR) module, with point-wise predictions modulated by Mouth Open Limitation (MOL), a clinical indicator associated with disc displacement without reduction. On a large-scale cohort of 2,488 PD MRI volumes from 1,300 patients, our method achieves up to a 4.96 Dice improvement over strong baselines across diverse architectures, delivering more anatomically coherent and clinically reliable TMJ disc segmentation.

Speaker-decoupled speech codecs can reduce bitrate by separating global speaker attributes from local content and prosody, while supporting voice conversion. Existing speaker-decoupled codecs face a trade-off: methods that explicitly suppress speaker leakage often rely on multi-stage or auxiliary training, whereas simpler designs can leave residual speaker information in local tokens. We propose SDP-Codec, a speaker-decoupled, pitch-injected codec trained with a single-stage optimization pipeline. SDP-Codec derives local tokens from continuous pre-quantization features of a pretrained self-supervised encoder and injects normalized F0 via a pitch encoder-decoder with global-conditioned denormalization and soft-label pitch reconstruction objective. Across 16 kHz and 24 kHz settings, SDP-Codec achieves competitive reconstruction and strong zero-shot voice conversion at comparable bitrates, with the lowest speaker-probing accuracy among compared systems, suggesting reduced speaker leakage.

The Virginia Tech Transportation Safety Index (VTTSI) is a real-time, cloud-native framework for quantifying intersection safety using multimodal connected-vehicle telemetry and multi-year VDOT crash history. Traditional crash-based methods rely on lagged, aggregated data and cannot reflect rapidly changing operational conditions. VTTSI addresses this gap through a hybrid modeling approach that fuses Empirical Bayes (EB) crash stabilization, uplift factors derived from speed and conflict behavior, and a CRITIC-weighted multi-criteria decision-making (MCDM) module combining SAW, EDAS, and CODAS. The system produces interpretable, exposure-adjusted safety scores on a 0–100 scale every 15 minutes. A cloud-deployed architecture built on FastAPI, PostgreSQL, PostGIS, and Streamlit supports interactive visualization of traffic volumes, VRU exposure, speed variance, and real-time incident activity. Validation across intersections demonstrates coherent diurnal patterns, consistency among MCDM methods, and sensitivity to observable operational turbulence. Sensitivity analysis further shows that the RT–SI is robust to parameter perturbations, with deviations typically remaining below one point on the 0–100 scale. By integrating long-term crash risk with short-term behavioral dynamics, VTTSI provides a transparent, adaptive, and proactive safety-monitoring framework suitable for transportation agencies, traffic management centers, fleet operators, and autonomous vehicle systems.%~\cite{persaud2007, montella2020systemic, schultz2025_surrogate, Amraji2025CombinedSafetyIndex}.

Remote photoplethysmography (rPPG) is a camera-based technique for measuring physiological signals, particularly cardiac activity. From the remotely measured signals, heart rate can be estimated, which is crucial for health monitoring. In this study, we investigate a driver health monitoring system based on remote heart rate estimation. However, driving environments represent uncontrolled settings where videos are subject to varying illumination conditions and frequent head movements. We introduce MS-rPPG, a multi-spectral framework that combines RGB with near-infrared (NIR) face video to alleviate rPPG estimation under challenging driving conditions. To combine the complementary features from two spectral videos, we propose a cross-spectral linear modulation (CSLM) strategy based on frequency-domain analysis. Moreover, we introduce MS-Mamba, a novel state space model designed to effectively model long-range temporal dependencies while jointly capturing cross-channel interactions between multi-spectral features. We collected a real-world dataset called MS-Drive, which was recorded from 50 participants while driving the vehicle. The proposed method was evaluated on the MR-NIRP Car dataset and MS-Drive datasets. The experimental results indicate that MS-rPPG shows better robustness and heart rate estimation accuracy than previous methods, highlighting its promise for driver health monitoring. The codes are available at github.com/ziiho08/MS-rPPG.

