Papers

Event cameras offer distinctive advantages, including microsecond-level latency and high dynamic range, rendering them promising for challenging perception tasks. Inspired by biological vision, they output asynchronous and sparse event streams rather than dense image frames, creating a fundamental mismatch with mainstream neural networks. This survey reviews recent advances in event camera representation learning from the perspective of converting raw event streams into learnable representations. We first organize existing methods according to whether they rely on a single principal event representation or jointly exploit multiple complementary representations. Single-representation methods are further categorized into dense-based representations, which regularize events into structured grid-like forms, and sparse-based representations, which preserve event-native discrete spatio-temporal structures. Multi-representation methods are organized into dense-dense and dense-sparse hybrid formulations that exploit complementary representation properties. This representation-centric taxonomy clarifies how different paradigms balance structural regularity, temporal fidelity, sparsity preservation, and architectural compatibility. For each paradigm, we examine the underlying design choices, modeling principles, and task-level implications. We further summarize standard benchmarks and evaluation settings across representative high-level perception and low-level vision tasks. Finally, we discuss open problems and outline future directions from fixed representation design toward adaptive representation optimization, improved fidelity-efficiency trade-offs, and more scalable event-based perception systems.

This work presents a method for constructing nonlinear reduced-order models from input-output time-domain data. The proposed approach, termed Mixed Interpolatory Inference with Variable Projection (MIIvp), exploits the fact that the considered class of nonlinear state-space models is linear in the output equation parameters. By applying the Variable Projection (VarPro) algorithm, the optimization is restricted to the state equation parameters alone, while the output equation parameters are recovered via linear least squares. As a consequence, the output dimension does not enter the nonlinear optimization parameter vector, making the method well suited for systems with very high-dimensional outputs, a setting where many other approaches become computationally prohibitive. Under mild assumptions, it is shown that MIIvp can recover the true model parameters up to similarity. The method is first validated on a synthetic bilinear system, where it achieves machine-precision accuracy and recovers the true eigenvalues. MIIvp is then compared with existing methods on two experimental benchmarks from the nonlinear system identification literature. These numerical experiments showcase both the validity and the limitations of the proposed approach. Finally, directions for improvements and future work are outlined.

Covert communications aim to hide the existence of wireless transmissions from unauthorized adversaries. However, conventional designs based on blind interference or passive uncertainty can be ineffective in dynamic propagation environments. This article investigates sensing-empowered covert communications, where adversary and environmental information are used to guide transmission and jamming control. We show how sensing changes covert system design from passive concealment to state-aware decision-making, while also introducing new challenges related to exposure and resource consumption. We further discuss several intelligent sensing paradigms that extract task-relevant information with limited active probing. A case study in low-altitude wireless networks illustrates that sensing-assisted beamforming can improve spatial resource utilization and the reliability of covert data delivery in time-varying channels. Finally, several open issues are discussed to support more adaptive covert wireless systems.

Stochastic programming has become a popular tool for supporting decision-making under uncertainty in the long-term planning of energy systems. Existing scenario reduction methods, however, are naive about the long-term temporal nature of scenarios, which limits their efficiency in reducing model size. In this paper, we overcome this inefficiency by proposing a novel time-variant scenario reduction framework that explicitly allows for varying scenario aggregations over time. As a result, scenario probabilities become time-variant, enabling not only the accurate capture of scenario realizations but also their probabilities at the time steps that drive investment decisions. This substantially increases flexibility compared to traditional time-invariant methods, which we demonstrate on a two-stage stochastic generation expansion planning problem with uncertain renewable power production.

