Image and Video Processing
Digital Elevation Model Estimation from RGB Satellite Imagery using Generative Deep Learning
Alif Ilham Madani, Riska A. Kuswati, Alex M. Lechner et al. (+1 more)
Digital Elevation Models (DEMs) are vital
datasets for geospatial applications such as hydrological modeling and environmental monitoring. However, conventional methods to generate DEM, such as using
LiDAR and photogrammetry, require specific types of data that are often inaccessible in resource-constrained settings. To alleviate this problem, this study proposes an approach to generate DEM from freely available RGB satellite imagery using generative deep learning, particularly based on a conditional Generative
Adversarial Network (GAN). We first developed a global
dataset consisting of 12K RGB-DEM pairs using Landsat satellite imagery and NASA's SRTM digital elevation data, both from the year 2000. A unique preprocessing pipeline was implemented to select high-quality, cloud-free regions and aggregate normalized RGB composites from Landsat imagery. Additionally, the model was trained in a two-stage process, where it was first trained on the complete
dataset and then fine-tuned on high-quality samples filtered by Structural Similarity Index Measure (SSIM) values to improve performance on challenging terrains. The results demonstrate promising performance in mountainous regions, achieving an overall mean root-mean-square error (RMSE) of 0.4671 and a mean SSIM score of 0.2065 (scale -1 to 1), while highlighting limitations in lowland and residential areas. This study underscores the importance of meticulous preprocessing and iterative refinement in generative modeling for DEM generation, offering a cost-effective and adaptive alternative to conventional methods while emphasizing the challenge of generalization across diverse terrains worldwide.
Read the full abstract ▾
All authors Alif Ilham Madani, Riska A. Kuswati, Alex M. Lechner, Muhamad Risqi U. Saputra
Signal Processing · Cross-listed from cs.LG
MOTIF-RF: Multi-template On-chip Transformer Synthesis Incorporating Frequency-domain Self-transfer Learning for RFIC Design Automation
Houbo He, Yizhou Xu, Lei Xia et al. (+3 more)
This paper presents a systematic study on developing multi-template machine learning (ML) surrogate models and applying them to the inverse design of
transformers (XFMRs) in
radio-frequency integrated
circuits (RFICs). Our study starts with benchmarking four widely used ML architectures, including MLP-, CNN-, UNet-, and GT-based models, using the same
datasets across different XFMR
topologies. To improve modeling accuracy beyond these baselines, we then propose a new frequency-domain self-transfer learning technique that exploits correlations between adjacent frequency bands, leading to around 30%-50% accuracy improvement in the S-parameters prediction. Building on these models, we further develop an inverse design framework based on the covariance
matrix adaptation evolutionary strategy (CMA-ES)
algorithm. This framework is validated using multiple impedance-matching tasks, all demonstrating fast convergence and trustworthy performance. These results advance the goal of AI-assisted specs-to-GDS automation for RFICs and provide RFIC designers with actionable tools for integrating AI into their workflows.
Read the full abstract ▾
All authors Houbo He, Yizhou Xu, Lei Xia, Yaolong Hu, Fan Cai, Taiyun Chi
Systems and Control
Communication-Aware Dissipative Control for Networks of Heterogeneous Nonlinear Agents
Ingyu Jang and Leila J. Bridgeman
Communication-aware control is essential to reduce costs and complexity in large-scale networks. However, it is challenging to simultaneously determine a sparse communication
topology and achieve high performance and robustness. This work achieves all three objectives through dissipativity-based, sparsity-promoting controller synthesis. The approach identifies an optimal sparse structure using either weighted l1 penalties or alternating direction methods of multipliers (ADMM) with a cardinality term, and iteratively solves a convexified version of the NP hard structured optimal control problem. The proposed methods are demonstrated on heterogeneous networks with uncertain and unstable
agents.
Read the full abstract ▾
Systems and Control
Learning Neural Network Safe Tracking Controllers from Backward Reachable Sets
Yuezhu Xu, Mohamed Serry, Jun Liu et al. (+1 more)
The design of tracking controllers that closely follow a reference trajectory while ensuring safety and robustness against disturbances is a challenging problem in the control of autonomous systems. In this work, we propose a neural network-based safe tracking control framework for nonlinear discrete-time systems with reach-avoid specifications in the presence of disturbances. Our approach begins with generation of a nominal trajectory using standard trajectory synthesis approaches, followed by construction of safe zonotopic backward reachable sets along the nominal trajectory. The states lying within the backward reachable sets are guaranteed to satisfy safe reachability specifications. Then, our key insight is to leverage the computed backward reachable sets to inform the architecture and training of a neural network-based tracking controller such that the neural network drives the system's states through these backward reachable sets, thereby improving the likelihood of safe reachability. We perform formal verification with conformal prediction to achieve statistical safety guarantees on the performance of the learned neural controller. The performance of our approach is illustrated through a numerical example on the discrete-time Dubin's car model.
