Audio and Speech Processing · Cross-listed from cs.LG
CSyMR: Benchmarking Compositional Music Information Retrieval in Symbolic Music Reasoning
Boyang Wang, Yash Vishe, Xin Xu et al. (+4 more)
Natural
language information needs over symbolic music scores rarely reduce to a single step lookup. Many queries require compositional Music Information
Retrieval (MIR) that extracts multiple pieces of evidence from structured notation and aggregates them to answer the question. This setting remains challenging for Large
Language Models due to the mismatch between natural
language intents and symbolic representations, as well as the difficulty of reliably handling long structured contexts. Existing
benchmarks only partially capture these
retrieval demands, often emphasizing isolated theoretical knowledge or simplified settings. We introduce CSyMR-Bench, a
benchmark for compositional MIR in symbolic music
reasoning grounded in authentic user scenarios. It contains 126 multiple choice questions curated from community discussions and professional examinations, where each item requires chaining multiple atomic analyses over a score to derive implicit musical evidence. To support diagnosis, we provide a taxonomy with six query intent categories and six analytical dimension tags. We further propose a tool-augmented
retrieval and
reasoning framework that integrates a ReAct-style controller with deterministic symbolic analysis operators built with music21. Experiments across prompting baselines and
agent variants show that tool-grounded compositional
retrieval consistently outperforms Large
Language Model-only approaches, yielding 5-7% absolute accuracy gains, with the largest improvements on analysis-heavy categories.
Read the full abstract ▾
All authors Boyang Wang, Yash Vishe, Xin Xu, Zachary Novack, Xunyi Jiang, Julian McAuley, Junda Wu
Systems and Control
Optimizing Sensor Node Localization for Achieving Sustainable Smart Agriculture System Connectivity
Mohamed Naeem
The innovative agriculture system is revolutionizing how we farm, making it one of the most critical innovations of our time! Yet it faces significant connectivity challenges, particularly with the sensors that power this technology. An efficient sensor deployment solution is still required to maximize the network's detection capabilities and efficiency while minimizing resource consumption and operational costs. This paper introduces an innovative sensor allocation optimization method that employs a Gradient-Based Iteration with Lagrange. The proposed method enhances coverage by utilizing a hybrid approach while minimizing the number of sensor nodes required under grid-based allocation. The proposed sensor distribution outperformed the classic deterministic deployment across coverage, number of sensors, cost, and power consumption. Furthermore, scalability is enhanced by extending sensing coverage to the remaining area via Bluetooth, which has a shorter communication range. Moreover, the proposed
algorithm achieved 98.5% wireless sensor coverage, compared with 95% for the particle swarm distribution.
Read the full abstract ▾
Audio and Speech Processing · Cross-listed from cs.CV
Adaptive Multimodal Person Recognition: A Robust Framework for Handling Missing Modalities
Aref Farhadipour, Teodora Vukovic, Volker Dellwo et al. (+2 more)
Person identification systems often rely on audio, visual, or behavioral cues, but real-world conditions frequently present with missing or degraded modalities. To address this challenge, we propose a
multimodal person identification framework incorporating upper-body motion, face, and voice. Experimental results demonstrate that body motion outperforms traditional modalities such as face and voice in within-session evaluations, while serving as a complementary cue that enhances performance in multi-session scenarios. Our model employs a unified hybrid fusion strategy, fusing both feature-level and score-level information to maximize representational richness and decision accuracy. Specifically, it leverages multi-task learning to process modalities independently, followed by cross-
attention and gated fusion mechanisms to exploit both unimodal information and cross-modal interactions. Finally, a confidence-weighted strategy and mistake-correction mechanism dynamically adapt to missing data, ensuring that our single classification head achieves optimal performance even in unimodal and bimodal scenarios. We evaluate our method on CANDOR, a newly introduced interview-based
multimodal dataset, which we
benchmark in this work for the first time. Our results demonstrate that the proposed trimodal system achieves 99.51% Top-1 accuracy on person identification tasks. In addition, we evaluate our model on the VoxCeleb1
dataset as a widely used evaluation protocol and reach 99.92% accuracy in bimodal mode, outperforming conventional approaches. Moreover, we show that our system maintains high accuracy even when one or two modalities are unavailable, making it a robust solution for real-world person recognition applications. The code and data for this work are publicly available.
Read the full abstract ▾
All authors Aref Farhadipour, Teodora Vukovic, Volker Dellwo, Petr Motlicek, Srikanth Madikeri
Image and Video Processing · Cross-listed from physics.med-ph
Vector Flow Imaging in Layered Models With a High Speed of Sound Contrast Using Pulse-Echo Ultrasound and Photoacoustics
Caitlin Smith, Guillaume Renaud, Kasper van Wijk et al. (+1 more)
In this study, we develop vector flow imaging techniques for multi-layered models with a high wavespeed contrast using photoacoustic and ultrasonic imaging. We use refraction-corrected delay-and-sum image reconstruction (RC-DAS), which enforces Snell's law to accurately calculate time delays within each layer. We compare RC-DAS against conventional delay-and-sum for vector flow imaging in benchtop phantoms made of transparent polymethyl methacrylate (PMMA) in a water bath. We study the flow beneath a PMMA layer using two phantoms, where the PMMA layer has different shapes and thicknesses. We image a slow-moving suspension of carbon microspheres (~4 mm/s) using interleaved photoacoustic and multi-angle plane wave ultrasound acquisitions measured with a 7.6 MHz linear ultrasound array. Photoacoustic waves are generated by a 1064 nm wavelength nanosecond-pulsed
laser at 50 Hz, and multi-angle plane wave ultrasound data are acquired at 100 Hz for eleven steering angles between $\pm$10$^\circ$. RC-DAS improves the flow speed accuracy, reducing the mean absolute error by 0.41-0.63 mm/s compared to the expected flow profile. The error in direction estimates improves when we use RC-DAS, with the interdecile range reducing by up to 17$^\circ$. This work emphasises the importance of refraction correction for accurate flow measurements in layered media with photoacoustics and ultrasonic imaging. While both imaging modalities can quantify flow in these multi-layered models, the modality best suited for a specific application will depend on the imaging target and flow dynamics. These techniques show promise for biomedical applications such as intraosseous and transcranial blood flow quantification, and in nondestructive testing to monitor fluid motion.