The rapid expansion of artificial intelligence (AI) has driven unprecedented growth in data center electricity demand. The scale and pace of this load growth carry significant implications for the sustainability of electric power systems. On the one hand, rapid, spatially concentrated data center load growth is outpacing clean energy deployment in several major regions, raising emissions and challenging both grid flexibility and reliability. On the other hand, this fast-developing and capital-intensive sector offers abundant opportunities to advance sustainability through clean energy integration and operational innovations. This article provides an overview of the mechanisms through which data center affect power system sustainability, underscoring both risks and the potential. Specifically, this article (i) characterizes AI data center load behavior and categorizes electricity supply configurations by function and sustainability profile, as well as situates these loads within global and regional electricity demand trends; (ii) analyzes sustainability impacts across short-run operational and long-run planning mechanisms, evaluates effects on grid carbon emissions and renewable energy utilization, and feasibility of offering system flexibility and participating in ancillary service; and (iii) evaluates real-world corporate sustainability pathways and highlighting both the system benefits and feasibility limits of current carbon accounting practices. The goal of this work is to synthesize existing knowledge and technological developments and to guide research and development toward a more sustainable integration of AI data centers and electric power systems.

Image compression for machines calls for a unified codec that serves multiple downstream vision tasks. Existing approaches either adopt task-specific end-to-end designs, raising parameter and deployment overhead, or rely on transfer-based adaptations that remain externally attached and heuristic task design. A key limitation shared by both lines of work is their largely static computation pattern, which applies similar transformations across tokens despite the fact that different image regions exhibit markedly different semantic importance and complexity for machine perception. We propose MoECodec, a token-aware image compression framework that supports multiple downstream tasks within a single model. MoECodec replaces the FFN layers in transformer-based compression model token-wise Mixture-of-Experts (MoE), enabling dynamic, token-level computation conditioned on the input content and task objective. To make MoE effective in compression model, we introduce a stable routing strategy that combines expert-choice routing with spatial total variation regularization to encourage spatially coherent assignments, and we propose a lightweight expert architecture, Group Shuffle MLP (GShMLP), to control parameter growth. Extensive experiments show consistent improvement against baselines on both conventional image reconstruction and machine tasks.

Multiparametric programming offers a powerful solution to the computational burden of solving optimal control problems repeatedly online. By solving the problem once offline, it yields the optimal control laws as explicit closed-form functions of the initial system state, reducing online execution to a direct evaluation with no iterations required. Most existing works build this idea on a discrete-time foundation, slicing the time horizon into intervals and applying KKT conditions to the resulting algebraic system. This discretization forces a tradeoff: too few intervals and the model fails to capture the true system dynamics, while too many cause the problem size, the number of decision variables, and the number of critical regions to grow rapidly, making both offline preparation and online lookup increasingly expensive. This work develops a multiparametric framework that works directly with the continuous-time problem. Pontryagin's Maximum Principle (PMP) is applied without any model discretization, and the optimal control is recovered as an explicit function of the initial state. Compared to the discrete-time formulation, the continuous-time approach produces substantially fewer critical regions, and this number remains fixed regardless of accuracy requirements, since it reflects the structure of the problem itself rather than a discretization grid. The framework also yields the switching times as explicit functions of the initial state, directly exposing when and how the optimal control structure changes over the horizon. Knowing these switching times in advance allows the real-time controller to skip unnecessary computations between them, further reducing the online execution cost. Results from a PAROC framework case study demonstrate that the continuous-time multiparametric approach is a rigorous alternative to the conventional discrete-time formulations.

