Peer Revenue Equity in science

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Electrical Engineering and Systems Science

eess Watch Electrical Engineering and Systems Science

Saturday 4 July

Audio and Speech Processing · Cross-listed from cs.SD

TokAN: Accent Normalization Using Self-Supervised Speech Tokens

Qibing Bai, Shuai Wang, Yuhan Du et al. (+3 more)

Accent normalization (AN) seeks to convert non-native (L2) accented speech into standard (L1) speech while preserving speaker identity. The current techniques either require naturally recorded parallel L1-L2 speech for training, or suffer from quality degradation when supervised by synthesized targets. In this paper, we present TokAN, a token-based accent normalization framework that operates on self-supervised discrete speech tokens extracted from a L1-L2 jointly trained vector-quantization (VQ) tokenizer, without the need of synthetic supervisory speech. An autoregressive encoder-decoder model performs token-to-token conversion, translating L2-accented token sequences into the tokens of standard voice. We also introduce reinforcement learning (RL) post-training based on Group Relative Policy Optimization (GRPO), using word error rate and accent classifier confidence as complementary rewards. A non-autoregressive flow-matching synthesizer recovers the Mel-spectrogram from the converted tokens, conditioned on the source speaker embedding. We also develop a flow-matching duration predictor that supports total-duration-aware synthesis, making TokAN applicable to duration-critical tasks such as voice dubbing and live casting. Experiments on seven English accents demonstrate that TokAN reduced the word error rate from 12.40% to 9.89% after supervised fine-tuning, and further to 9.23% after RL post-training, consistently outperforming frame-to-frame, direct flow-matching, and prompt-based token-conversion baselines in terms of accent reduction and intelligibility.
Read the full abstract

All authors Qibing Bai, Shuai Wang, Yuhan Du, Bohan Li, Yannan Wang, Haizhou Li

Signal Processing

A Triple-Band Bandpass Filter Using Square Open-Loop Resonators

Rashmi Ravi, Bhaskarareddy N C and Augustine O. Nwajana

The rapid advancement of multi-band wireless communication systems has driven demand for compact, high-performance multi-band bandpass filters (BPFs) capable of isolating specific frequency bands within a wider spectrum. This paper reviews the current state-of-the-art in multi-band filter design and implementation techniques and presents the design, simulation, and characterisation of a prototype triple-band bandpass filter to validate one of the investigated techniques. A triple-band BPF operating at 2.1 GHz, 2.2 GHz, and 2.3 GHz is designed using Keysight ADS software and implemented on Rogers RT/Duroid 6010LM substrate (dielectric constant = 10.7, loss tangent = 0.0023, thickness = 1.27 mm). The design employs nine square open-loop resonators three per passband with a characteristic impedance of 50 {\Omega} and a fractional bandwidth of 6%. Simulation results demonstrate insertion losses of 0.674 dB, 0.976 dB, and 1.314 dB, and return losses of 16.785 dB, 33.609 dB, and 17.162 dB across the three passbands respectively. The 2.2 GHz centre frequency is achieved with high precision, confirming the effectiveness of the square open-loop resonator transformation technique for compact multi-band filter design in modern wireless communication systems.
Read the full abstract

Signal Processing

Task-Oriented Multimodal Edge Intelligence via Integrated Sensing-Communication-Computation

Weiwei Chen, Yinghui He, Zhong Ye et al. (+2 more)

Integrated sensing, communication, and computation (ISCC) has recently emerged as a unified framework for enabling edge intelligence. However, existing ISCC designs predominantly rely on single-modal sensing, which is inherently vulnerable to occlusions, environmental uncertainties, and modality-specific failures, leading to degraded robustness in real-world deployments. This motivates the need for multi-modal ISCC, yet its design remains insufficiently explored. Compared with the single-modal case, multi-modal ISCC is more challenging because heterogeneous modalities enlarge data dimensionality and tighten communication/computation/energy budgets, while inter-modal correlations further complicate performance characterization. To address these challenges, we propose a task-oriented multi-modal ISCC framework that integrates device-side feature extraction with edge-side joint multi-modal inference. A central component of our approach is the maximal coding rate reduction (MCR^2) criterion, which enables each device to learn compact and discriminative task-relevant features, offering clear advantages over conventional cross-entropy-based extractors. We further leverage MCR^2 as a principled metric for edge-side sensing evaluation. On this basis, we formulate a sensing accuracy maximization problem under delay and resource constraints and develop an efficient block coordinate descent (BCD) algorithm after transforming the problem into a more tractable equivalent form. Focusing on a human activity recognition task, we conduct extensive experiments on publicly available datasets to evaluate the performance of the proposed ISCC framework. The results demonstrate that our approach consistently outperforms three baseline schemes under limited resource conditions.
Read the full abstract

All authors Weiwei Chen, Yinghui He, Zhong Ye, Dingzhu Wen, Guanding Yu

Systems and Control

Approximate Dynamic Optimization via Deep Neural Operators

Amin Nassaji, Ilias Mitrai and Prodromos Daoutidis

This paper addresses the solution of nonlinear dynamic optimization problems that compute optimal manipulated input profiles to enforce desired output profiles. Such trajectory optimization problems commonly arise in chemical process applications, for example, batch processes where optimal temperature or feeding profiles (in case of fed-batch processes) are calculated to enforce time-varying product quality profiles, tightly controlling the reaction rate or rate of heat generation. We propose deep neural operators that approximate function to function mappings as surrogates for the solution of such dynamic optimization problems. We specifically employ deep operator networks (DeepONets) and Fourier-enhanced DeepONets in a batch polymerization reactor case study for which number-average and weight-average molecular weight profiles, together with a final conversion target, are enforced through an optimal temperature program. Our results show that the Fourier-enhanced DeepONet architecture performs very well in approximating the solution of the dynamic optimization problem for different instances, achieving a lower prediction error compared to the standard DeepONet architecture and standard feedforward neural networks.
Read the full abstract