Read the full abstract ▾
All authors Caitlin Smith, Guillaume Renaud, Kasper van Wijk, Jami Shepherd
Systems and Control · Cross-listed from stat.AP
Restless Multi-Process Multi-Armed Bandits with Applications to Self-Driving Microscopies
Jaume Anguera Peris, Songtao Cheng, Hanzhao Zhang et al. (+2 more)
High-content screening microscopy generates large amounts of live-
cell imaging data, yet its potential remains constrained by the inability to determine when and where to image most effectively. Optimally balancing acquisition time, computational capacity, and photobleaching budgets across thousands of dynamically evolving regions of interest remains an open challenge, further complicated by limited field-of-view adjustments and sensor sensitivity. Existing approaches either rely on static sampling or heuristics that neglect the dynamic
evolution of biological processes, leading to inefficiencies and missed events. Here, we introduce the restless multi-process multi-armed bandit (RMPMAB), a new decision-theoretic framework in which each experimental region is modeled not as a single process but as an ensemble of Markov chains, thereby capturing the inherent heterogeneity of biological systems such as asynchronous
cell cycles and heterogeneous drug responses. Building upon this foundation, we derive closed-form expressions for transient and asymptotic behaviors of aggregated processes, and design scalable Whittle index policies with sub-linear complexity in the number of imaging regions. Through both simulations and a real biological live-
cell imaging
dataset, we show that our approach achieves substantial improvements in throughput under resource constraints. Notably, our
algorithm outperforms Thomson Sampling,
Bayesian UCB, epsilon-Greedy, and Round Robin by reducing cumulative regret by more than 37% in simulations and capturing 93% more biologically relevant events in live imaging experiments, underscoring its potential for transformative smart microscopy. Beyond improving experimental efficiency, the RMPMAB framework unifies
stochastic decision theory with optimal autonomous microscopy control, offering a principled approach to accelerate discovery across multidisciplinary sciences.
Read the full abstract ▾
All authors Jaume Anguera Peris, Songtao Cheng, Hanzhao Zhang, Wei Ouyang, Joakim Jaldén
Image and Video Processing
Deep learning water-unsuppressed MRSI at ultra-high field for simultaneous quantitative metabolic, susceptibility and myelin water imaging
Paul J. Weiser, Jiye Kim, Jongho Lee et al. (+6 more)
Purpose:
Magnetic Resonance Spectroscopic Imaging (MRSI) maps endogenous
brain metabolism while suppressing the overwhelming water signal. Water-unsuppressed MRSI (wu-MRSI) allows simultaneous imaging of water and metabolites, but large water sidebands cause challenges for metabolic fitting. We developed an end-to-end deep-learning pipeline to overcome these challenges at ultra-high field. Methods:Fast high-resolution wu-MRSI was acquired at 7T with non-cartesian ECCENTRIC sampling and ultra-short echo time. A water and lipid removal network (WALINET+) was developed to remove lipids, water signal, and sidebands. MRSI reconstruction was performed by DeepER and a physics-informed network for metabolite fitting. Water signal was used for absolute metabolite quantification, quantitative susceptibility mapping (QSM), and myelin water fraction imaging (MWF). Results: WALINET+ provided the lowest NRMSE (< 2%) in simulations and in vivo the smallest bias (< 20%) and limits-of-agreement (+-63%) between wu-MRSI and ws-MRSI scans. Several metabolites such as creatine and glutamate showed higher SNR in wu-MRSI. QSM and MWF obtained from wu-MRSI and GRE showed good agreement with 0 ppm/5.5% bias and +-0.05 ppm/ +- 12.75% limits-of-agreement. Conclusion: High-quality metabolic, QSM, and MWF mapping of the human
brain can be obtained simultaneously by ECCENTRIC wu-MRSI at 7T with 2 mm isotropic resolution in 12 min. WALINET+ robustly removes water sidebands while preserving metabolite signal, eliminating the need for water suppression and separate water acquisitions.
Read the full abstract ▾
All authors Paul J. Weiser, Jiye Kim, Jongho Lee, Amirmohammad Shamaei, Gulnur Ungan, Malte Hoffmann, Antoine Klauser, Berkin Bilgic, Ovidiu C. Andronesi
Systems and Control
Assessing the Frequency Response Potential of Heavy-Duty Electric Vehicles with Vehicle-to-Grid Integration in the California Power System
Xiaojie Tao, Yaoyu Fan, Zhaoyi Ye et al. (+1 more)
The integration of heavy-duty electric vehicles (EVs) with Vehicle-to-Grid (V2G) capability can enhance primary frequency response and improve stability in power systems with high renewable penetration. This study evaluates the technical potential of heavy-duty EV fleets to support the California power grid under three practical charging strategies: immediate charging, delayed charging, and constant-minimum-power charging. We develop a simulation framework that couples aggregated frequency dynamics with battery and charger constraints, state-of-charge management, and fleet-availability profiles. Performance is assessed using standard frequency security metrics, including nadir, rate-of-change-of-frequency, overshoot, and settling time, across credible contingency scenarios and renewable generation conditions. Results indicate that both non-V2G modes and V2G-enabled operation can contribute meaningful primary response, with V2G providing the strongest and fastest support while respecting mobility and network limits. Sensitivity analyses show that the relative benefits depend on charging strategy, control parameters, and renewable output, highlighting design trade-offs between response magnitude, duration, and battery usage. Overall, heavy-duty EV fleets-when coordinated by appropriate charging and V2G controls-offer a viable resource for strengthening primary frequency control on the California grid and mitigating stability challenges associated with increasing renewable penetration.