Diffusion-based image compression has achieved strong perceptual quality at ultra-low bitrates. However, existing codecs are often tied to specific backbones and specialized components, making diverse, rapidly evolving generative models difficult to reuse. This raises a natural question: Can modern generative foundation models be connected to image compression through a simple and extensible interface? Two insights guide our design: stronger generative priors make a simpler codec interface viable, and generation and compression can be intrinsically linked through latent transport. We therefore propose Gen2-IC with two stages: (1) Latent Compression maps clean image latents to entropy-constrained latents; and (2) Latent Transport refines them with one near-terminal update based on the pretrained model. Gen2-IC requires neither auxiliary conditioning signals nor task-specific backbone modifications. With lightweight adaptation and no distillation, it supports fast encoding and one-step decoding across multiple bitrates. We validate Gen2-IC on SD-2.1, SANA-1.5, FLUX.1-dev, and Qwen-Image-2512, spanning U-Net and Transformer architectures as well as diffusion and flow-matching formulations. With stronger priors, Gen2-IC delivers gains below 0.05 bpp: the Qwen variant leads diffusion-based generative codecs in reconstruction fidelity (PSNR), perceptual similarity (LPIPS and DISTS), and recognizer-based semantic fidelity (OCR CER/WER and face-ROI similarity) across four benchmarks.

The Unlimited Sampling Framework (USF) overcomes the traditional trade-off between dynamic range and digital resolution, achieving performance unattainable with standard ADCs. Its multi-channel extension (MC-USF) enables reconstruction from multiple folded measurements at critical sampling rates. Existing MC-USF methods typically rely on Chinese Remainder Theorem (CRT)-based unfolding, which requires strict channel-level sampling synchronization and is therefore vulnerable to timing mismatch, jitter, and drift. This paper introduces an asynchronous MC-USF architecture that eliminates the need for synchronization. By viewing spatial-temporal signal lifting as inducing smoothness over a graph of sensing channels, we develop a reconstruction strategy robust to temporal misalignment. Numerical experiments validate the approach, demonstrating accurate recovery and enabling more practical multi-channel USF implementations.

Poultry production operates under tightly coupled environmental and biological dynamics, yet commercial climate control remains largely heuristic, limiting welfare assurance and operational efficiency. We introduce an edge-cloud digital twin framework for real-time, welfare-constrained environmental control in poultry facilities. The framework integrates distributed sensing, on-device state estimation, a hybrid physics-data model, and model predictive control to enable anticipatory and adaptive management under practical farm constraints. A grey-box thermodynamic and mass-balance formulation is augmented with a learned residual that captures unmodeled biological variability, including activity-dependent metabolic heat. This hybrid model is embedded within a state-space representation for real-time estimation and control at the edge, while cloud coordination supports cross-farm learning and long-horizon optimization. Bandwidth-aware processing and asynchronous synchronization enable deployment in connectivity-limited environments. Evaluation in a high-fidelity broiler production testbed demonstrates substantial gains over rule-based control and physics-only modeling. Temperature prediction error is reduced from 1.8 degrees Celsius to 0.4 degrees Celsius, ammonia constraint violations decrease by 90 percent, and communication requirements are lowered approximately 30-fold through edge-first processing. A Domain Transfer Score of 0.92 further indicates strong robustness across facility conditions. These results show that physically grounded digital twins, coupled with real-time control, enable scalable and welfare-aware management of biological production systems.

We study the convergence and acceleration of Qvalue iteration (QVI) with heavy-ball momentum. Although acceleration of value iteration has been studied extensively, there has been less work on the acceleration of heavy-ball QVI for control tasks. We analyze heavy-ball QVI from the viewpoint of switched linear system (SLS) theory and the joint spectral radius (JSR). First, we convert heavy-ball QVI into an exact SLS and interpret its convergence through the JSR. We also give JSR-based conditions under which heavy-ball QVI can be faster than standard QVI.

Sleep spindle is a physiologically significant biomedical signal in electroencephalographic (EEG) waveforms, which is typically a low-amplitude event in sleep. Due to the small signal ratio in the overall EEG, previous detection methods have limited capability to capture its start and end points and lack flexibility in handling multi-spindle scenarios. To address the gap, we address the problem from a new perspective and introduce SpindleFlexNet, the first framework in this field to apply deep learning-based one-dimensional object detection, leveraging an adapted one-dimensional RetinaNet architecture. The framework employs one-dimensional anchor generation, matching, and regression, along with a customized one-dimensional loss function. Analyses were conducted on two public datasets: the Montreal Archive of Sleep Studies and DREAMS, from which a total of 11,061 and 335 segments were obtained, respectively. When trained on these datasets, SpindleFlexNet achieved an average recall, precision, and F1-score of 0.61, 0.76, 0.67, and 0.58, 0.80, 0.67 in five-fold cross-validation. The model demonstrates stable detection performance and good generalization, making it a practical tool for sleep research. Potential applications include automated spindle labeling in clinical settings and as a reference for studies combining EEG with simultaneous functional magnetic resonance imaging.