Common synaptic input is considered a fundamental principle of motor neuron control and represents the dominant component of the neural drive transmitted from the motor neurons to muscle. Recent advances in High-Density surface Electromyography (HDsEMG) and motor unit (MU) decomposition algorithms have enabled the concurrent identification of increasingly large populations of MUs and substantially expanded the possibility of estimating common synaptic input from MU spike trains, making this approach widely used to investigate the neural control of movement in humans. However, multiple analytical approaches are currently available, each relying on different physiological assumptions, mathematical formulations, and parameter choices. The lack of practical guidelines and open-source implementations has also limited the accessibility and reproducibility of these analyses. In this tutorial, we provide a practical, physiologically grounded guide to estimating common synaptic input from populations of MU spike trains using openhdemg, an open-source Python framework. We organize the available methods into three complementary categories: time-domain approaches applied to smoothed discharge rates, frequency-domain approaches based on coherence between cumulative spike trains, and a network-information approach based on nonlinear pairwise dependencies and graph theory. For each method, we describe its physiological interpretation, step-by-step estimation, and systematically examine how key parameter choices influence the resulting estimates, providing practical recommendations for their selection. Finally, we present a complete workflow from HDsEMG decomposition and MU cleaning to common synaptic input estimation, demonstrating that decomposition quality directly affects these estimates.

Continuous Variational Autoencoders (VAEs) serve as the fundamental continuous tokenizer for modern neural audio generation systems, enabling high-fidelity reconstruction while providing a compact, smooth latent space for downstream generative priors. However, continuous VAEs face a fundamental conflict among compression rate, reconstruction fidelity, and latent space topology, which we formalize as the Rate-Distortion-Regularity Trilemma. This trilemma stems from a topological mismatch: the isotropic Gaussian prior in standard VAEs imposes a flat latent geometry that fails to accommodate audio's hierarchical nature, where low-frequency components are structured and compressible while high-frequency components are stochastic and incompressible, leading to disordered information packing in which crucial semantic features are interleaved with high-entropy noise. To address this challenge, we propose Structured Topology-Aware Regularization (STAR), a general training strategy that reshapes latent space geometry by imposing a growth-based constraint field, routing structural and textural information into channel subspaces with matching capacities. STAR is applicable to any VAE architecture and effectively resolves the trilemma, as demonstrated in CNN-based VAEs. We further present STAR-VAE, which combines STAR with a hybrid CNN-Mamba architecture for local feature extraction and linear-complexity global context modeling, and STAR-Gen, an LLM-based Flow Matching framework that leverages STAR-VAE's structured latent space for high-fidelity generation without vector quantization artifacts. Experiments across diverse audio domains show that STAR-VAE achieves state-of-the-art reconstruction fidelity and enhanced semantic information preservation, while the structured latent space improves both traditional diffusion models and STAR-Gen for text-to-audio generation.

Hallucinations of ASR models - fluent transcriptions with no basis in audio - degrade system performance and pose risks in downstream applications. Robust detection of such errors remains a challenge. This paper studies Whisper large v3 hallucination detection on real-speech human-annotated data across three paradigms: text-based, LLM-based, and internal decoder state probing. Text classifiers utilizing metrics for text evaluation achieve high recall but degrade without reference transcripts. LLM-based detection improves precision with domain-specific prompt conditioning, yet remains less competitive than the lightweight text-based methods. Probing Whisper's decoder representations, without a ground-truth reference, yields the strongest performance, revealing that hallucination traits are encoded across intermediate decoding layers. A late-fusion meta-classifier combining text and internal-state outputs achieves the best overall detection performance.

Speech Language Models achieve reasoning capabilities, but are often hindered by massive parameter counts and a tendency to prioritize linguistic priors over acoustic features. While contrastive decoding enhances grounding by contrasting audio-aware and text-only logits, it increases inference latency. We propose Contrastive Audio-Aware Distillation (CAAD), a framework that internalizes the teacher's contrastive reasoning into the student model's weights. To overcome the high computational training overhead in the dual-path token-by-token contrastive distillation process, we introduce a synchronized teacher-forcing strategy. Anchored by unified Pseudo-Ground Truths, this mechanism enables simultaneous full-sequence generation of the teacher's contrastive distributions, allowing student to distill the audio-aware signal efficiently. Overall, CAAD yields a ~8% relative gain over standard knowledge distillation on Dynamic-SUPERB and successfully reduces linguistic bias in MCR-BENCH.