Read the full abstract ▾
All authors Xiaojie Tao, Yaoyu Fan, Zhaoyi Ye, Rajit Gadh
Signal Processing
Compensating Coarse Quantization in Massive MIMO: Channel Estimation and BER under Imperfect CSI
Reza Mohammadkhani, Azad Azizzadeh, Seyed Vahab Al-Din Makki et al. (+2 more)
Low-resolution quantization is essential to reduce implementation cost and power consumption in massive multiple-input multiple-output (
MIMO) systems for 5G and 6G. While most existing studies assume perfect channel state information (CSI), we model the impact of coarse quantization noise on both channel estimation and data transmission, yielding a more realistic assessment of system performance under imperfect CSI conditions in the uplink. We develop a tight approximation for the bit-error ratio (BER) of uncoded M-QAM with zero-forcing detection, based on the linear minimum mean-square error (LMMSE) channel estimate. These analytical results enable compensation strategies that jointly optimize quantization resolution, transmit power, and pilot length across different numbers of users and base station
antennas. We further demonstrate the applicability of the proposed framework through several design scenarios that highlight its effectiveness in optimizing system parameters and improving energy efficiency under quantization constraints. For example, in a 16-QAM system, extending the pilot sequence by 2.5 times and lowering transmit power by 0.5 dB enables a 3-bit quantized system to match the BER of the full-resolution case. The proposed framework offers a fast and accurate alternative to Monte Carlo simulations, enabling practical system optimization under realistic quantization constraints.
Read the full abstract ▾
All authors Reza Mohammadkhani, Azad Azizzadeh, Seyed Vahab Al-Din Makki, John Thompson, Maziar Nekovee
Image and Video Processing · Cross-listed from cs.CV
HERBench: A Benchmark for Multi-Evidence Integration in Video Question Answering
Dan Ben-Ami, Gabriele Serussi, Kobi Cohen et al. (+1 more)
Video Large
Language Models (
Video-LLMs) are improving rapidly, yet current
Video Question Answering (VideoQA)
benchmarks often admit single-cue shortcuts, under-testing
reasoning that must integrate evidence across time. We introduce HERBench, a
benchmark designed to make multi-evidence integration unavoidable: each question requires at least three non-overlapping cues drawn from distinct
video segments. HERBench contains 26,806 five-way multiple-choice questions across 12 compositional tasks. To make evidential demand measurable, we introduce the Minimum Required Frame-Set (MRFS), the smallest number of frames a model must fuse to answer correctly, and show that HERBench imposes higher evidential demand than prior
benchmarks. Evaluating 13 state-of-the-art
Video-LLMs yields only 31-42% accuracy, only modestly above the 20% random-guess baseline. We disentangle this failure into two critical bottlenecks: (1) a
retrieval deficit, where frame selectors overlook key evidence, and (2) a fusion deficit, where models fail to integrate information even when all necessary evidence is provided. HERBench thus provides a principled
benchmark for studying robust multi-evidence
video understanding.
Read the full abstract ▾
All authors Dan Ben-Ami, Gabriele Serussi, Kobi Cohen, Chaim Baskin
Image and Video Processing
Artificial Intelligence for the Assessment of Peritoneal Carcinosis during Diagnostic Laparoscopy for Advanced Ovarian Cancer
Riccardo Oliva, Farahdiba Zarin, Alice Zampolini Faustini et al. (+9 more)
Advanced Ovarian Cancer (AOC) is often diagnosed at an advanced stage with peritoneal carcinosis (PC). Fagotti score (FS) assessment at diagnostic laparoscopy (DL) guides treatment planning by estimating
surgical resectability, but its subjective and operator-dependent nature limits reproducibility and widespread use.
Videos of patients undergoing DL with concomitant FS assessments at a referral center were retrospectively collected and divided into a development
dataset, for data annotation, AI training and evaluation, and an independent test
dataset, for internal validation. In the development
dataset, FS-relevant frames were manually annotated for anatomical structures and PC. Deep learning models were trained to automatically identify FS-relevant frames, segment structures and PC, and predict
video-level FS and indication to surgery (ItS). AI performance was evaluated using Dice score for
segmentation, F1-scores for anatomical stations (AS) and ItS prediction, and root mean square error (RMSE) for final FS estimation. In the development
dataset, the
segmentation model trained on 7,311 frames, achieved Dice scores of 70$\pm$3% for anatomical structures and 56$\pm$3% for PC.
Video-level AS classification achieved F1-scores of 74$\pm$3% and 73$\pm$4%, FS prediction showed normalized RMSE values of 1.39$\pm$0.18 and 1.15$\pm$0.08, and ItS reached F1-scores of 80$\pm$8% and 80$\pm$2% in the development (n=101) and independent test
datasets (n=50), respectively. This is the first AI model to predict the feasibility of cytoreductive surgery providing automated FS estimation from DL
videos. Its reproducible and reliable performance across
datasets suggests that AI can support surgeons through standardized intraoperative tumor burden assessment and clinical decision-making in AOC.