This paper presents a modular S-/X-band shared-aperture phased-array antenna (SAPAA) for satellite-communication ground-station reception. The proposed architecture uses a repeatable unit cell that supports independent S- and X-band operation within the same physical aperture and enables arbitrary aperture scaling. Dual-polarized radiators are combined with calibrated complex receive coefficients to synthesize linear polarization (LP), right-hand circular polarization (RHCP), and left-hand circular polarization (LHCP). The design burden of the electrically large shared aperture is reduced by using theoretical estimates for scan matching and inter-band isolation before full shared-aperture verification. Simulated and measured results demonstrate axial ratios below 0.1 dB in the target S- and X-band receiving bands over a +/-50 deg scan range. The prototypes are validated using two approaches: passive measurements, where the element responses are measured individually, and RF system-on-chip-based active measurements, where all available receive channels are measured simultaneously. The results confirm that the proposed SAPAA provides wide-angle scanning, very high polarization purity, and polarization-reconfigurable operation for multi-mission SATCOM ground terminals.

Policies must operate across diverse conditions, yet a single policy is often conservative while fully adaptive schemes can be complex. We study zero-shot generalization in contextual dynamical systems and introduce a performance-centric, directional task dissimilarity–the signed divergence–that upper bounds the generalization gap from a source context to a target context. The signed divergence induces $\varepsilon$-tolerance sets that certify when a source policy class generalizes, and it yields a concrete notion of task-space complexity: the minimum number of source contexts needed so that every target context incurs at most $\varepsilon$ generalization gap. Under a mild local smoothness assumption on performance, the induced tolerance sets admit certified inner/outer balls and instance-dependent volume bounds on task-space complexity. In the finite-oracle setting, source selection reduces to set cover; a greedy strategy inherits the standard $H(n)$ approximation guarantee. Using a Mass-Spring-Damper system with linear-quadratic regulator (LQR) controllers and a nonlinear CartPole system with deep reinforcement learning controllers, we show that greedy selection achieves the same $\varepsilon$-coverage with fewer policies than uniform or random baselines. Our approach delivers a performance-based task similarity measure and practical certificates for building generalizable control with simple policies.

Executable evaluation – checking the consequences of an agent's actions with a program rather than grading its prose – has become a prominent way to assess tool-using AI agents in software settings. Electric power engineering has not yet had an analogous benchmark: language-model use is still dominated by retrieval and text question answering, while agents acting on power-system artifacts remain mostly academic prototypes. We introduce the Power Systems Agent Benchmark, an executable benchmark for power-engineering agents. An agent receives a structured task and returns a structured solution; a deterministic evaluator recomputes the engineering quantities, checks operational constraints, and returns a feasibility flag, a normalized score, and explicit violations. The benchmark contains 41 task families across eight areas of power engineering, from power flow and protection to stability, microgrids, reliability, power quality, and forecasting. Each task is grounded in a citable source, standard, or documented engineering formulation. To resist contamination, held-out cases are synthesized on demand by per-family generators from private seeds: the construction is inspectable, but the instances remain private. In a reference evaluation with three command-line agents, the strongest score near the compact tier's ceiling, a smaller open model trails, and public and held-out performance are broadly consistent; a separate public-split grid with OpenCode and Aider probes harness effects. The reference evaluation doubles as quality control: unanimous failures flag candidate task or evaluator defects, and it exposed a latent evaluator bug missed by self-consistency checks. The evaluators are compact deterministic surrogates, but the task contract allows their internals to be upgraded to simulator-backed checks without changing how tasks are posed or solved.