End-to-end Automatic Speech Recognition (ASR) systems hallucinate on natural speech, yet existing mitigation methods are typically evaluated on non-speech or artificially corrupted audio. We introduce HALAS, the first human-annotated dataset of naturally occurring hallucinations from seven state-of-the-art ASR models on real unprocessed earnings call recordings. HALAS provides span-level labels, enabling analysis of hallucination patterns and their severity. Our analysis reveals strong cross-model vocabulary overlap and confirms that hallucinations also occur for almost correctly transcribed speech (characterized by a low Word Error Rate). The proposed benchmark with HALAS shows that the character and semantic-level metrics used as a proxy for hallucination detection reach 81% ROC-AUC, while state-of-the-art detection methods achieve an F1 score of only 53.1%. As such, HALAS establishes the first rigorous non-artificial benchmark for the detection and mitigation of ASR hallucinations.

This paper presents a data-driven framework for decentralized Nash equilibrium (NE) seeking in multi-agent systems with unknown linear dynamics subject to exogenous disturbances, operating under partial-decision information (where agents lack direct access to the decisions of all others) and equality constraints. The proposed framework integrates an NE model, a distributed communication protocol, an internal model for disturbance rejection, and a data-driven stabilization strategy. By reformulating the problem as a cooperative output regulation problem, we synthesize controllers directly from noisy input-state data via semi-definite programs (SDPs), providing formal guarantees for closed-loop stability and asymptotic convergence to the NE. The approach is further extended to a class of nonlinear systems with constant disturbances by leveraging integral control and describing nonlinearities via quadratic constraints. Numerical simulations involving unmanned aerial vehicle networks and a rotary-wing aerial vehicle formation validate the efficacy and robustness of the proposed method.

As Urban Air Mobility (UAM) scales toward high-density operations, generating collision-free trajectories within complex 3D cityscapes is a critical safety requirement. This paper proposes a scalable Sequential Quadratic Programming (SQP) framework that integrates geometric environmental constraints, operational limits, and vehicle dynamics within a single online trajectory optimization process. Rather than precomputing obstacle-free corridors ahead of time, our method encodes obstacle avoidance as live separating-hyperplane constraints regenerated at every solver iteration, so that dense urban geometry and full-DOF vehicle dynamics are resolved jointly and online as the reference and environment evolve. A variable-scale quadtree decomposition keeps computation bounded, enabling the framework to scale to city-wide environments while preserving real-time performance for high-speed flight. We validate the framework against conventional SQP, Iterative Linear Quadratic Regulator, and Differential Dynamic Programming across flights in five real-world urban centers, attaining 100% success and clearance rates on CPU-only hardware.

This technical report presents submission systems for Task 7(domain-incremental audio classification) of the DCASE 2026 Challenge. The main obstacle is that, the system is unable to access to past or future domain's data at once. We approached domain-incremental learning (DIL) as a frozen-feature replay problem. At each incremental stage, one or two compact experts are trained and then kept fixed; at the final stage, the penultimate features from all frozen experts are concatenated and used to train a lightweight per-class prototype classifier solely on cached features. This design prevents catastrophic forgetting by preserving each expert models at inference. To retain earlier-domain knowledge without storing raw audio, some experts were trained with DeepInversion-based generative replay. A cross-stage regression imputer was trained to fill the expert feature slots that did not yet exist at an ealier stage. We submit four fully DIL-compliant systems: three systems based on diverse frozen five-expert backbones and their cross-stack ensemble achieving 78.15% micro / 77.03% macro on the development set, outperforming every individual backbone on both evaluations.