Read the full abstract ▾
All authors Riccardo Oliva, Farahdiba Zarin, Alice Zampolini Faustini, Armine Vardazaryan, Andrea Rosati, Vinkle Srivastav, Nunzia Del Villano, Jacques Marescaux, Giovanni Scambia, Pietro Mascagni, Nicolas Padoy, Anna Fagotti
Image and Video Processing
Magnification-Aware Distillation (MAD): A Self-Supervised Framework for Unified Representation Learning in Gigapixel Whole-Slide Images
Mahmut S. Gokmen, Mitchell A. Klusty, Peter T. Nelson et al. (+8 more)
Whole-slide images (WSIs) contain tissue information distributed across multiple magnification levels, yet most self-supervised methods treat these scales as independent views. This separation prevents models from learning representations that remain stable when resolution changes, a key requirement for practical neuropathology workflows. This study introduces Magnification-Aware
Distillation (MAD), a self-supervised strategy that links low-magnification context with spatially aligned high-magnification detail, enabling the model to learn how coarse tissue structure relates to fine cellular patterns. The resulting foundation model, MAD-NP, is trained entirely through this cross-scale correspondence without annotations. A linear classifier trained only on 10x
embeddings maintains 96.7% of its performance when applied to unseen 40x tiles, demonstrating strong resolution-invariant representation learning.
Segmentation outputs remain consistent across magnifications, preserving anatomical boundaries and minimizing noise. These results highlight the feasibility of scalable, magnification-robust WSI analysis using a unified
embedding space
Read the full abstract ▾
All authors Mahmut S. Gokmen, Mitchell A. Klusty, Peter T. Nelson, Allison M. Neltner, Sen-Ching Samson Cheung, Thomas M. Pearce, David A Gutman, Brittany N. Dugger, Devavrat S. Bisht, Margaret E. Flanagan, V. K. Cody Bumgardner
Signal Processing · Cross-listed from cs.CV
Spherical Leech Quantization for Visual Tokenization and Generation
Yue Zhao, Hanwen Jiang, Zhenlin Xu et al. (+3 more)
Non-parametric quantization has received much
attention due to its efficiency on parameters and scalability to a large codebook. In this paper, we present a unified formulation of different non-parametric quantization methods through the lens of
lattice coding. The geometry of
lattice codes explains the necessity of auxiliary loss terms when training auto-encoders with certain existing lookup-free quantization variants such as BSQ. As a step forward, we explore a few possible candidates, including random
lattices, generalized Fibonacci
lattices, and densest sphere packing
lattices. Among all, we find the Leech
lattice-based quantization method, which is dubbed as Spherical Leech Quantization ($\Lambda_{24}$-SQ), leads to both a simplified training recipe and an improved reconstruction-compression tradeoff thanks to its high
symmetry and even distribution on the hypersphere. In image tokenization and compression tasks, this quantization approach achieves better reconstruction quality across all metrics than BSQ, the best prior art, while consuming slightly fewer bits. The improvement also extends to state-of-the-art auto-regressive image generation frameworks.
Read the full abstract ▾
All authors Yue Zhao, Hanwen Jiang, Zhenlin Xu, Chutong Yang, Ehsan Adeli, Philipp Krähenbühl
Audio and Speech Processing · Cross-listed from cs.CL
Spoken DialogSum: An Emotion-Rich Conversational Dataset for Spoken Dialogue Summarization
Yen-Ju Lu, Kunxiao Gao, Mingrui Liang et al. (+5 more)
Recent audio
language models can follow long conversations. However, research on
emotion-aware or spoken dialogue summarization is constrained by the lack of data that links
speech, summaries, and paralinguistic cues. We introduce Spoken DialogSum, the first corpus aligning raw conversational audio with factual summaries,
emotion-rich summaries, and utterance-level labels for speaker age, gender, and
emotion. The
dataset is built in two stages: first, an LLM rewrites DialogSum scripts with Switchboard-style fillers and back-channels, then tags each utterance with
emotion, pitch, and speaking rate. Second, an expressive TTS engine synthesizes
speech from the tagged scripts, aligned with paralinguistic labels. Spoken DialogSum comprises 13,460
emotion-diverse dialogues, each paired with both a factual and an
emotion-focused summary. We release an online demo at https://fatfat-emosum.github.io/EmoDialog-Sum-Audio-Samples/, with plans to release the full
dataset in the near future. Baselines show that an Audio-LLM raises emotional-summary ROUGE-L by 28% relative to a cascaded ASR-LLM system, confirming the value of end-to-end
speech modeling.
Read the full abstract ▾
All authors Yen-Ju Lu, Kunxiao Gao, Mingrui Liang, Helin Wang, Thomas Thebaud, Laureano Moro-Velazquez, Najim Dehak, Jesus Villalba
Systems and Control · Cross-listed from math.OC
Enhancing Orbital Debris Remediation with Reconfigurable Space-Based Laser Constellations
David O. Williams Rogers and Hang Woon Lee
Orbital debris poses an escalating threat to space missions and the long-term sustainability of Earth's orbital environment. The literature proposes various approaches for orbital debris remediation, including the use of multiple space-based
lasers that collaboratively engage debris targets. While the proof of concept for this
laser-based approach has been demonstrated, critical questions remain about its scalability and responsiveness as the debris population continues to expand rapidly. This paper introduces constellation reconfiguration as a system-level strategy to address these limitations. Through coordinated orbital maneuvers,
laser-equipped satellites can dynamically adapt their positions to respond to evolving debris distributions and time-critical events. We formalize this concept as the Reconfigurable
Laser-to-Debris Engagement Scheduling Problem (R-L2D-ESP), an optimization framework that determines the optimal sequence of constellation reconfigurations and
laser engagements to maximize debris remediation capacity, which quantifies the constellation's ability to nudge, deorbit, or perform just-in-time collision avoidance maneuvers on debris objects. To manage the complexity of this combinatorial optimization problem, we employ a receding horizon approach. Our experiments reveal that reconfigurable constellations significantly outperform static ones, achieving greater debris remediation capacity and successfully deorbiting substantially more debris objects. Additionally, our sensitivity analyses identify the key parameters that influence remediation performance the most, providing essential insights for future system design. These findings demonstrate that constellation reconfiguration represents a promising advancement for
laser-based debris removal systems, offering the adaptability and scalability necessary to enhance this particular approach to orbital debris remediation.