Read the full abstract ▾
All authors Yuezhu Xu, Mohamed Serry, Jun Liu, S. Sivaranjani
Signal Processing · Cross-listed from cs.LG
Deep Learning Architectures for Code-Modulated Visual Evoked Potentials Detection
Kiran Nair and Hubert Cecotti
Non-invasive
Brain-Computer Interfaces (BCIs) based on Code-Modulated Visual Evoked Potentials (C-VEPs) require highly robust decoding methods to address temporal variability and session-dependent noise in
EEG signals. This study proposes and evaluates several deep learning architectures, including convolutional neural networks (CNNs) for 63-bit m-sequence reconstruction and classification, and Siamese networks for similarity-based decoding, alongside canonical correlation analysis (CCA) baselines.
EEG data were recorded from 13 healthy adults under single-target flicker stimulation. The proposed deep models significantly outperformed traditional approaches, with distance-based decoding using Earth Mover's Distance (EMD) and constrained EMD showing greater robustness to latency variations than Euclidean and Mahalanobis metrics. Temporal data augmentation with small shifts further improved generalization across sessions. Among all models, the multi-class Siamese network achieved the best overall performance with an average accuracy of 96.89%, demonstrating the potential of data-driven deep architectures for reliable, single-trial C-VEP decoding in adaptive non-invasive BCI systems.
Read the full abstract ▾
Systems and Control · Cross-listed from cs.NI
Secure Command, Control and Communications Systems (C3) for Army UxVs
T. Rebolo, A. Grilo and C. Ribeiro
Unmanned Vehicles (UxVs) are increasingly used in modern military operations for reconnaissance, surveillance, and strike missions, enhancing situational awareness while reducing risk to personnel. Their affordability and rapid deployment have encouraged the adoption of commercial solutions. However, many rely on insecure protocols such as MAVLink, which lack authentication and encryption mechanisms. This paper designed, implemented, and evaluated a new secure command-and-control architecture that ensures confidentiality, integrity, and authentication (CIA) while supporting real-time control delegation between Ground Control Stations (GCSs). The proposed solution, named New Command and Control System (NC2S), enforces a zero-trust model integrating hierarchical credential-based privileges to regulate access and control among Tactical Commanders (TC), GCSs, and UxVs. It employs mutual Transport Layer Security (mTLS) with
Elliptic Curve Digital Signature
Algorithm (ECDSA) certificates and
Elliptic Curve Diffie-Hellman (ECDH) key exchange, while message integrity is ensured through Hash-based Message Authentication Codes (HMAC). Multiple lightweight protocols were developed for credential management, key renewal, and control handover. The NC2S prototype was experimentally validated over Wi-Fi and Rohde&Schwarz HR-5000H tactical
radios. Results showed that HR-5000H links introduce latencies roughly two orders of magnitude higher than broadband technologies (e.g., Wi-Fi or 5G&Beyond technologies) but are still able to maintain stable communication with minimal message loss, making them suitable for the NC2S links among TC terminals and GCSs.
Read the full abstract ▾
Image and Video Processing
Comparing SAM 2 and SAM 3 for Zero-Shot Segmentation of 3D Medical Data
Satrajit Chakrabarty and Ravi Soni
Foundation models, such as the Segment Anything Model (SAM), have heightened interest in promptable zero-shot
segmentation. Although these models perform strongly on natural images, their behavior on medical data remains insufficiently characterized. While SAM 2 has been widely adopted for annotation in 3D medical workflows, the recently released SAM 3 introduces a new architecture that may change how visual
prompts are interpreted and propagated. Therefore, to assess whether SAM 3 can serve as an out-of-the-box replacement for SAM 2 for zero-shot
segmentation of 3D medical data, we present the first controlled comparison of both models by evaluating SAM 3 in its Promptable Visual
Segmentation (PVS) mode using a variety of prompting strategies. We
benchmark on 16 public
datasets (CT, MRI, Ultrasound, endoscopy) covering 54 anatomical structures, pathologies, and
surgical instruments. We further quantify three failure modes:
prompt-frame over-
segmentation, over-propagation after object disappearance, and temporal retention of well-initialized predictions. Our results show that SAM 3 is consistently stronger under click prompting across modalities, with fewer
prompt-frame over-
segmentation failures and slower prediction retention decay compared to SAM 2. Under bounding-box and mask
prompts, performance gaps narrow in few structures of CT/MR and the models trade off termination behavior, while SAM 3 remains stronger on ultrasound and endoscopy sequences. The overall results position SAM 3 as the superior default choice for most medical
segmentation tasks, while clarifying when SAM 2 remains a preferable propagator.