Image and Video Processing · Cross-listed from cs.CV

ContiStain: Cross-Domain Relation-Preserving Distillation for Continual Multi-Domain Virtual IHC Staining

Fuqiang Chen, Yifeng Wang, Hongpeng Wang et al. (+1 more)

A unified multiplex virtual staining model enables scalable and non-destructive multiplex analysis from H&E slides while promoting parameter efficiency, shared pathological knowledge, and consistent cross-biomarker representations. However, in clinical practice, data for new biomarkers are typically acquired sequentially over time. Fine-tuning on such temporally arriving data leads to severe performance degradation on previously learned biomarkers, as sequential optimization disrupts the structured relationships among biomarker representations in the latent space. To address this issue, we propose ContiStain, an IHC multi-domain relational distillation framework for continual virtual staining. We first (i) construct a domain-aware structured feature space using a mixture-of-experts (MoE) feature extractor to reduce representation interference across biomarker domains. Based on this stabilized feature space, we then (ii) propose a relation-preserving distillation strategy that explicitly enforces the consistency of cross-domain token-level cosine similarity matrices between learned biomarker domains during continual adaptation. By maintaining cross-domain structural coherence, ContiStain mitigates forgetting while retaining adaptability to new domains. Experiments on the MIST dataset under a four-domain sequential virtual IHC staining setting show improved stability, reducing FID and ConchFID by 11.1 and 60.9 compared to sequential fine-tuning, enabling scalable and robust multi-domain virtual staining. Code is released at https://github.com/ccitachi/ContiStain.
Read the full abstract

All authors Fuqiang Chen, Yifeng Wang, Hongpeng Wang, Yongbing Zhang

Systems and Control · Cross-listed from math.OC

Stability of input-output maps and their minimal realizations in state-linear, state-affine, LPV, and linear switched systems

Mihály Petreczky, Juan-Pablo Ortega, Florian Rossmannek et al. (+1 more)

Stability is often assumed in learning and identification, yet it is rarely characterized directly from input–output data. We show that an input–output family admits a stable finite-dimensional state-linear realization iff it has finite Hankel-rank and its response decays uniformly with time; for state-linear realizable maps this decay is necessarily exponential. We extend these results to state-affine, LPV, and linear switched systems via suitable input-forgetting notions, and relate forgetting to decay of impulse responses (sub-Markov parameters). In all cases, the decay/forgetting rate determines the decay rate of every minimal realization.
Read the full abstract

All authors Mihály Petreczky, Juan-Pablo Ortega, Florian Rossmannek, Bálint Daróczy

Signal Processing

LAMBDA: A Low-Altitude Multimodal Base Dataset for UAV Sensing and Communication

Lin Zhou, Peichuan Rao, Chenshuo Zhang et al. (+4 more)

Research on low-altitude integrated sensing and communication (ISAC) requires aligned multimodal data that jointly describe wireless propagation, visual appearance, unmanned aerial vehicle (UAV) motion, light detection and ranging (LiDAR) perception, and radar sensing under common trajectories and timestamps. To address this need, a low-altitude multimodal base dataset, named LAMBDA, is introduced. LAMBDA is characterized by high fidelity, modality diversity, scenario richness, and configuration flexibility. It is generated through a high-fidelity digital-twin pipeline with detailed scene geometry, refined material assignment, and electromagnetic modeling of UAVs. LAMBDA provides synchronized RGB images, depth maps, LiDAR point clouds, inertial measurement unit states, UAV poses, channel state information (CSI), and radar-synthesis resources across matched low-altitude operating conditions, shared coordinate systems, and synchronized frame indices. The dataset covers urban, suburban, and campus scenes, multi-UAV/multi-base-station settings, nighttime conditions, and sunny, rainy, snowy, and foggy weather variations. Its CSI and radar resources support user-defined antenna-array sizes, bandwidths, subcarrier spacings, chirp parameters, and plane-wave or spherical-wavefront channel synthesis. The reliability and usability of LAMBDA are assessed through quality control, weather and multimodal visualization, and two UAV ISAC-related use cases: RGB-aided beam prediction and RGB-LiDAR-based UAV localization.
Read the full abstract

All authors Lin Zhou, Peichuan Rao, Chenshuo Zhang, Jianhua Mo, Shu Sun, Zhiyong Chen, Meixia Tao

Audio and Speech Processing

Probing Low-Level Acoustic Attribute Encoding in CLAP Audio Embeddings

Héctor Martel, Joe Hennessy-Priest and Taemin Cho

Audio foundation models are widely adopted as general-purpose feature extractors, yet the internal structure of their learned representations remains insufficiently understood. In this work, we analyze CLAP audio embeddings through a probing framework, studying the encoding of three fundamental perceptual dimensions: reverberation (RT60), loudness (LUFS), and spectral content, measured via spectral centroid (SC) and relative pitch (RP). Probes of increasing complexity are trained to predict each attribute from frozen embeddings across five datasets spanning noise, speech, monophonic musical notes, and music mixtures. Our primary finding is that all of these attributes are reliably recoverable from the CLAP embedding space across the examined datasets. Within this global picture, two encoding regimes emerge: RT60, LUFS, and RP are approximately linearly encoded, while SC requires non-linear probes. Both regimes generalize across eight additional audio foundation models, with the notable exception that amplitude-invariant architectures discard loudness entirely by construction. The identified linear feature directions are geometrically consistent across datasets for RT60 and LUFS, while highly domain-specific for RP. Finally, we provide a qualitative demonstration of cross-modal consistency, showing that text embeddings of acoustic descriptors align geometrically with the identified RT60 feature direction.
Read the full abstract