Multi-zone HVAC control is a spatial decision problem in which indoor thermal evolution and control decisions depend not only on outdoor conditions and internal heat gains but also on zone layout, physical adjacency, and delayed thermal interactions across the building. Recent LLM-based HVAC controllers have shown that prompt-based control is feasible. However, these methods typically rely on task descriptions, observation values, short textual feedback, or unstructured retrieval, which limits their ability to reason about zone coupling, thermal response, and building dynamics. This paper presents a thermodynamics-aware LLM control framework for a five-zone EnergyPlus building simulation. The controller is grounded in a physics-informed spatial knowledge graph derived from Brick-style building semantics and linked with recent interaction history. At each control step, the model receives the current building state, graph-structured spatial context, and recent environment-controller history, enabling it to make decisions that reflect both building structure and short-term thermal evolution. We evaluate the framework against standard control baselines and several LLM-based alternatives. Results show that the proposed approach achieves the best overall energy-comfort trade-off and the lowest PMV violation while maintaining energy-efficient operation.

Explaining and understanding the decision-making process of artificial intelligence (AI) systems, particularly those implemented by neural networks, falls within the field of explainable AI (XAI). Analogous to the human attention mechanism, neural networks are assumed to possess their own attention mechanisms that selectively process information during decision-making. This work proposes to study one XAI topic: analysing and visualising the attention mechanisms of neural networks. Our experiments are performed on speaker recognition neural networks that are trained to identify speaker identity from a given utterance. Previous studies have widely used class activation map (CAM)-based methods to analyse and visualise the attention mechanisms of neural networks. Each of these methods produces an attention map for each network input, highlighting which input regions are selectively processed when the speaker recognition network makes decisions. However, the evaluation of attention maps produced by these methods remains largely underexplored. This work systematically reviews an existing attention map evaluation algorithm, establishing key concepts and identifying its shortcomings. On the basis of this existing evaluation algorithm, a new version is then proposed to address the identified shortcomings, called the Modified Randomised Input Sampling for Explanation - Evaluation algorithm (Modified RISE-eval). Using Modified RISE-eval, we evaluate the attention maps produced by two representative CAM-based methods, GradCAM and LayerCAM, applied to a certain speaker recognition network. The evaluation results demonstrate that GradCAM and LayerCAM each exhibit distinct advantages when applied under different experimental conditions in the speaker recognition task.

Low Earth Orbit (LEO) satellite networks face critical challenges in radio resource allocation due to dynamic traffic demands and stringent interference constraints. Beam-hopping (BH) technology offers a promising solution by enabling dynamic beam resource allocation across spatial and temporal domains. In this paper, we propose a Tabu Search-based spatio-temporal BH resource allocation strategy for LEO satellite communication systems. Specifically, the BH scheduling problem is formulated to maximize user demand satisfaction under interference constraints. To solve this problem efficiently, the proposed Tabu Search framework integrates adaptive tabu tenure control, greedy-based initialization with interference-aware beam selection, and Simulated Annealing acceptance criteria. Extensive simulation results demonstrate that the proposed method consistently improves system throughput by 17.2% and user satisfaction by 11.7% compared with greedy-based BH strategies. These results indicate that the proposed approach provides a scalable and robust solution for dynamic resource allocation in interference-limited LEO satellite networks.

Fixed-Position Antennas (FPAs) are constrained by static physical topologies and struggle to adapt to rapidly varying wireless environments. By dynamically reconfiguring the antenna positions, Fluid Antenna Systems (FASs) introduce additional spatial Degrees of Freedom (DoF) for wireless optimization. This paper investigates the joint optimization of Fluid Antenna System (FAS) topology reconfiguration and active beamforming for mobile Integrated Sensing and Communication (ISAC) systems. To enable real-time decision making, an end-to-end optimization framework based on the Soft Actor-Critic (SAC) algorithm is proposed. Simulation results show that the proposed scheme achieves an online inference latency of only 4 ms. Compared to the widely used alternating optimization, it improves communication performance by 42%. Moreover, it achieves performance comparable to the SCA-SDR benchmark while requiring 57% fewer antennas, demonstrating superior hardware efficiency.