Read the full abstract ▾
Systems and Control
Quantized Distributed Estimation with Event-triggered Communication and Packet Loss
Ying Wang, Yanlong Zhao, Ji-Feng Zhang et al. (+1 more)
This paper focuses on the problem of quantized distributed estimation with event-triggered communication and packet loss, aiming to reduce the number of transmitted bits. The main challenge lies in the inability to differentiate between an untriggered event and a packet loss occurrence. This paper proposes an event-triggered distributed estimation
algorithm with quantized communication and quantized measurement, in which it introduces a one-bit information reconstruction method to deal with packet loss. The almost sure convergence and convergence rate of the proposed
algorithm are established. Besides, it is demonstrated that the global average communication bit-rate decreases to zero over time. Moreover, the trade-off between communication rate and convergence rate is revealed, providing guidance for designing the communication rate required to achieve the
algorithm's convergence rate. A numerical example is supplied to validate the findings.
Read the full abstract ▾
All authors Ying Wang, Yanlong Zhao, Ji-Feng Zhang, Karl Henrik Johansson
Audio and Speech Processing · Cross-listed from cs.SD
Advancing Marine Bioacoustics with Deep Generative Models: A Hybrid Augmentation Strategy for Southern Resident Killer Whale Detection
Bruno Padovese, Fabio Frazao, Michael Dowd et al. (+1 more)
Automated detection and classification of marine mammals vocalizations is critical for conservation and management efforts but is hindered by limited annotated
datasets and the acoustic complexity of real-world marine environments. Data augmentation has proven to be an effective strategy to address this limitation by increasing
dataset diversity and improving model generalization without requiring additional field data. However, most augmentation techniques used to date rely on effective but relatively simple transformations, leaving open the question of whether deep generative models can provide additional benefits. In this study, we evaluate the potential of deep generative for data augmentation in marine mammal call detection including: Variational Autoencoders, Generative
Adversarial Networks, and Denoising
Diffusion Probabilistic Models. Using Southern Resident Killer Whale (Orcinus orca) vocalizations from two long-term hydrophone deployments in the Salish Sea, we compare these approaches against traditional augmentation methods such as time-shifting and vocalization masking. While all generative approaches improved classification performance relative to the baseline,
diffusion-based augmentation yielded the highest recall (0.87) and overall F1-score (0.75). A hybrid strategy combining generative-based synthesis with traditional methods achieved the best overall performance with an F1-score of 0.81. We hope this study encourages further exploration of deep generative models as complementary augmentation strategies to advance acoustic monitoring of threatened marine mammal populations.
Read the full abstract ▾
All authors Bruno Padovese, Fabio Frazao, Michael Dowd, Ruth Joy
Signal Processing
Internodal Distance Distributions for Static and Mobile Nodes in 2D/3D Wireless Networks
Nicholas Vaiopoulos, Alexander Vavoulas, Harilaos G. Sandalidis et al. (+2 more)
This letter presents a unified analytical framework for internodal distance distributions in 2D and 3D wireless networks, with nodes confined to concentric circular or spherical regions. Four deployment scenarios are considered, covering all combinations of static (uniform) and mobile (random waypoint-based) nodes. For each scenario, closed-form expressions for the internodal distance probability density functions are derived, incorporating both
geometric constraints and spatial effects introduced by mobility. Equal-radius cases are also addressed. Beta-distribution approximations and Monte Carlo simulations demonstrate the accuracy and validity of the analytical results.
Read the full abstract ▾
All authors Nicholas Vaiopoulos, Alexander Vavoulas, Harilaos G. Sandalidis, Konstantinos K. Delibasis, Dimitris Varoutas
Systems and Control
Automated Enumeration of Reconfigurable Architectures for Thermal Management Systems in Battery Electric Vehicles
Reihaneh Jahedan, Satya Peddada, Mark Jennings et al. (+3 more)
As the automotive industry moves towards vehicle electrification, designing and optimizing thermal management systems (TMSs) for Battery Electric Vehicles (BEVs) has become a critical focus in recent years. The dependence of battery performance on operating temperature, the lack of waste combustion heat, and the significant effect of TMS energy consumption on driving range make the design of BEV TMSs highly complicated compared to conventional vehicles. Although prior research has focused on optimizing the configuration of thermal systems for varying ambient conditions, a holistic approach to studying the full potential of reconfigurable TMS architectures has not yet been fully explored. The complex design landscape of multi-mode reconfigurable systems is difficult to navigate. Relying solely on expert intuition and creativity to identify new architectures both restricts progress and leaves significant performance improvements unrealized. In this study, using graph modelling of TMS architectures, we propose a systematic method to automatically enumerate and simulate reconfigurable architectures for a TMS, given the desired operating modes, along with a framework to conduct transient performance analysis and optimization-based trade-off studies among system performance, energy consumption, and complexity. We explored more than 150 operating mode sequences, retaining 39 unique architectures for further evaluation. MATLAB Simscape models of these architectures were automatically created and their performance evaluated. The multi-objective optimization results provide decision support for selecting the best architecture based on user priorities.