Image and Video Processing · Cross-listed from cs.CV

Probabilistic Robustness in Medical Image Classification

Yi Zhang, Siddartha Khastgir and Xingyu Zhao

Deep learning (DL) has shown strong performance in medical image classification, but its trustworthy deployment remains challenging in safety-critical clinical settings, where prediction errors under perturbations may lead to severe consequences. Existing studies mainly focus on adversarial robustness (AR) from a worst-case perspective; however, such settings may be less representative of real medical applications. In this work, we investigate probabilistic robustness (PR) as a more practical measure of model trustworthiness. To this end, we construct a set of natural corruption settings for medical image classification and systematically evaluate commonly used DL models on MedMNIST v2 dataset. Our study provides a statistically grounded perspective on assessing the trustworthiness of DL models, thereby supporting their more trustworthy deployment in medical imaging applications.
Read the full abstract

Systems and Control

Handover-Aware Trajectory Planning for Cellular-Connected UAVs under STL Specifications and URLLC Constraints

Yuqi Ping, Tianhao Liang, Bofeng Zheng et al. (+2 more)

This paper investigates handover-aware trajectory planning for cellular-connected UAVs executing mission-centric tasks under ultra-reliable low-latency communication (URLLC) constraints. Signal temporal logic (STL) provides a formal specification layer for translating mission semantics into time-bounded trajectory requirements, while finite-blocklength URLLC feasibility characterizes reliable command-and-control (C2) links with serving base stations (BSs). We formulate a joint planning problem that optimizes the UAV trajectory, STL mission satisfaction, serving-BS association, and handover behavior. To solve this mixed discrete-continuous problem, we adopt and integrate a Logic Network Flow (LNF) based STL reformulation with Bézier-parameterized motion, disk-shaped URLLC service regions, and binary association variables, so that the resulting mixed-integer quadratically constrained formulation can be solved by standard branch-and-bound solvers. Numerical simulations over a library of STL missions show that the proposed planner can execute different mission specifications under the same cellular map while maintaining URLLC feasibility. The results further reveal how mission timing, handover-aware association, and finite-blocklength stringency jointly affect trajectory shape, serving margin, and computational complexity.
Read the full abstract

All authors Yuqi Ping, Tianhao Liang, Bofeng Zheng, Bingyan Xie, Tingting Zhang

Image and Video Processing

GALOSH: Blind, Training-Free Denoising of Raw Bayer and sRGB Images by Parallel-Friendly Local Shrinkage

Yoshiro Sato

Classical training-free denoisers such as BM3D and non-local means owe much of their strength to search: content-dependent block matching whose memory traffic and data-dependent control flow parallelize poorly and preclude fixed-latency implementations. Learned denoisers reach the highest quality, but they need training data, degrade outside their training domain (which we also observe), and carry per-pixel compute budgets that effectively require a GPU. We present GALOSH (Generalized Anscombe LOcal SHrinkage), a redesign of training-free denoising that removes the search entirely and aims at multi-domain coverage, speed, and quality at once: a blind per-image Poisson-Gaussian noise fit, a generalized Anscombe transform, a two-pass local Walsh-Hadamard shrinkage of luminance, and a luminance-guided local regression of chrominance – two deliberately different operators for the two perceptually different noise components, each with its own strength control. Every stage is local, data-independent, and regular – the same computation graph for every pixel of every image. One core serves two domains: raw Bayer mosaics and sRGB/YUV images. On four real-noise benchmarks (SIDD Medium and RawNIND, raw and sRGB) GALOSH is consistently the strongest among the tested blind, training-free methods – surpassing BM3D- and NLM-family baselines even when those are given an oracle noise level – and approaches trained networks on raw data while remaining below in-domain trained networks at high ISO in sRGB. Being search-free makes it fast: 7x-650x faster than the DL baselines on the same GPU at full benchmark size, and the only strong method in the comparison that also runs practically on plain CPUs. The fixed, data-independent structure is designed to map naturally onto fixed-point and streaming hardware, supported by an operation-count analysis and a working INT16 fixed-point realization.
Read the full abstract

Image and Video Processing

Deep Learning-Based Characterization of Detonation-Cell Size Distributions in Soot-Foil Records

Mingyang Bu, Robson A. Schneider, Karl P. Chatelain et al. (+5 more)

The geometric size and regularity of detonation cells are key physical parameters for characterizing detonation waves. Traditional manual measurement of soot foils is time-consuming and subjective, while existing computer vision techniques often exhibit poor generalization on real experimental images with high noise, blurred boundaries, and severe overlapping. To address this, we propose a novel method for automated recognition and high-order feature extraction of detonation cells based on deep learning instance segmentation (Mask R-CNN). By constructing a custom heterogeneous dataset (numerical simulations and physical experiments) and integrating transfer learning, the model achieves accurate pixel-level mask prediction within highly noisy flow fields. Results indicate high pixel-level agreement in benchmark validations and strong robustness against noise in complex real-world soot foils. Predicted average cell sizes agree well with manual measurements, yielding relative errors under 2% and 3.5% for regular and irregular conditions, respectively. Sensitivity ablation experiments confirm the model's scale adaptability and guided the establishment of a standardized preprocessing paradigm for appropriate image patching. Overcoming the limitation of extracting only global average sizes, this model achieves automated tracking of the transient spatial evolution of cell sizes along the propagation direction. Furthermore, it quantitatively extracts high-order regularity features, such as the irregularity index (RI) and standard deviation of cell deflection angles, demonstrating consistency with theoretical expectations. The proposed method enhances the efficiency and objectivity of statistical analysis, providing a powerful data extraction tool for experimental and numerical soot foils.
Read the full abstract