The clinical diagnosis of skin diseases is susceptible to interference from inter-class similarity of skin lesions, and over-reliance on clinicians'experience easily leads to subjective bias. Although existing deep learning aided diagnosis methods achieve competitive accuracy, they suffer from the black-box opacity of Vision Transformer (ViT) and poor adaptability to medical few-shot scenarios. Moreover, mainstream explainable algorithms generally face the bottleneck of significant accuracy degradation when improving interpretability. This paper proposes an interpretable ViT (IViT) constrained by Quadratic Programming (QP). The introduced pre-trained transfer learning adapts to few-shot feature extraction. A discrete QP feature selection framework is constructed to screen generic and discriminative features consistent with clinical diagnostic logic. A multi-objective loss function is designed to reduce feature redundancy and optimize activation distribution while preserving classification performance. Experimental results on six standard skin disease datasets show that IViT achieves an accuracy of 93.80%, only 0.21% lower than the baseline, with feature redundancy reduced by 29.5%. Its core activation regions are consistent with clinically concerned lesion areas. The proposed model balances accuracy and interpretability, providing a reliable solution for the clinical deployment of few-shot intelligent skin disease diagnosis.

LiDAR (Light Detection and Ranging) sensors capture the surrounding environment as dense 3D point clouds by measuring the time-of-flight of emitted laser pulses, making them foundational across autonomous vehicles, robotics, and large-scale mapping. PCAP (Packet Capture) files from these sensors are the starting point of most 3D perception pipelines, yet internal packet structures, UDP (User Datagram Protocol) port conventions and encoding schemes differ enough across manufacturers that no single tool reads them all. Ouster, Velodyne, Hesai, and Livox each require their own SDK (Software Development Kit), their own environment setup, and their own conversion workflow. Supporting all four means maintaining four disconnected pipelines with no shared infrastructure. The pipeline described here takes a raw PCAP as input and handles vendor identification automatically, scoring six independent file characteristics through a weighted multi-signal approach to determine the source sensor. C++ SDKs handle Ouster and Velodyne, while Hesai and Livox rely on Python-based dpkt parsing where no open source SDK exists. From there, a single command writes output to any of five industry-standard formats. We tested on real outdoor captures. Ouster peaks at 2.08M points per second, Velodyne at 1.47M, both running through native C++ packet decoding. Hesai and Livox land at 110K and 150K respectively, where Python-layer parsing introduces overhead that compounds under sustained load. The 8-10x gap held consistently across runs. Tested on a consumer-grade i3 with 8GB RAM, no vendor configuration required

Spoken Language Understanding (SLU) is moving from task-specific pipelines toward large audio language models (LALMs) that generate natural-language responses. However, existing speech benchmarks mainly focus on single-speaker settings or isolated subtasks, leaving speaker-centric understanding in realistic multi-speaker conversations insufficiently evaluated. We introduce MSU-Bench, a diagnostic benchmark for multi-speaker conversational understanding, covering 16 speaker-centric tasks and 2,300 QA instances in a two-tier framework from speaker grounding to dialogue reasoning. We build a Gemini-assisted annotation and QA generation pipeline with human-in-the-loop verification, achieving high QA validity and strong agreement between human answers and verified labels. We further analyze speaker-referencing schemes and diagnostic error types to reveal bottlenecks in speaker grounding and reasoning. Experiments reveal clear gaps across model families, with closed-source systems leading overall but all models still facing challenges in complex speaker grounding and multi-speaker reasoning. The benchmark annotations, metadata, and evaluation scripts will be available at the GitHub repository: https://github.com/ASLP-lab/MSU-Bench.

Building on the collective advancements in the literature \cite{trinh1, trinh2, trinhnn26, trinhnam26, trinhnam1}, this paper proposes the design of delayed functional observers to asymptotically estimate a generalized delayed control law under significant input and output delays. This framework enables designers to extend the allowable bounds for input delays while ensuring that the observer-based control scheme stabilizes the system despite simultaneous mismatched input and output time-delays.