Read the full abstract ▾
All authors Reihaneh Jahedan, Satya Peddada, Mark Jennings, Sunil Katragadda, James Allison, Nenad Miljkovic
Systems and Control
LILAD: Learning In-context Lyapunov-stable Adaptive Dynamics Models
Amit Jena, Na Li and Le Xie
System identification in control theory aims to approximate dynamical systems from trajectory data. While neural networks have demonstrated strong predictive accuracy, they often fail to preserve critical physical properties such as stability and typically assume stationary dynamics, limiting their applicability under distribution shifts. Existing approaches generally address either stability or adaptability in isolation, lacking a unified framework that ensures both. We propose LILAD (Learning In-Context Lyapunov-stable Adaptive Dynamics), a novel framework for system identification that jointly guarantees adaptability and stability. LILAD simultaneously learns a dynamics model and a Lyapunov function through in-context learning (ICL), explicitly accounting for parametric uncertainty. Trained across a diverse set of tasks, LILAD produces a stability-aware, adaptive dynamics model alongside an adaptive Lyapunov certificate. At test time, both components adapt to a new system instance using a short trajectory
prompt, which enables fast generalization. To rigorously ensure stability, LILAD also computes a state-dependent attenuator that enforces a sufficient decrease condition on the Lyapunov function for any state in the new system instance. This mechanism extends stability guarantees even under out-of-distribution and out-of-task scenarios. We evaluate LILAD on
benchmark autonomous systems and demonstrate that it outperforms adaptive, robust, and non-adaptive baselines in predictive accuracy.
Read the full abstract ▾
Signal Processing · Cross-listed from physics.med-ph
Effective Hyper-clutter Artifacts Suppression for Ultrafast Ultrasound Doppler Imaging
Lijie Huang, Jingyi Yin, Jingke Zhang et al. (+13 more)
Objective: Hyper-clutter artifacts (HCA), arising from strong tissue reflections or physiological motion, present persistent challenges in ultrafast ultrasound Doppler imaging, often obscuring surrounding small vessel flow signals, especially in fascial regions such as the renal capsule. This study proposes U-profile-based decluttering (UPBD), a robust and computationally efficient method that exploits singular value decomposition (SVD)-derived spatial singular vectors to suppress HCA in ultrafast Doppler imaging. Methods: UPBD analyzes intensity profile of each pixel along the singular-order dimension of the SVD-derived left singular vectors U. A pixel-wise clutter-energy ratio is computed to derive a spatially adaptive declutter weighting map, which is applied to the SVD-filtered flow signals. Results: UPBD was evaluated on multiple in vivo
datasets. Quantitative assessments based on contrast-to-noise ratio (CNR) and contrast-to-tissue ratio (CTR) demonstrated significant improvements over conventional SVD filtering. For example, UPBD enhanced CTR from 7.3 dB to 21.7 dB in contrast-free pig kidney, 17.8 dB to 42.1 dB in contrast-enhanced pig kidney, 8.2 dB to 32.8 dB in human kidney, and -4.9 dB to 3.7 dB in 3D human liver. Conclusion: The proposed UPBD method effectively suppresses HCA while preserving blood flow signals with minimal extra computational cost and no need for extensive parameter tuning. Significance: UPBD serves as a lightweight, easily integrated post-processing method that enhances HCA suppression, enabling broader application of SVD-based ultrafast Doppler imaging.