All authors Mingyang Bu, Robson A. Schneider, Karl P. Chatelain, Mhedine Alicherif, Yingchen Shi, Andrés Z. Mendiburu, Deanna A. Lacoste, Bing Wang

Image and Video Processing · Cross-listed from q-bio.QM

Triple-Phase Multimodal Knowledge Aggregation Framework for Microbial Keratitis Subtype Diagnosis on Slit-Lamp Photography

Yiqing Wang, Maria A. Woodward, Ziyun Yang et al. (+11 more)

Microbial keratitis requires rapid pathogen identification to guide treatment, but culture- and PCR-based diagnostics are slow and resource-intensive. We developed a triple-phase multimodal framework for bacterial-versus-fungal keratitis classification using slit-lamp photographs acquired under blue-light, sclerotic-scatter, and white-light illumination, together with clinical metadata. The model combines cross-modality contrastive learning, modality-specific fine-tuning, and feature-level multimodal ensemble learning for patient-level prediction. We evaluated the framework on a multicenter dataset of 1,645 patients and 17,158 images from India and the United States. The model achieved 85.84% accuracy, 84.46% average F1-score, and 0.885 AUC. Site-specific evaluation showed that pooled results were overly optimistic, whereas resampling- and balance-based re-evaluation provided a more realistic assessment of cross-site generalization. Under all settings, our framework remained the top-performing approach. The code is available at https://github.com/yqwang01/TPMKA and dataset access will be provided subject to University of Michigan data-sharing clearance.
Read the full abstract

All authors Yiqing Wang, Maria A. Woodward, Ziyun Yang, N. Venkatesh Prajna, Chunming He, Leslie M. Niziol, Mercy Pawar, Ming-Chen Lu, Guillermo Amescua, Rachel Wozniak, Sejal Amin, Abinaya Krishnan, Prabhleen Kochar, Sina Farsiu

Systems and Control

Learning Robust Control Lyapunov Functions through Lipschitz Neural Networks

Shiqing Wei, Prashanth Krishnamurthy and Farshad Khorrami

This work presents a novel framework for learning robust control Lyapunov functions and stabilizing controllers for nonlinear dynamical systems subject to additive disturbances upper bounded by a state-dependent function. We leverage recent advances in Lipschitz neural networks to jointly learn both the Lyapunov functions and state-feedback controllers. We establish explicit bounds on the Hessian and third-order derivatives of these neural networks in the spectral norm, and introduce a GPU-friendly branch-and-bound algorithm that utilizes higher-order bounds to significantly accelerate the verification of the Lyapunov conditions. Finally, we validate the proposed approach through extensive simulations on six different dynamical systems.
Read the full abstract

Signal Processing

Virtual Reality-Simulated Interaction Between Micro-Mobility Vehicles and Pedestrians: A Biomechanical Analysis of Human Gait and Movement Responses

Sahan Siriwardena, Omar Aboelrous, Anas Madkoor et al. (+7 more)

Pedestrian walking is a fundamental activity of daily living and a key component of first and last-mile urban mobility. The rapid adoption of e-scooters has increased pedestrian-vehicle interactions on shared sidewalks and crossings, raising collision risks. However, most previous studies have relied on trajectory-based observations, providing limited insight into biomechanical gait responses. This study investigated pedestrian gait adaptations during simulated e-scooter interactions using immersive virtual reality (VR) and markerless pose estimation. Twelve healthy male university students (21-23 years) completed four VR walking scenarios: normal walking, e-scooter encounters at 10-25 km/h, crossing encounters, and near-crash encounters. Sagittal-plane videos were analyzed using the OpenPose 25-point model. Step length, gait cycle time, walking velocity, stance and swing phases, and lower-limb joint trajectories were extracted using Kinovea and custom JSON-based analysis tools. Statistical analyses included ANOVA, MANOVA, and non-parametric tests Crossing and near-crash scenarios significantly reduced step length (p<0.001), from 226.5 cm during normal walking to 204.7 cm during near-crash simulations. Although gait velocity and timing were not significantly affected, participants consistently exhibited shorter stance phases, longer swing phases, and restricted knee motion during stressful encounters, indicating reflexive gait adaptations to perceived collision risk. These findings demonstrate that immersive VR combined with markerless pose estimation effectively quantifies pedestrian biomechanical responses to micro-mobility interactions. Gait adaptations identified in this study may serve as sensitive indicators of collision risk and support the development of proactive pedestrian safety measures and intelligent micro-mobility control systems.
Read the full abstract

All authors Sahan Siriwardena, Omar Aboelrous, Anas Madkoor, Mohamed Elansari, Aiman Alhetari, Jassim Al Jufairi, Mohammed Al-Kuwari, Qinaat Hussain, Charitha Dias, Ajith Sominanda Herath

Systems and Control · Cross-listed from cs.RO

Lost in Time? Continuous Symmetry and Identifiability in Aided Inertial Navigation with Unknown Measurement Delays