Read the full abstract ▾
Image and Video Processing
Configurable {\gamma} Photon Spectrometer to Enable Precision Radioguided Tumor Resection
Rahul Lall, Youngho Seo, Ali M. Niknejad et al. (+1 more)
Surgical tumor resection aims to remove all cancer
cells in the tumor margin and at centimeter-scale depths below the tissue surface. During surgery, microscopic clusters of disease are intraoperatively difficult to visualize and are often left behind, significantly increasing the risk of cancer recurrence. Radioguided surgery (RGS) has shown the ability to selectively tag cancer
cells with gamma ({\gamma})
photon emitting radioisotopes to identify them, but require a mm-scale {\gamma}
photon spectrometer to localize the position of these
cells in the tissue margin (i.e., a function of incident {\gamma}
photon energy) with high specificity. Here we present a 9.9 mm2 integrated
circuit (IC)-based {\gamma} spectrometer implemented in 180 nm CMOS, to enable the measurement of single {\gamma}
photons and their incident energy with sub-keV energy resolution. We use small 2 2 um reverse-biased diodes that have low depletion region capacitance, and therefore produce millivolt-scale voltage signals in response to the small charge generated by incident {\gamma}
photons. A low-power energy spectrometry method is implemented by measuring the decay time it takes for the generated voltage signal to settle back to DC after a {\gamma} detection event, instead of measuring the voltage drop directly. This spectrometry method is implemented in three different pixel architectures that allow for configurable pixel sensitivity, energy-resolution, and energy dynamic range based on the widely heterogenous
surgical and patient presentation in RGS. The spectrometer was tested with three common {\gamma}-emitting radioisotopes (64Cu, 133Ba, 177Lu), and is able to resolve activities down to 1 uCi with sub-keV energy resolution and 1.315 MeV energy dynamic range, using 5-minute acquisitions.
Read the full abstract ▾
All authors Rahul Lall, Youngho Seo, Ali M. Niknejad, Mekhail Anwar
Systems and Control · Cross-listed from cs.LG
gridfm-datakit-v1: A Python Library for Scalable and Realistic Power Flow and Optimal Power Flow Data Generation
Alban Puech, Matteo Mazzonelli, Celia Cintas et al. (+11 more)
We introduce gridfm-datakit-v1, a Python library for generating realistic and diverse Power Flow (PF) and Optimal Power Flow (OPF)
datasets for training Machine Learning (ML) solvers. Existing
datasets and libraries face three main challenges: (1) lack of realistic
stochastic load and
topology perturbations, limiting scenario diversity; (2) PF
datasets are restricted to OPF-feasible points, hindering generalization of ML solvers to cases that violate operating limits (e.g., branch overloads or voltage violations); and (3) OPF
datasets use fixed generator cost functions, limiting generalization across varying costs. gridfm-datakit addresses these challenges by: (1) combining global load scaling from real-world profiles with localized noise and supporting arbitrary N-k
topology perturbations to create diverse yet realistic
datasets; (2) generating PF samples beyond operating limits; and (3) producing OPF data with varying generator costs. It also scales efficiently to large grids (up to 10,000 buses). Comparisons with OPFData, OPF-Learn, PGLearn, and PF$\Delta$ are provided. Available on GitHub at https://github.com/gridfm/gridfm-datakit under Apache 2.0 and via `pip install gridfm-datakit`.
Read the full abstract ▾
All authors Alban Puech, Matteo Mazzonelli, Celia Cintas, Tamara R. Govindasamy, Mangaliso Mngomezulu, Jonas Weiss, Matteo Baù, Anna Varbella, François Mirallès, Kibaek Kim, Le Xie, Hendrik F. Hamann, Etienne Vos, Thomas Brunschwiler
Audio and Speech Processing
Segmental Attention Decoding With Long Form Acoustic Encodings
Pawel Swietojanski, Xinwei Li, Mingbin Xu et al. (+3 more)
We address the fundamental incompatibility of
attention-based encoder-decoder (AED) models with long-form acoustic encodings. AED models trained on segmented utterances learn to encode absolute frame positions by exploiting limited acoustic context beyond segment boundaries, but fail to generalize when decoding long-form segments where these cues vanish. The model loses ability to order acoustic encodings due to permutation invariance of keys and values in cross-
attention. We propose four modifications: (1) injecting explicit absolute positional encodings into cross-
attention for each decoded segment, (2) long-form training with extended acoustic context to eliminate implicit absolute position encoding, (3) segment concatenation to cover diverse
segmentations needed during training, and (4) semantic
segmentation to align AED-decoded segments with training segments. We show these modifications close the accuracy gap between continuous and segmented acoustic encodings, enabling auto-regressive use of the
attention decoder.