We present HERCULES, an open-source simulator and data-collection pipeline for heterogeneous multi-robot autonomy. Built upon the Unreal Engine 5 (UE5)-based simulators AirSim and Cosys-AirSim, HERCULES resolves key architectural limitations of prior frameworks to enable concurrent unmanned aerial and ground vehicle (UAV-UGV) operation in large-scale, photorealistic, dynamic environments. It introduces a new waypoint-tracking UGV controller that mirrors existing UAV control interfaces, and provides a shared navigation stack for mapping, traversability analysis, planning, and control across heterogeneous platforms. Expanding inherited sensor suites, it adds physics-based long-wave infrared (LWIR) cameras and configurable night-vision modes for degraded visual environments. HERCULES provides lightweight APIs, ROS 2 wrappers, and rigorous time synchronization across sensors and platforms, and brings state-of-the-art game-engine capabilities into robotics simulation, integrating intelligent agents such as pedestrians, traffic, and wildlife with high-fidelity dynamic phenomena, including fire, flooding, and crop disease spread. HERCULES runs in two modes: passively, replaying offline-designed trajectories to generate reproducible multi-modal datasets, and actively, running an online planner in closed loop from live observations. Our experiments in heterogeneous multi-robot SLAM, collaborative perception, and exploration, using both HERCULES-generated data and active closed-loop execution, demonstrate its utility for advancing heterogeneous multi-robot autonomy. We publicly release our source code, experiment code, documentation, and datasets, including a heterogeneous multi-robot SLAM benchmark collected with two UAVs and two UGVs across kilometer-scale desert, forest, and city environments, at https://lunarlab-gatech.github.io/HERCULES-website.

This paper introduces a low-complexity technique named quasi-direct geolocation (QDG) to perform passive radio-frequency (RF) geolocation of emitters directly in the position domain, akin to direct geolocation (DG) and direct position determination (DPD). The proposed technique drastically reduces the complexity of DG/DPD and is experimentally demonstrated in geolocating a terrestrial jammer at Jammertest 2025 from a repurposed satellite in low Earth orbit (LEO): OPS-SAT PRETTY. The goal of QDG is to enable satellites to contribute to a multi-constellation system for RF interference (RFI) monitoring as opportunistic spectrum sensors in global navigation satellite systems (GNSS) bands, even if these are constrained by low size, weight, and power (SWaP). They can serve as data collectors and/or edge computers. In the former case, QDG can be used to compress large volumes of I/Q samples into minimal signal information, which can be relayed to ground for post-processing via low-capacity downlinks. In the latter case, QDG can be used to compute the geolocation of RFI sources in orbit on low-power on-board computers (OBC). The drawback of these capabilities is lower sensitivity and accuracy than DG/DPD plus limitations on the types of signal sources that can be geolocated, which, nonetheless, include the most common GNSS jammers.

Wave energy converters (WECs) require system-level techno-economic analysis to balance power production, cost, and survivability. Existing simulation tools are either too computationally costly for large-scale optimization or too narrow in disciplinary scope to support integrated design studies. This work presents MDOcean, a novel open-source WEC simulation framework for rapid early-stage design exploration, parametric analysis, and multidisciplinary optimization. MDOcean integrates hydrodynamics, dynamics, structures, and economics in a computationally efficient architecture based on analytical and semi-analytical methods that substantially reduce runtime while maintaining near-numerical accuracy. The framework includes an eigenfunction-based linear hydrodynamic solver, a quasi-linearized frequency-domain dynamics engine capable of modeling drag and saturation nonlinearities, a structural sizing module incorporating realistic yield, ultimate, buckling, storm, and fatigue design criteria, and a simple cost model for techno-economic assessment. Particular emphasis is placed on the linearized pseudo-spectral optimal control formulation, which extends frequency-domain constraint-handling approaches with a unified describing-function and analytical quadratically-constrained quadratic program framework. This formulation efficiently treats nonlinearities and constraints while preserving compatibility with optimization and frequency-domain analysis techniques. Validation and benchmarking demonstrate that MDOcean's 151 ms runtime is orders of magnitude faster than leading WEC simulation tools while maintaining agreement with higher-fidelity baselines to within a few percent in most cases. The framework also provides insight into limiting behaviors, scaling laws, subsystem interactions, and key tradeoffs governing WEC design and techno-economic performance.