Jonathan Kelly, Phone Thiha Kyaw and Matthew Giamou

In many multisensor systems, measurements from different sensors are subject to unknown relative time delays. Accurate state estimation requires that delays be accounted for and, when possible, calibrated online. We consider the case of aided inertial navigation, where measurements from a single aiding sensor are subject to an unknown but constant delay relative to the inertial measurement stream, and study the identifiability of the resulting system. Critically, identifiability depends not only on the temporal structure of the measurements, but also on the shape of the vehicle trajectory: some trajectories are sufficiently informative to support unique recovery of the delay and the navigation state, while others are not. Using the special Galilean Lie group, we characterize a broad family of uninformative trajectories, each generated by a constant element of the Galilean Lie algebra. We show that, along any such trajectory, the delayed measurement model admits a continuous symmetry that prevents unique recovery of the delay and the navigation state. We connect this symmetry-based characterization to the familiar linearized, Jacobian-based analysis. Although our development is motivated by aided navigation, the underlying ideas apply more generally to estimation problems on Lie groups with delayed measurements.
Read the full abstract

Audio and Speech Processing

CHILDES-Aligned: A Curated Children's Speech Dataset via Multi-Model Timestamp Ensembling

Haolong Zheng, Yuanzhuo Hu, Xinyu Liang et al. (+7 more)

CHILDES is a large-scale child speech corpus containing long-form recordings of naturalistic child-adult interactions, making it a valuable resource for studying child speech and language development. However, utterance-level timestamps provided in this corpus are often noisy, incomplete, or misaligned with the audio. As a result, utterances cannot always be reliably localized within long recordings, which limits the direct use of these data for training and evaluating speech models. In this work, we propose BEACON (Boundary Estimation via Alignment CONsensus), an ensemble timestamp-curation framework that refines utterance-level timestamps by aggregating knowledge from multiple off-the-shelf ASR models. Specifically, each model's word-level timestamp predictions are first aligned to provided human transcripts, and the final utterance time boundaries are determined by a consensus voting strategy. The framework is corpus-agnostic and applies to any long-form recording paired with a trusted transcript whose timestamps are unreliable or missing, offering a general recipe for timestamp curation. Leveraging this pipeline, we curate and release a 413-hour general-purpose child-speech dataset with corrected utterance-level timestamps, together with a 283-hour quality-controlled subset for ASR training. Fine-tuning on this subset yields up to an average 19.5% relative WER reduction on four out-of-domain child-speech benchmarks.
Read the full abstract

All authors Haolong Zheng, Yuanzhuo Hu, Xinyu Liang, Vishal Sunder, Dancheng Liu, Jinjun Xiong, Samuel Thomas, Brian Kingsbury, Zhizheng Wu, Mark A. Hasegawa-Johnson

Audio and Speech Processing

TRACE-EVC: Text-Guided Relative Affective Control for Zero-Shot Emotional Voice Conversion

Zihan Zhang, Shreeram Suresh Chandra, Zongyang Du et al. (+5 more)

Traditional emotional voice conversion (EVC) conditions generation on explicit target emotions like labels or references, defining the target affective state but omitting the direction or nature of the transition. We introduce instruction-guided relative emotional voice conversion, a task where natural-language instructions specify source-conditioned affective transformations (e.g., "make the speech slightly calmer" or "sound noticeably more confident") instead of fixed targets. To support this task, we construct TRACE-Instruct, a dataset of relative emotion instructions covering categorical transitions, intensity modifications, and open-ended affective changes. We propose TRACE-EVC, a framework built around Emo-Compass, a module that models each conversion as a source-anchored rectified flow. Rather than conditioning on an explicit target, it predicts the direction and degree of the affective change. Experiments demonstrate that TRACE-EVC accurately follows relative emotion instructions while preserving speaker identity, linguistic content, and speech quality, generalizes to unseen speakers, and remains competitive with conventional EVC systems on standard categorical emotion conversion.
Read the full abstract

All authors Zihan Zhang, Shreeram Suresh Chandra, Zongyang Du, Xiutian Zhao, Aurosweta Mahapatra, Hao Zhang, Philipp Koehn, Berrak Sisman

Audio and Speech Processing

QuaSR: Quality-Aware Sample Reweighting for Pacific Indigenous Speech Recognition

Yishun Li, Yang Xiao, Gongping Huang et al. (+3 more)

Training automatic speech recognition (ASR) models for low-resource languages is challenging due to limited data and highly variable supervision quality. In particular, Pacific Indigenous speech corpora often exhibit heterogeneous acoustic conditions, transcript inconsistencies, and varying degrees of acoustic-text alignment reliability, making standard fine-tuning approaches sensitive to noisy or misleading supervision signals. In this work, we propose QuaSR, a simple yet effective weighting framework that combines data-side reliability with model-side learnability to improve ASR adaptation. Specifically, we estimate data reliability from acoustic, transcription, and alignment, while measuring learnability using training loss from the model. These two complementary signals are integrated into a unified sample utility score to produce training weights for the samples. We also evaluated across four Pacific Indigenous languages, which shows that the proposed utility scores reliably correlate with adaptation performance. Furthermore, QuaSR consistently improves ASR adaptation over standard fine-tuning and alternative data selection strategies, highlighting a new way to leverage difficulty scores for low-resource speech learning.
Read the full abstract

All authors Yishun Li, Yang Xiao, Gongping Huang, Eun-Jung Holden, Nick Thieberger, Ting Dang

Friday 3 July

Systems and Control · Cross-listed from q-bio.NC

Real-time closed-loop protocol to assess neural variability in temporal coding

Alberto Ayala, Angel Lareo, Pablo Varona et al. (+1 more)