Read the full abstract ▾
All authors Lijie Huang, Jingyi Yin, Jingke Zhang, U-Wai Lok, Ryan M. DeRuiter, Kaipeng Ji, Yanzhe Zhao, Tao Wu, James D. Krier, Xiang-yang Zhu, Andrew J. Bentall, Andrew D. Rule, Thomas D. Atwell, Lilach O. Lerman, Shigao Chen, Chengwu Huang
Image and Video Processing
Attention-Guided Fair AI Modeling for Skin Cancer Diagnosis
Mingcheng Zhu, Mingxuan Liu, Han Yuan et al. (+4 more)
Artificial intelligence (AI) has shown remarkable promise in dermatology, offering accurate and non-invasive diagnosis of skin cancer. While extensive research has addressed skin tone-related bias, gender bias in dermatologic AI remains underexplored, leading to unequal care and reinforcing existing gender disparities. In this study, we developed LesionAttn, a
fairness-aware
algorithm that integrates clinical knowledge into model design by directing
attention toward lesion regions, mirroring the diagnostic focus of clinicians. Combined with Pareto-frontier optimization for dual-objective model selection, LesionAttn balances
fairness and predictive accuracy. Validated on two large-scale dermatological
datasets, LesionAttn significantly mitigates gender bias while maintaining high diagnostic performance, outperforming existing bias mitigation
algorithms. Our study highlights the potential of
embedding clinical knowledge into AI development to advance both model performance and
fairness, and further to foster interdisciplinary collaboration between clinicians and AI developers.
Read the full abstract ▾
All authors Mingcheng Zhu, Mingxuan Liu, Han Yuan, Yilin Ning, Zhiyao Luo, Tingting Zhu, Nan Liu
Signal Processing
Optimal Bit Detection in Thermal Noise Communication Systems Under Rician Fading
Mohamed El Jbari, Fernando D. A. García, Hugerles S. Silva et al. (+2 more)
Thermal noise communication (TNC) enables ultra-low-power wireless links for Internet of Things (IoT) devices by modulating the variance of thermal noise, rather than using active carriers. Existing analyses often rely on Gaussian approximations and overlook fading effects, which limits their accuracy. This paper presents an accurate analytical framework for optimal bit detection in TNC systems under Rician fading. Using chi-squared statistics, we derive the optimal maximum-likelihood detection threshold and an expression for the bit error probability (BEP) via Gauss-Laguerre quadrature. The proposed model eliminates approximation errors and accurately characterizes performance for finite sample sizes. Monte Carlo simulations confirm the analytical results and demonstrate significant improvements in BEP compared with suboptimal Gaussian-based detection. Furthermore, the influence of key parameters, sample size, resistance ratio, and Rician K-factor, is quantified. The proposed framework provides a solid foundation for designing energy-efficient TNC receivers in future B5G/6G and large-scale IoT systems.
Read the full abstract ▾
All authors Mohamed El Jbari, Fernando D. A. García, Hugerles S. Silva, Felipe A. P. de Figueiredo, Rausley A. A. de Souza
Systems and Control
Model-free practical PI-Lead control design by ultimate sensitivity principle
Michael Ruderman
Practical design and tuning of feedback controllers has often to get by without a model of the dynamic process at hand. Only some general assumptions about the system dynamics, in this work type-one stable, can be available for engineers, for instance in motion control applications and many others. This paper proposes a practical and simple in realization procedure for designing a robust PI-Lead control without modeling. The developed method derives from the ultimate sensitivity principles, known in empirical Ziegler-Nichols tuning of PID controllers, and makes use of some general characteristics of the loop shaping. A three-steps procedure is proposed to determine the integration time constant, control gain, and Lead-element in a way to guarantee a sufficient phase margin, while all steps are served by only experimental monitoring of the output value. Proposed method is demonstrated and discussed with experiments accomplished on a noise-perturbed electro-mechanical actuator system.
Read the full abstract ▾
Systems and Control
Bang-Bang Evasion: Its Stochastic Optimality and a Terminal-Set-Based Implementation
Liraz Mudrik and Yaakov Oshman
We address the problem of optimal evasion in a planar endgame engagement, where a target with bounded lateral acceleration seeks to avoid interception by a missile guided by a linear feedback law. Contrary to existing approaches, that assume perfect information or use heuristic maneuver models in
stochastic settings, we formulate the problem in an inherently
stochastic framework involving imperfect information and bounded controls. Complying with the generalized separation
theorem, the control law factors in the posterior distribution of the state. We extend the well-known optimality of bang-bang evasion maneuvers in deterministic settings to the realm of realistic,
stochastic evasion scenarios. First, we prove that an optimal evasion strategy always exists, and that the set of optimal solutions includes at least one bang-bang policy, rendering the resulting optimal control problem finite-dimensional. Second, leveraging this structure, we propose the closed-loop terminal-set-based evasion (TSE) strategy, and demonstrate its effectiveness in simulation against a proportional
navigation pursuer. Monte Carlo simulations show that the TSE strategy outperforms traditional
stochastic evasion strategies based on random telegraph, Singer, and weaving models.