Read the full abstract ▾
All authors Pawel Swietojanski, Xinwei Li, Mingbin Xu, Takaaki Hori, Dogan Can, Xiaodan Zhuang
Image and Video Processing · Cross-listed from cs.CV
Adaptable Segmentation Pipeline for Diverse Brain Tumors with Radiomic-Guided Subtyping and Lesion-Wise Model Ensemble
Daniel Capellán-Martín, Abhijeet Parida, Zhifan Jiang et al. (+6 more)
Robust and generalizable
segmentation of
brain tumors on multi-parametric
magnetic resonance imaging (MRI) remains difficult because tumor types differ widely. The BraTS 2025 Lighthouse Challenge
benchmarks segmentation methods on diverse high-quality
datasets of adult and pediatric tumors: multi-consortium international pediatric
brain tumor
segmentation (PED), preoperative meningioma tumor
segmentation (MEN), meningioma radiotherapy
segmentation (MEN-RT), and
segmentation of pre- and post-treatment
brain metastases (MET). We present a flexible, modular, and adaptable pipeline that improves
segmentation performance by selecting and combining state-of-the-art models and applying tumor- and lesion-specific processing before and after training. Radiomic features extracted from MRI help detect tumor subtype, ensuring a more balanced training. Custom lesion-level performance metrics determine the influence of each model in the ensemble and optimize post-processing that further refines the predictions, enabling the workflow to tailor every step to each case. On the BraTS testing sets, our pipeline achieved performance comparable to top-ranked
algorithms across multiple challenges. These findings confirm that custom lesion-aware processing and model selection yield robust
segmentations yet without locking the method to a specific network architecture. Our method has the potential for quantitative tumor measurement in clinical practice, supporting diagnosis and prognosis.
Read the full abstract ▾
All authors Daniel Capellán-Martín, Abhijeet Parida, Zhifan Jiang, Nishad Kulkarni, Krithika Iyer, Austin Tapp, Syed Muhammad Anwar, María J. Ledesma-Carbayo, Marius George Linguraru
Image and Video Processing
An Energy-Efficient Adiabatic Capacitive Neural Network Chip
Himadri Singh Raghav, Sachin Maheshwari, Mike Smart et al. (+2 more)
Recent advances in artificial intelligence, coupled with increasing data bandwidth requirements, in applications such as
video processing and high-resolution sensing, have created a growing demand for high computational performance under stringent energy constraints, especially for battery-powered and edge devices. To address this, we present a mixed-signal adiabatic capacitive neural network chip, designed in a 130$nm$ CMOS technology, to demonstrate significant energy savings coupled with high image classification accuracy. Our dual-layer hardware chip, incorporating 16 single-cycle multiply-accumulate engines, can reliably distinguish between 4 classes of 8x8 1-bit images, with classification results over 95%, within 2.7% of an equivalent software version. Energy measurements reveal average energy savings between 2.1x and 6.8x, compared to an equivalent CMOS capacitive implementation.
Read the full abstract ▾
All authors Himadri Singh Raghav, Sachin Maheshwari, Mike Smart, Patrick Foster, Alex Serb
Signal Processing
Tunable Gaussian Pulse for Delay-Doppler ISAC
Bruno Felipe Costa, Anup Mishra, Israel Leyva-Mayorga et al. (+2 more)
Integrated sensing and communication (
ISAC) for next-generation networks targets robust operation under high mobility and high Doppler spread, leading to severe inter-carrier interference (ICI) in systems based on orthogonal frequency-division multiplexing (OFDM) waveforms. Delay–Doppler (DD)-domain
ISAC offers a more robust foundation under high mobility, but it requires a suitable DD-domain pulse-shaping filter. The prevailing DD pulse designs are either communication-centric or static, which limits adaptation to non-stationary channels and diverse application demands. To address this limitation, this paper introduces the tunable Gaussian pulse (TGP), a DD-native, analytically tunable pulse shape parameterized by its aspect ratio \( \gamma \), chirp rate \( \alpha_c \), and phase coupling \( \beta_c \). On the sensing side, we derive closed-form Cramér–Rao lower bounds (CRLBs) that map \( (\gamma,\alpha_c,\beta_c) \) to fundamental delay and Doppler precision. On the communications side, we show that \( \alpha_c \) and \( \beta_c \) reshape off-diagonal covariance, and thus inter-symbol interference (ISI), without changing received power, isolating capacity effects to interference structure rather than power loss. A comprehensive trade-off analysis demonstrates that the TGP spans a flexible operational region from the high capacity of the Sinc pulse to the high precision of the root raised cosine (RRC) pulse. Notably, TGP attains near-RRC sensing precision while retaining over \( 90\% \) of Sinc's maximum capacity, achieving a balanced operating region that is not attainable by conventional static pulse designs.
Read the full abstract ▾
All authors Bruno Felipe Costa, Anup Mishra, Israel Leyva-Mayorga, Taufik Abrão, Petar Popovski
Signal Processing
Fusion of Cellular ISAC and Passive RF Sensing for UAV Detection and Tracking
Cole Dickerson, Sean Kearney, Sultan Manjur et al. (+5 more)
The rapid growth of unmanned aerial vehicles (
UAVs) in civilian and critical-infrastructure airspace has created a need for reliable detection and tracking systems that operate under diverse environmental and sensing conditions. This paper presents a
UAV detection and tracking system that fuses measurements from a network of passive Keysight N6841A RF sensors and a Ku-band Fortem TrueView R20 radar operating in the FR3 spectrum (16.3 GHz) as an
ISAC proxy. Real-world experiments at the NSF AERPAW testbed demonstrate that radar and RF sensing provide complementary strengths under varying
geometric, range, and line-of-sight conditions. A Kalman filter using a constant-velocity motion model integrates the asynchronous 2D RF and 3D radar observations, suppressing large standalone errors, improving accuracy over individual modalities, and increasing tracking coverage without degrading performance. These results demonstrate the effectiveness of multi-modal,
ISAC-oriented sensing for robust
UAV tracking in outdoor environments.