Understanding temporal coding in neural systems is essential for decoding brain communication and advancing knowledge of neural information processing. Neural activity often conveys information through spike sequences with stereotypical temporal structures linked to specific functions. However, these sequences are subject to variability introduced by neural dynamics. Real-time closed-loop stimulation is a powerful approach to study temporal coding through adaptive control. In this work, we evaluate how a closed-loop protocol adapts to this variability to drive neural dynamics toward a desired state. It computes the Victor-Purpura distance to quantify similarity between spike sequences generated by the neural system and a triggering pattern. If the protocol determines that a neural sequence is similar to the trigger pattern, it applies stimulation to the system. This allows for an analysis of whether the system's responses are consistent and facilitates the identification of varying spike sequences that can be considered instances of the same functional temporal code. We designed two validation experiments using the Hindmarsh-Rose model: (i) detection of a temporal code and delivery of stimulation to produce brief interspersed bursts, and (ii) detection of burst onset in chaotic activity followed by inhibitory stimulation to regularize it. Gaussian noise was progressively injected to increase variability. The protocol exhibited high degree of adaptability to variability and was effective in achieving the target dynamics. The results reported in this paper suggest that adaptive closed-loop stimulation can enhance experimental methodologies for studying neural coding under realistic variability conditions.
Read the full abstract

All authors Alberto Ayala, Angel Lareo, Pablo Varona, Francisco B. Rodriguez

Systems and Control · Cross-listed from cs.AI

Reinforcement Learning for Heterogeneous Sensor Selection in Maritime Surveillance

Andrei Starodubov, Yaqub Aris Prabowo, Andreas Hadjipieris et al. (+2 more)

This paper presents an information-gain-guided reinforcement-learning sensor-selection framework for single-vessel tracking in heterogeneous maritime sensor networks. The proposed approach is motivated by information-theoretic sensor management: instead of activating all sensors or repeatedly performing computationally expensive online expected-information-gain evaluation, a learned policy selects one tracking-relevant sensor at each decision epoch. A Bayesian sequential Monte Carlo tracker estimates the vessel state from noisy measurements and provides a belief representation for scheduling under nonlinear and non-Gaussian conditions. A Proximal Policy Optimization agent selects one of five sensors in a georeferenced simulation of the CMMI Smart Marina testbed at Ayia Napa Marina, Cyprus. The policy is trained on the testbed's actual five-sensor configuration. The agent observes belief-state, detection-history, coverage, sensor-geometry, and realized-information-gain features. The reward is defined as a realized-information-gain term gated by an observability mask. Final-test simulations compare the proposed framework with random single-sensor selection, always-on sensing using all sensors simultaneously, and the expected-information-gain sensor-selection baseline proposed in our previous work. Results show that the learned policy achieves tracking performance close to always-on sensing while activating only one sensor per decision time step and avoiding the computationally expensive online entropy search required by expected-information-gain selection. Additional zero-shot evaluation without retraining on ten moderately perturbed versions of actual layout configuration showed broadly stable tracking, with any increase in positional tracking error remaining below 1 meter across all perturbations.
Read the full abstract

All authors Andrei Starodubov, Yaqub Aris Prabowo, Andreas Hadjipieris, Roberto Galeazzi, Ioannis Kyriakides

Signal Processing

MorphologyFM: A Foundation Model for Morphology-Aware Representation Learning from ECG and Pulse Oximetry Waveforms

Saiyang Feng, Yuanyun Zhang and Shi Li

Foundation models have recently emerged as a powerful paradigm for learning transferable representations from large scale biomedical data, yet existing approaches for physiological waveforms primarily optimize reconstruction or forecasting objectives that do not explicitly preserve clinically meaningful waveform morphology. Electrocardiograms (ECGs) and pulse oximetry (SpO2) waveforms encode rich cardiovascular and hemodynamic information through their morphological structure. In this work, we introduce MorphologyFM, a multimodal foundation model pretrained on paired ECG and SpO2 waveforms from the MIMIC critical care database using a morphology aware self supervised learning objective. MorphologyFM combines morphology guided masking, cross modal representation learning, and contrastive latent alignment to learn representations that capture clinically relevant physiological structure without requiring manual annotations. We evaluate MorphologyFM across multiple downstream prediction tasks, including arrhythmia classification, hypoxemia prediction, mortality prediction, and length of stay estimation, demonstrating consistent improvements over representative self supervised learning methods, including Masked Autoencoders (MAE), contrastive learning, Barlow Twins, and Joint Embedding Predictive Architectures (JEPA). Furthermore, we show that jointly modeling ECG and SpO2 waveforms produces more transferable representations than single modality pretraining. Our results establish waveform morphology as a powerful inductive bias for self supervised physiological representation learning and introduce MorphologyFM as a general purpose foundation model for continuous physiological monitoring.
Read the full abstract

Image and Video Processing · Cross-listed from q-bio.QM

Longitudinal MRI template of the baboon brain from birth to adolescence

Katherine L. Bryant, Arnaud Le Troter, David Meunier et al. (+8 more)

The baboon (Papio) is an invaluable resource within nonhuman primate research, having the advantage of being a cercopithecoid (Old World monkey) with one of the largest brains among non-hominid primates. In order to facilitate comparative developmental neuroscience research, we present the BABACOOL (BAby Brain Atlas COnstruction for Optimized Labeled segmentation) approach for creating multi-modal developmental atlases, which we used to produce BaBa21, a population-based longitudinal developmental baboon template. BaBa21 is a spatio-temporal template that consists of structural (T1- and T2-weighted) images and tissue probability maps from a population of 21 baboons (Papio anubis) scanned at 4 timepoints beginning from about 2 weeks after birth and continuing to sexual maturity (5 years). Further, his study offers a fully automatic method for generating a template at any intermediate age for future age-specific group studies. This resource is made available to provide a normalization target for baboon data across the lifespan, including intermediate timepoints, and moreover facilitate neuroimaging research in baboons, comparative research with humans and nonhuman primate species for which developmental templates are available (e.g., macaques).
Read the full abstract