Read the full abstract ▾
Systems and Control
Robust Rule-Based Sizing and Control of Batteries for Peak Shaving Applications
Lorenzo Nespoli and Vasco Medici
As the cost of batteries lowers, sizing and control methods that are both fast and can achieve their promised performances when deployed are becoming more important. In this paper, we show how stochastically tuned rule based controllers (RBCs) can be effectively used to achieve both these goals, providing more realistic estimates in terms of achievable levelised cost of energy (LCOE), and better performances while in operation when compared to deterministic model predictive control (MPC). We test the proposed methodology on yearly profiles from real meters for peak shaving applications and provide strong evidence about these claims.
Read the full abstract ▾
Signal Processing
SIR Analysis for Affine Filter Bank Modulation
Henrique L. Senger, Gustavo P. Gonçalves, Bruno S. Chang et al. (+4 more)
The signal-to-interference ratio (SIR) of the Affine Filter Bank Modulation (AFBM) waveform is analyzed under minimum mean square error (MMSE) equalization in two domains; namely, the affine domain and the filtered time-domain (TD). Due to the incorporation of the discrete affine
Fourier transform (DAFT) and despreading/mapping, an interesting and counter-intuitive cancellation of the unwanted combination of the channel induced interference with the orthogonality approximation error is seen in the filtered TD, a process which does not occur in the affine domain. The direct impact on bit error rate (BER) provides a thorough validation of the proposed analysis and explains the substantial gains in performance of the filtered TD detection scheme as opposed to its affine domain equivalent
Read the full abstract ▾
All authors Henrique L. Senger, Gustavo P. Gonçalves, Bruno S. Chang, Hyeon Seok Rou, Kuranage Roche Rayan Ranasinghe, Giuseppe Thadeu Freitas de Abreu, Didier Le Ruyet
Image and Video Processing
Entropy Coding for Non-Rectangular Transform Blocks using Partitioned DCT Dictionaries for AV1
Priyanka Das, Tim Classen and Mathias Wien
Recent
video codecs such as VVC and AV1 apply a Non-rectangular (NR) partitioning to combine prediction signals using a smooth blending around the boundary, followed by a rectangular transform on the whole block. The NR signal transformation is not yet supported. A transformation technique that applies the same partitioning to the 2D Discrete Cosine Transform (DCT) bases and finds a sparse representation of the NR signal in such a dictionary showed promising gains in an experimental setup outside the reference software. This method uses the regular inverse transformation at the decoder to reconstruct a rectangular signal and discards the signal outside the region of interest. This design is appealing due to the minimal changes required at the decoder. However, current
entropy coding schemes are not well-suited for optimally encoding these coefficients because they are primarily designed for DCT coefficients. This work introduces an
entropy coding method that efficiently codes these transform coefficients by effectively modeling their properties. The design offers significant theoretical rate savings, estimated using conditional
entropy, particularly for scenarios that are more dissimilar to DCT in an experimental setup.
Read the full abstract ▾
Systems and Control · Cross-listed from math.OC
Closed Form HJB Solution for Continuous-Time Optimal Control of a Non-Linear Input-Affine System
Akash Vyas, Shreyas Kumar, Jayant Kumar Mohanta et al. (+1 more)
Designing optimal controllers for nonlinear dynamical systems often relies on reinforcement learning and adaptive dynamic programming (ADP) to approximate solutions of the Hamilton Jacobi Bellman (HJB) equation. However, these methods require iterative training and depend on an initially admissible policy. This work introduces a new analytical framework that yields closed-form solutions to the HJB equation for a class of continuous-time nonlinear input-affine systems with known dynamics. Unlike ADP-based approaches, it avoids iterative learning and numerical approximation. Lyapunov theory is used to prove the asymptotic stability of the resulting closed-loop system, and theoretical guarantees are provided. The method offers a closed-form control policy derived from the HJB framework, demonstrating improved computational efficiency and optimal performance on state-of-the-art optimal control problems in the literature.
Read the full abstract ▾
All authors Akash Vyas, Shreyas Kumar, Jayant Kumar Mohanta, Ravi Prakash
Systems and Control · Cross-listed from cs.LG
Predictive Safety Shield for Dyna-Q Reinforcement Learning
Jin Pin, Krasowski Hanna and Vanneaux Elena
Obtaining safety guarantees for reinforcement learning is a major challenge to achieve applicability for real-world tasks. Safety shields extend standard reinforcement learning and achieve hard safety guarantees. However, existing safety shields commonly use random sampling of safe actions or a fixed fallback controller, therefore disregarding future performance implications of different safe actions. In this work, we propose a predictive safety shield for model-based reinforcement learning
agents in discrete space. Our safety shield updates the Q-function locally based on safe predictions, which originate from a safe simulation of the environment model. This shielding approach improves performance while maintaining hard safety guarantees. Our experiments on gridworld environments demonstrate that even short prediction horizons can be sufficient to identify the optimal path. We observe that our approach is robust to distribution shifts, e.g., between simulation and reality, without requiring additional training.