Growing reports of global navigation satellite system (GNSS) radio-frequency interference (RFI) highlight the need for scalable wide-area sensing for situational awareness. Crowdsourced Automatic Dependent Surveillance-Broadcast (ADS-B) receiver networks form a large-scale opportunistic sensor network for GNSS RFI monitoring, but ADS-B quality indicators may remain high during abnormal reported-position behavior, and heterogeneous receiver timestamping can produce apparent speed spikes. This letter proposes a three-stage framework that screens position-jump candidates, verifies local track consistency to suppress timing artifacts, and groups confirmed anomalies into traffic-adaptive multi-aircraft events. Using 605 million 1090-MHz ADS-B reports over Northeast Asia from December 2025 to February 2026, the framework identified 166 event clusters within the validity window of Notice to Airmen (NOTAM) RKRR Z1401/25 and none in the pre-NOTAM period. More than 99% of confirmed anomalies remained in high quality-indicator regimes, suggesting that track-consistency verification provides a complementary sensing criterion for GNSS RFI monitoring.

Jamming and spoofing pose significant threats to wireless and satellite navigation by disrupting radio-frequency (RF) signals and compromising availability and integrity. Robust RF interference direction finding through angle-of-arrival (AoA) estimation is therefore essential for detecting and localizing anomalous signals. Although data-driven methods perform well under line-of-sight (LoS) conditions, their performance degrades in practical environments due to non-line-of-sight (NLoS) multipath propagation. In this work, we propose a hybrid learning framework that incorporates physics-informed constraints into deep neural networks to improve the robustness of AoA estimation. A neural network is trained to estimate the azimuth and elevation of incoming signals received by a four-element antenna array, while a physics-informed loss enforces consistency between the predicted angles and inter-antenna phase differences under a plane-wave model. We further introduce a latent-space classifier to distinguish LoS from NLoS samples. Since inter-antenna phase differences under LoS propagation exhibit domain-invariant structure across environments, the physics-based loss is applied only to LoS samples, promoting physically consistent and domain-invariant representations without over-constraining the model in NLoS scenarios. In addition, domain-incremental learning (DIL) across NLoS environments with varying scatterer distributions improves cross-domain generalization. Evaluations on real-world datasets show that the proposed method reduces AoA estimation error by up to 6{\deg} in low-exemplar settings compared with DIL baselines.

Urban deceleration is one of the most empirically studied yet least taxonomically organized behaviors in car-following research. Recent perception-equipped autonomous-vehicle datasets enable trajectory-anchored mode discovery. We extract 1,219 sustained deceleration events from 234 urban driving logs of the Argoverse 2 Sensor dataset, encode each event in a 19-dimensional kinematic feature vector, discover behavioral modes via K-means clustering with bootstrap stability analysis, and quantify modulation by eleven scene-context variables. A HistGradientBoosting classifier predicts mode membership from the first 1.0 s of each event. Four stable modes emerge with a bootstrap Adjusted Rand Index of 0.897 across 50 resamples: anticipatory soft (62.8%), reactive closing (30.6%), brake-like jerk (4.8%), and an outlier category (1.8%). Only pair age shows a medium effect (epsilon^2 = 0.085); scene geometry and vulnerable-road-user proximity show negligible effects. The early-event classifier achieves macro-F1 = 0.758 at 1.0 s, with scene context contributing +0.059 F1 over kinematics alone. Modes are regime-invariant in medium-speed driving (ARI = 0.817) but regime-dependent at low speed (ARI = 0.166). A small set of stable kinematic modes structures urban deceleration; early-window jerk dominates predictive signal; and pair age is the primary contextual modulator.