All authors Katherine L. Bryant, Arnaud Le Troter, David Meunier, Yannick Becker, Scott A. Love, Siham Bouziane, Kep Kee Loh, Julien Sein, Luc Renaud, Olivier Coulon, Adrien Meguerditchian

Systems and Control · Cross-listed from econ.EM

Retrieval over Reasoning: A Cost-Controlled Benchmark of Language Models for Energy-Retrofit Recommendation

Eliseo Curcio

Recommending the correct set of energy conservation measures (ECMs) for a building is a structured, multi-label prediction problem in which a task-specific supervised model has weak training signal and a general language model has no grounding in the local building stock. We study this problem on 10,422 real New York City Local Law 87 (LL87) energy-audit records, taking as ground truth the set of ECM categories that certified auditors actually recommended. We make four contributions. First, we establish that energy-use-intensity (EUI) prediction - the upstream task - is effectively solved by tree ensembles: across fifteen trained models, a stacking ensemble reaches a coefficient of determination R^2 = 0.757, and every one of six neural architectures is outperformed by gradient-boosted trees. Second, we show that the framing of the recommendation task dominates model choice: recasting ECM recommendation as 19-way multi-label classification rather than single-label categorization lifts a gradient-boosted-tree baseline from a previously reported 25.9% accuracy to a micro-F1 of 0.571. Third, we benchmark eight large language models (LLMs) from four providers in a 2x2 design that independently toggles retrieval grounding and explicit reasoning, scoring each arm on per-label F1, U.S.-dollar cost per building, and latency; retrieval-augmented generation (RAG) improves micro-F1 by +0.11 to +0.20 on every model, while explicit reasoning yields no measurable accuracy change (-0.018 to +0.010) at up to 8.4x the cost. Fourth, we show LLMs systematically over-recommend - high recall, low precision - and that retrieval closes the gap chiefly by improving precision. A 70-billion-parameter open-weight model with a fifteen-line nearest-neighbor retrieval step reaches 0.511 micro-F1 at $0.00032 per building, comparable to a frontier model at roughly 10.1x lower cost.
Read the full abstract

Image and Video Processing

Model Confidence-Guided Multi-Image Fusion of Fundus Images for Diabetic Retinopathy Diagnosis

Ananya Raghu, Anisha Raghu, Alice S. Tang et al. (+3 more)

Purpose: Early screening for eye diseases is critical in low- and middle-income countries where access to care is limited. We investigate whether a confidence-guided, multi-image diabetic retinopathy diagnosis framework can integrate image filtering with confidence-aware predictions for reliable screening at capture. Methods: We develop a multi-image fusion method that aggregates retinal views to improve confidence and balanced accuracy. Our method uses confidence to identify unreliable predictions, prompting retakes when needed. We compare: (1) a cascaded image-quality and disease diagnosis pipeline using a single image per patient, (2) confidence-based prediction, and (3) our confidence-based multi-image fusion pipeline. All methods are evaluated using a RETFoundGreen backbone on the mBRSET (n = 1,234) and BRSET (n = 7,599) datasets. Results: At 70% coverage, our method achieves 91% balanced accuracy on mBRSET and 97% on BRSET, improvements of ~12% and ~6%, respectively, over cascade filtering. The image-quality cascade reaches sensitivities of 61% on mBRSET and 86% on BRSET, whereas our framework reaches 94% and 96%, respectively, at 50% coverage. Conclusions: Human-annotated quality labels are weakly associated with diagnostic performance, and confidence-based filtering consistently outperforms image quality-based cascaded pipelines. Translational Relevance: Using confidence-based multi-image fusion, patients receive more reliable predictions, reducing incorrect diagnoses during screening. The lightweight backbone and single inference pass per image make the framework compatible with low-latency mobile screening systems in resource-limited settings.
Read the full abstract

All authors Ananya Raghu, Anisha Raghu, Alice S. Tang, Yannis M. Paulus, Tyson N. Kim, Tomiko T. Oskotsky

Signal Processing

STAR-RIS-Assisted Integrated Sensing, Secure Communication, and Power Transfer: A Transmit Power Minimization Framework

Ling He, Vaibhav Kumar, Yingyang Chen et al. (+3 more)

Evolving wireless networks call for architectures that unify sensing, communication, and wireless power transfer. Although integrated sensing and communication (ISAC) and simultaneous wireless information and power transfer (SWIPT) have validated dual-function transmission, the combination of integrated sensing, secure communication, and power transfer (ISSCPT) remains largely unexplored, in part due to the tight coupling among design variables. To address this coupling and expand spatial degrees of freedom, we turn to intelligent metasurfaces: while a conventional reconfigurable intelligent surface (cRIS) reflects only to one side and thus limits coverage and flexibility, a simultaneously transmitting and reflecting RIS (STAR-RIS) enables full-space wave control, making it a natural vehicle for power-efficient ISSCPT. We study a STAR-RIS-assisted ISSCPT system and pose a central question: How much transmit power is required to operate such a system? We formulate a transmit-power minimization problem that jointly optimizes transmit and receive beamforming and the STAR-RIS configuration, and solve it via alternating optimization with successive convex approximation, second-order cone programming, and eigenvalue decomposition. Simulations show that the proposed STAR-RIS-assisted design outperforms cRIS and no-RIS baselines, and quantify the additional transmit power required by ISSCPT relative to ISAC and secure SWIPT, clarifying security-sensing-power tradeoffs in metasurface-assisted systems.
Read the full abstract