Read the full abstract ▾
Signal Processing · Cross-listed from cs.AR
A 0.32 mm$^2$ 100 Mb/s 223 mW ASIC in 22FDX for Joint Jammer Mitigation, Channel Estimation, and SIMO Data Detection
Jonas Elmiger, Fabian Stuber, Oscar Castañeda et al. (+2 more)
We present the first single-input multiple-output (SIMO) receiver ASIC that jointly performs jammer mitigation, channel estimation, and data detection. The ASIC implements a recent
algorithm called siMultaneous mitigAtion, Estimation, and Detection (MAED). MAED mitigates smart jammers via spatial filtering using a nonlinear optimization problem that unifies jammer estimation and nulling, channel estimation, and data detection to achieve state-of-the-art error-rate performance under jamming. The design supports eight receive
antennas and enables mitigation of smart jammers as well as of barrage jammers. The ASIC is fabricated in 22 nm FD-SOI, has a core area of 0.32 mm$^2$, and achieves a throughput of 100 Mb/s at 223 mW, thus delivering 3$\times$ higher per-user throughput and 4.5$\times$ higher area efficiency than the state-of-the-art jammer-resilient detector.
Read the full abstract ▾
All authors Jonas Elmiger, Fabian Stuber, Oscar Castañeda, Gian Marti, Christoph Studer
Systems and Control
VibraWave: Sensing the Pulse of Polluted Waters
Sagnik Ghosh and Sandip Chakraborty
Conventional methods for water pollutant detection, such as chemical assays and optical spectroscopy, are often invasive, expensive, and unsuitable for real-time, portable monitoring. In this paper, we introduce VibraWave, a novel non-invasive sensing framework that combines mmWave radar with controlled acoustic excitation, tensor decomposition, and deep learning to detect and quantify a wide range of water pollutants. By capturing radar reflections as a three-dimensional tensor encoding phase dynamics, range bin power, and angle-of-arrival (AoA), we apply PARAFAC decomposition with non-negative constraints to extract compact, interpretable pollutant fingerprints. These are used to train a lightweight student neural network via knowledge
distillation, enabling joint classification and quantification of heavy metals (Cu, Fe, Mg), oil emulsions, and sediments. Extensive experiments show that VibraWave achieves high accuracy and low RMSE across pure, binary, and tertiary mixtures, while remaining robust and computationally efficient, making it well-suited for scalable, real-time water quality monitoring.
Read the full abstract ▾
Image and Video Processing
Semantic-Enhanced Feature Matching with Learnable Geometric Verification for Cross-Modal Neuron Registration
Wenwei Li, Lingyi Cai, Hui Gong et al. (+2 more)
Accurately registering in-vivo two-
photon and ex-vivo fluorescence micro-optical sectioning
tomography images of individual
neurons is critical for structure-function analysis in neuroscience. This task is profoundly challenging due to a significant cross-modality appearance gap, the scarcity of annotated data and severe tissue deformations. We propose a novel deep learning framework to address these issues. Our method introduces a semantic-enhanced hybrid feature descriptor, which fuses the
geometric precision of local features with the contextual robustness of a
vision foundation model DINOV3 to bridge the modality gap. To handle complex deformations, we replace traditional RANSAC with a learnable
Geometric Consistency Confidence Module, a novel classifier trained to identify and reject physically implausible correspondences. A data-efficient two-stage training strategy, involving pre-training on synthetically deformed data and fine-tuning on limited real data, overcomes the data scarcity problem. Our framework provides a robust and accurate solution for high-precision registration in challenging biomedical imaging scenarios, enabling large-scale correlative studies.
Read the full abstract ▾
All authors Wenwei Li, Lingyi Cai, Hui Gong, Qingming Luo, Anan Li
Signal Processing
Design Of A Communication System To Send Text Using Lora At 400 MHz
Fabrizio André Farfán Prado, William César Pérez Campos, Steisy Anahi Carreño Tacuri et al. (+2 more)
This work describes the design and implementation of a low-power wireless communication system for transmitting text using ESP32 modules and the LoRa DXLR01. The proposal arises as a solution to connectivity and energy-efficiency problems commonly found in rural areas and certain urban environments where Wi-Fi or mobile networks are unavailable or operate with limitations. To address this, LoRa technology known for its long-range capability and low power consumption is integrated with an ESP32 responsible for capturing, processing, and sending messages. The LoRa DXLR01 module, which operates in the 433 MHz band, is configured with parameters aimed at maximising both transmission range and efficient energy usage. Messages are sent using Chirp Spread Spectrum (CSS) modulation, improving signal penetration in obstructed areas and reducing the likelihood of errors. On the receiving end, the ESP32 interprets the data and displays it on an LCD screen. Additionally, the received information is sent to the ThingSpeak platform, allowing remote storage and visualisation without relying on conventional network infrastructure. Tests conducted in a controlled environment show an average latency of 3.2 seconds for text transmission. It was also verified that the system can be used in applications such as remote monitoring, infrastructure management, and access control.