Read the full abstract ▾
All authors Cole Dickerson, Sean Kearney, Sultan Manjur, Ismail Guvenc, Sevgi Gurbuz, Ali Gurbuz, Ozgur Ozdemir, Mihail Sichitiu
Image and Video Processing
Test Time Optimized Generalized AI-based Medical Image Registration Method
Sneha Sree C., Dattesh Shanbhag and Sudhanya Chatterjee
Medical image registration is critical for aligning anatomical structures across imaging modalities such as computed
tomography (CT),
magnetic resonance imaging (MRI), and ultrasound. Among existing techniques, non-rigid registration (NRR) is particularly challenging due to the need to capture complex anatomical deformations caused by physiological processes like respiration or contrast-induced signal variations. Traditional NRR methods, while theoretically robust, often require extensive parameter tuning and incur high computational costs, limiting their use in real-time clinical workflows. Recent deep learning (DL)-based approaches have shown promise; however, their dependence on task-specific retraining restricts scalability and adaptability in practice. These limitations underscore the need for efficient, generalizable registration frameworks capable of handling heterogeneous imaging contexts. In this work, we introduce a novel AI-driven framework for 3D non-rigid registration that generalizes across multiple imaging modalities and anatomical regions. Unlike conventional methods that rely on application-specific models, our approach eliminates anatomy- or modality-specific customization, enabling streamlined integration into diverse clinical environments.
Read the full abstract ▾
Signal Processing · Cross-listed from cs.LG
Synthetic Electrogram Generation with Variational Autoencoders for ECGI
Miriam Gutiérrez-Fernández, Karen López-Linares, Carlos Fambuena-Santos et al. (+3 more)
Atrial fibrillation (AF) is the most prevalent sustained cardiac arrhythmia, and its clinical assessment requires accurate characterization of atrial electrical activity. Noninvasive electrocardiographic imaging (ECGI) combined with deep learning (DL) approaches for estimating intracardiac electrograms (EGMs) from body surface potentials (BSPMs) has shown promise, but progress is hindered by the limited availability of paired BSPM-EGM
datasets. To address this limitation, we investigate variational autoencoders (VAEs) for the generation of synthetic multichannel atrial EGMs. Two models are proposed: a sinus rhythm-specific VAE (VAE-S) and a class-conditioned VAE (VAE-C) trained on both sinus rhythm and AF signals. Generated EGMs are evaluated using morphological, spectral, and distributional similarity metrics. VAE-S achieves higher fidelity with respect to in silico EGMs, while VAE-C enables rhythm-specific generation at the expense of reduced sinus reconstruction quality. As a proof of concept, the generated EGMs are used for data augmentation in a downstream noninvasive EGM reconstruction task, where moderate augmentation improves estimation performance. These results demonstrate the potential of VAE-based generative modeling to alleviate data scarcity and enhance deep learning-based ECGI pipelines.
Read the full abstract ▾
All authors Miriam Gutiérrez-Fernández, Karen López-Linares, Carlos Fambuena-Santos, María S. Guillem, Andreu M. Climent, Óscar Barquero-Pérez
Systems and Control
Scalable Nonlinear DeePC: Bridging Direct and Indirect Methods and Basis Reduction
Thomas O. de Jong, Mircea Lazar, Siep Weiland et al. (+1 more)
This paper studies regularized data-enabled predictive control (DeePC) within a nonlinear framework and its relationship to subspace predictive control (SPC). The $\Pi$-regularization is extended to general basis functions and it is shown that, under suitable conditions, the resulting basis functions DeePC formulation constitutes a relaxation of basis functions SPC. To improve scalability, we introduce an SVD-based dimensionality reduction that preserves the equivalence with SPC, and we derive a reduced {\Pi}-regularization. A LASSO based sparse basis selection method is proposed to obtain a reduced basis from lifted data. Simulations on a nonlinear van der Pol oscillator model indicate that, in the absence of noise, DeePC and SPC yield equivalent absolute mean tracking errors (AMEs) when large penalties are applied. In contrast, under noisy measurements, careful tuning of the DeePC regularization results in a reduced AME, outperforming SPC.
Read the full abstract ▾
All authors Thomas O. de Jong, Mircea Lazar, Siep Weiland, Florian Dörfler
Systems and Control · Cross-listed from math.OC
The Innovation Null Space of the Kalman Predictor: A Stochastic Perspective for DeePC
Aihui Liu and Magnus Jansson
Willems' fundamental lemma uses a key decision variable $g$ to combine measured input-output data and describe trajectories of a linear time-invariant system. In this paper, we ask: what is a good choice for this vector $g$ when the system is affected by noise? For a linear system with Gaussian noise, we show that there exists an optimal subspace for this decision variable $g$, which is the null space of the innovation Hankel
matrix. If the decision vector lies in this null space, the resulting predictor gets closer to the Kalman predictor. To show this, we use a result that we refer to as the Kalman Filter Fundamental Lemma (KFFL), which applies Willems' lemma to the Kalman predictor. This viewpoint also explains several existing data-driven predictive control methods: regularized DeePC schemes act as soft versions of the innovation null-space constraint, instrumental-variable methods enforce it by construction, and ARX-based approaches explicitly estimate this innovation null space.