All authors Ling He, Vaibhav Kumar, Yingyang Chen, Miaowen Wen, Christina Pöpper, Marwa Chafii

Systems and Control

On Determining the Convergence Rate of an Infinite Product of Stochastic Matrices

Ron Ofir and A. Stephen Morse

By a convergent set is meant a set of stochastic matrices where every infinite product of matrices from every compact subset converges to a rank one matrix. Well-known examples include the set of all scrambling matrices, the set of all stochastic matrices with all diagonal entries positive and a rooted graph, the set of all Sarymsakov matrices, and the set of doubly stochastic matrices with positive diagonal entries and a weakly connected graph. It is known that every infinite product from each compact set of every convergent set converges to its limit exponentially fast, but not much is known about the rate of convergence when not all matrices involved are scrambling matrices. This paper deals with bounding the rate of convergence in convergent sets using submultiplicative seminorms. It is shown that only in some convergent sets all matrices are contractions in the same seminorm, and in particular that this method cannot be used to determine the convergence rate for the class of matrices with positive diagonal entries and a rooted graph. As a second contribution, it is shown that for every compact convergent set and every submultiplicative seminorm, there is a finite number $k$ such that all products of $k$ matrices from the set are contractions in the seminorm. Finally, several open questions are posed for future research.
Read the full abstract

Signal Processing

Two-dimensional Fourier compressed sensing under a fixed readout budget per channel

Nitin Jonathan Myers

Recovering sparse signals from their subsampled Fourier representation is an important problem in communications, radar, and imaging. In this letter, we focus on reconstructing sparse 2D signals (matrices) under the constraint that only a fixed number of entries can be sampled from each channel, e.g., a row or a column in the Fourier domain. For a specified per-channel readout budget, we derive a lower bound on the mutual coherence of the corresponding compressed sensing matrix. We show that our bound is larger than the classical Welch bound, due to a limited readout budget. We also construct deterministic subsampling patterns that attain this bound for a class of matrix dimensions and readout budgets, and benchmark them against random subsampling through simulations.
Read the full abstract

Signal Processing

Improving the clinical utility of lower-limb surface electromyography (sEMG) by quantifying and correcting for location changes in inter-session recordings

Fraser Douglas, Mona Pei, Quoc Sy Vu et al. (+2 more)

Purpose: Surface electromyography (sEMG) can enable direct muscle activity measurement to support the recovery assessment of individuals with neurological and musculoskeletal disorders. Despite this, its broader adoption of sEMG has been limited given its sensitivity to changes in electrode location across sessions. To address this challenge and enable multi-session sEMG, this work develops a novel high-density sEMG (HDsEMG) algorithm to quantify changes in electrode location and mitigate its effects on common time and frequency domain sEMG features. Methods: 11 healthy participants performed isometric and dynamic exercises with HDsEMG on four lower limb muscles. These were repeated four times, reapplying arrays at shifted locations. The error between spatially-mapped HDsEMG metrics was then minimised to estimate the change in array location, with this compared against ground truth 3D scans. Lastly, relative feature differences across locations were computed at select electrodes to assess the degree to which inter-session sEMG effects were mitigated. Results: Electrode location estimates were improved over the assumption their location remained unchanged in 81.7% of cases, 37.6% identified within 1 cm of the ground truth. Feature differences computed between closest electrodes across locations per ground truth and algorithm estimates were statistically similar. Conversely, feature differences for the same electrode across locations were significantly greater, increasing the mean difference for the isometric max envelope amplitude from 15.9% with the algorithm to 21.1% without. Conclusions: The algorithm's application reduced inter-session feature differences arising from changes in electrode location. This can facilitate more direct cross-session feature comparisons, representing a promising step toward robust sEMG measurement for musculoskeletal and neurological recovery tracking.
Read the full abstract

All authors Fraser Douglas, Mona Pei, Quoc Sy Vu, Linh Le, Calvin Kuo

Image and Video Processing

An Interpretable Deep Learning Framework for Discovery and Clinical Validation of Deep Radiomic Signatures in Tumor Classification

Chengkun Sun, Jinqian Pan, Renjie Liang et al. (+7 more)

Imaging signatures are quantitative features extracted from medical images that provide clinically meaningful information for tumor diagnosis, characterization, prognosis, and treatment planning. Although deep learning has shown great potential for imaging signature discovery, its limited interpretability remains a major barrier to clinical adoption. Existing approaches often achieve high predictive performance but provide little biological insight into the identified signatures. We propose a unified framework for interpretable imaging signature discovery by integrating deep learning based segmentation, explainable classification, and radiomic analysis. A robust segmentation model is first used to accurately delineate tumors, followed by a Grad-CAM guided pipeline that identifies diagnostically important regions as candidate imaging signatures. A mutual information based adaptive thresholding strategy enables patient-specific signature extraction. The resulting signatures are validated using a downstream deep learning classification model, while radiomic features extracted from the signature regions are evaluated with traditional machine learning models and interpreted using SHAP to identify the most discriminative biomarkers. The proposed framework is evaluated on the public BUSI breast ultrasound, KiTS renal CT, and BraTS brain tumor datasets, as well as a private UF Health renal CT cohort. Compared with conventional whole-tumor radiomics, the proposed signature-based approach achieves improved discriminative performance while providing greater biological interpretability. By converting deep learning attention into reproducible quantitative imaging biomarkers, this framework offers an interpretable and reproducible solution for non-invasive tumor characterization and imaging biomarker discovery.
Read the full abstract

All authors Chengkun Sun, Jinqian Pan, Renjie Liang, Zhengkang Fan, Xin Miao, Yi Guo, Mei Liu, Muxuan Liang, Russell Terry, Jie Xu