Read the full abstract ▾
All authors Fabrizio André Farfán Prado, William César Pérez Campos, Steisy Anahi Carreño Tacuri, Favio David Cabrera Alva, Harold Jacobed Carhuas Lizarbe
Systems and Control
Multi-Hypotheses Navigation in Collaborative Localization subject to Cyber Attacks
Peter Iwer Hoedt Karstensen and Roberto Galeazzi
This paper addresses resilient collaborative localization in multi-
agent systems exposed to spoofed
radio frequency measurements. Each
agent maintains multiple hypotheses of its own state and exchanges selected information with neighbors using covariance intersection.
Geometric reductions based on distance tests and convex hull structure limit the number of hypotheses transmitted, controlling the spread of hypotheses through the network. The method enables
agents to separate spoofed and truthful measurements and to recover consistent estimates once the correct hypothesis is identified. Numerical results demonstrate the ability of the approach to contain the effect of
adversarial measurements, while also highlighting the impact of conservative fusion on detection speed. The framework provides a foundation for resilient multi-
agent navigation and can be extended with coordinated hypothesis selection across the network.
Read the full abstract ▾
Signal Processing
Group-wise Semantic Splitting Multiple Access for Multi-User Semantic Communication
Jungyeon Koh, Hyeonho Noh and Hyun Jong Yang
In this letter, we propose a group-wise semantic splitting multiple access framework for multi-user semantic communication in downlink scenarios. The framework begins by applying a balanced clustering mechanism that groups users based on the similarity of their semantic characteristics, enabling the extraction of group-level common features and user-specific private features. The base station then transmits the common features via multicast and the private features via unicast, effectively leveraging both shared and user-dependent semantic information. To further enhance semantic separability and reconstruction fidelity, we design a composite loss function that integrates a reconstruction loss with a repulsion loss, improving both the accuracy of semantic recovery and the distinctiveness of common
embeddings in the latent space. Simulation results demonstrate that the proposed method achieves up to 3.26x performance improvement over conventional schemes across various channel conditions, validating its robustness and semantic efficiency for next-generation wireless networks.
Read the full abstract ▾
Image and Video Processing
Knowledge Distillation for Continual Learning of Biomedical Neural Fields
Wouter Visser and Jelmer M. Wolterink
Neural fields are increasingly used as a light-weight, continuous, and differentiable signal representation in (bio)medical imaging. However, unlike discrete signal representations such as voxel grids, neural fields cannot be easily extended. As neural fields are, in essence, neural networks, prior signals represented in a neural field will degrade when the model is presented with new data due to catastrophic forgetting. This work examines the extent to which different neural field approaches suffer from catastrophic forgetting and proposes a strategy to mitigate this issue. We consider the scenario in which data becomes available incrementally, with only the most recent data available for neural field fitting. In a series of experiments on cardiac cine MRI data, we demonstrate how knowledge
distillation mitigates catastrophic forgetting when the spatiotemporal domain is enlarged or the dimensionality of the represented signal is increased. We find that the amount of catastrophic forgetting depends, to a large extent, on the neural fields model used, and that
distillation could enable continual learning in neural fields.
Read the full abstract ▾
Systems and Control
Decentralized Shepherding of Non-Cohesive Swarms Through Cluttered Environments via Deep Reinforcement Learning
Cristiana Punzo, Italo Napolitano, Cinzia Tomaselli et al. (+1 more)
This paper investigates decentralized shepherding in cluttered environments, where a limited number of herders must guide a larger group of non-cohesive, diffusive targets toward a goal region in the presence of static obstacles. A hierarchical control architecture is proposed, integrating a high-level target assignment rule, where each herder is paired with a selected target, with a learning-based low-level driving module that enables effective steering of the assigned target. The low-level policy is trained in a one-herder-one-target scenario with a rectangular obstacle using Proximal Policy Optimization and then directly extended to multi-
agent settings with multiple obstacles without requiring retraining. Numerical simulations demonstrate smooth, collision-free trajectories and consistent convergence to the goal region, highlighting the potential of reinforcement learning for scalable, model-free shepherding in complex environments.
Read the full abstract ▾
All authors Cristiana Punzo, Italo Napolitano, Cinzia Tomaselli, Mario di Bernardo