Read the full abstract ▾
Systems and Control
Closed-Loop Consistent, Causal Data-Driven Predictive Control via SSARX
Aihui Liu and Magnus Jansson
We propose a fundamental-lemma-free data-driven predictive control (DDPC) scheme for synthesizing model predictive control (MPC)-like policies directly from input-output data. Unlike the well-known DeePC approach and other DDPC methods that rely on Willems' fundamental lemma, our method avoids stacked Hankel representations and the DeePC decision variable g. Instead, we develop a closed-loop consistent, causal DDPC scheme based on the multi-step predictor Subspace-ARX (SSARX). The method first (i) estimates predictor/observer Markov parameters via a high-order ARX model to decouple the noise, then (ii) learns a multi-step past-to-future map by regression, optionally with a reduced-rank constraint. The SSARX predictor is strictly causal, which allows it to be integrated naturally into an MPC formulation. Our experimental results show that SSARX performs competitively with other methods when applied to closed-loop data affected by measurement and process noise.
Read the full abstract ▾
Audio and Speech Processing · Cross-listed from cs.CL
Linguists should learn to love speech-based deep learning models
Marianne de Heer Kloots, Paul Boersma and Willem Zuidema
Futrell and Mahowald present a useful framework bridging technology-oriented deep learning systems and explanation-oriented linguistic theories. Unfortunately, the target article's focus on generative text-based LLMs fundamentally limits fruitful interactions with linguistics, as many interesting questions on human
language fall outside what is captured by written text. We argue that audio-based deep learning models can and should play a crucial role.
Read the full abstract ▾
Signal Processing
Hybrid Cognitive IoT with Cooperative Caching and SWIPT-EH: A Hierarchical Reinforcement Learning Framework
Nadia Abdolkhani and Walaa Hamouda
This paper proposes a hierarchical deep reinforcement learning (DRL) framework based on the soft actor-critic (SAC)
algorithm for hybrid underlay-overlay cognitive Internet of Things (CIoT) networks with simultaneous wireless information and power transfer (SWIPT)-energy harvesting (EH) and cooperative caching. Unlike prior hierarchical DRL approaches that focus primarily on spectrum access or power control, our work jointly optimizes EH, hybrid access coordination, power allocation, and caching in a unified framework. The joint optimization problem is formulated as a weighted-sum multi-objective task, designed to maximize throughput and cache hit ratio while simultaneously minimizing transmission delay. In the proposed model, CIoT
agents jointly optimize EH and data transmission using a learnable time switching (TS) factor. They also coordinate spectrum access under hybrid overlay-underlay paradigms and make power control and cache placement decisions while considering energy, interference, and storage constraints. Specifically, in this work, cooperative caching is used to enable overlay access, while power control is used for underlay access. A novel three-level hierarchical SAC (H-SAC)
agent decomposes the mixed discrete-continuous action space into modular subproblems, improving scalability and convergence over flat DRL methods. The high-level policy adjusts the TS factor, the mid-level policy manages spectrum access coordination and cache sharing, and the low-level policy decides transmit power and caching actions for both the CIoT
agent and PU content. Simulation results show that the proposed hierarchical SAC approach significantly outperforms
benchmark and greedy strategies. It achieves better performance in terms of average sum rate, delay, cache hit ratio, and energy efficiency, even under channel fading and uncertain conditions.
Read the full abstract ▾
Signal Processing · Cross-listed from cs.LG
AnySleep: a channel-agnostic deep learning system for high-resolution sleep staging in multi-center cohorts
Niklas Grieger, Jannik Raskob, Siamak Mehrkanoon et al. (+1 more)
Sleep is essential for health, yet studying its dynamics requires manual sleep staging, a labor-intensive step in research and clinical care. Across centers, polysomnography (PSG) recordings are traditionally scored in 30-s epochs for pragmatic, not physiological, reasons and vary in electrode count, montage, and subject characteristics. These constraints challenge harmonized multi-center studies and the discovery of robust biomarkers on shorter timescales. We present AnySleep, a deep neural network that scores sleep from any electroencephalography (
EEG) or electrooculography (EOG) data at adjustable temporal resolutions. We trained and validated the model on over 20,000 overnight recordings (> 200,000 h of
EEG and EOG) from 28
datasets across multiple clinics to promote robust generalization across sites. The model attains state-of-the-art performance and surpasses or equals established baselines at 30-s epochs. Performance improves with more channels, yet remains strong when EOG is absent, or only EOG or single
EEG derivations (frontal, central, or occipital) are available. On sub-30-s timescales, the model captures short wake intrusions consistent with arousals and improves the prediction of pathophysiological conditions (obstructive sleep apnea, narcolepsy type 1, insomnia) over 30-s scoring. We make the model publicly available to facilitate large-scale studies with heterogeneous electrode setups and accelerate biomarker discovery in sleep.
Read the full abstract ▾
All authors Niklas Grieger, Jannik Raskob, Siamak Mehrkanoon, Stephan Bialonski
Systems and Control
Equivariant Observer for Bearing Estimation with Linear and Angular Velocity Inputs
Gil Serrano, Marcelo Jacinto, Bruno J. Guerreiro et al. (+1 more)
This work addresses the problem of designing an equivariant observer for a first order dynamical system on the unit-sphere. Building upon the established case of unit bearing vector dynamics with angular velocity inputs, we introduce an additional linear velocity input projected onto the unit-sphere tangent space. This extended formulation is particularly useful in image-based visual servoing scenarios where stable bearing estimates are required and the relative velocity between the vehicle and target features must be accounted for. Leveraging lifted kinematics to the Special Orthogonal group, we design an observer for the bearing vector and prove its almost global asymptotic stability. Additionally, we demonstrate how the equivariant observer can be expressed in the original state
manifold. Numerical simulation results validate the effectiveness of the proposed
algorithm.
Read the full abstract ▾
All authors Gil Serrano, Marcelo Jacinto, Bruno J. Guerreiro, Rita Cunha