Takens' theorem is the gold standard for attractor reconstruction from time series, but it guarantees only topological equivalence and does not preserve metric or group properties such as symmetries. We show that switching from delay-coordinate space to jet space (signal and its derivatives) allows one to exactly preserve the symmetry group of the original system. This statement is rigorously justified by a theorem on the isomorphism of Lie algebras under jet prolongation. Numerical experiments on the Lorenz and Rössler systems confirm that jet-space reconstruction preserves geometry and symmetries, whereas Takens embedding distorts them. As quantitative metrics we use a variational elastic energy functional and the correlation dimension. It is shown that jet-space reconstruction not only outperforms Takens embedding but in some cases yields more accurate estimates of invariants than projections of the original system. The proposed approach provides a coordinate-invariant criterion for the classification of strange attractors and can serve as a basis for detecting hidden attractors.

We present a geometric coordinate-free solution to the isotropic Schrödinger bridge problem (SBP) for the kinematic equation on the Lie group $\mathsf{SO}(2)$. We consider the angular velocity of the system as the control input and assume that the given initial and terminal state probability density functions defined on $\mathsf{SO}(2)$ in our SBP are continuous and strictly positive. We solve the SBP by proving the existence and uniqueness of a solution to the so-called Schrödinger system of equations on $\mathsf{SO}(2)$, by showing that a fixed-point recursion is contractive in a complete metric space with respect to the Hilbert's projective metric. The geometric controller thus designed only uses the intrinsic geometric structure of $\mathsf{SO}(2)$ and does not embed it in the Euclidean plane to achieve the optimal density control. The numerical simulation verifies the validity of the theoretical construction of the Schrödinger bridge. The code and animations are publicly available at \texttt{\href{https://gitlab.com/a5akhtar/sbp}{https://gitlab.com/a5akhtar/sbp}}.

Forecasting chaotic dynamical systems such as the Lorenz attractor is notoriously difficult: small numerical errors are amplified exponentially over long autoregressive rollouts. We study seven recurrent and convolutional architectures for the AI-DEEDS 2026 Chaotic Systems Challenge: a vanilla LSTM, an LSTM with additive attention, a Bidirectional LSTM (BiLSTM), a BiLSTM trained with the Huber loss, a Temporal Convolutional Network (TCN), a CNN front-end followed by an LSTM, and a CNN front-end followed by a BiLSTM. All models share the same pre-processing, sequence length, and rollout procedure, isolating the contribution of each design choice. The challenge scores predictions on a 0-100 scale where higher is better. We obtain leaderboard scores between 45.72 and 58.81, with the BiLSTM trained with Huber loss being the strongest configuration. Two findings stand out: (i) adding additive attention to the unidirectional baseline degraded performance by over ten points, and (ii) prepending a CNN front-end to either an LSTM or a BiLSTM did not help and slightly hurt the score. Per-pair RMSE measurements confirm that the BiLSTM family generalizes better in the harder pairs (6-7), while the LSTM + Attention model collapses there (RMSE up to 8.94 on pair 6). We discuss why bidirectional context and a robust loss help in chaotic regimes while attention and CNN front-ends fail in this setting.

Foundation models provide highly descriptive representations for medical images, yet their reliability degrades under distribution shifts arising from changes in patients, devices, or acquisition conditions. Reliable out-of-distribution (OOD) detection is therefore essential for safe deployment. Recent post-hoc detectors efficiently exploit frozen embeddings (e.g., kNN), whereas reconstruction-based OOD detection in latent feature space has seen limited adoption due to inconsistent performance. In this work, we show that the limitation of reconstruction-based methods in latent space does not stem from poor reconstruction quality, but from how reconstruction errors are scored. Standard L2 residual norms collapse the anisotropic residual structure, thereby suppressing informative deviations. To address this limitation, we introduce MaRS (Mahalanobis Residual Scoring), a label-free OOD detector that learns an in-distribution manifold using a lightweight autoencoder and measures deviation via a Mahalanobis distance on reconstruction residuals, yielding variance-aware OOD scores. Across three imaging modalities, multiple types of distribution shift, and different model families and scales, MaRS outperforms established confidence-, distance-, and reconstruction-based baselines, while remaining fully post-hoc and lightweight. The code is available at https://github.com/francescodisalvo05/mars.