The increasing proliferation of inverter-based resources (IBRs) in distribution networks is presenting a major challenge for phasor-based overcurrent protection. This challenge stems from IBRs' lack of short-circuit current sourcing capacity. As a result, traditional overcurrent protection functions (e.g., ANSI 51) are inadequate in such scenarios, and warrant alternative approaches. Time-domain protection, for example, shows promise in overcoming this challenge. In this paper we propose a pre-fault voltage discrimination (PVD) strategy whose role is to detect faults and discriminate normal switching and transformer inrush disturbances from actual faults. The use of PVD allows for the design of a simple, yet effective fault detection algorithm by using time-domain protection principles for distribution networks containing IBRs. The introduction of PVD provides for faster fault detection without reducing security and dependability. Offline simulation experiments and controller hardware-in-the-loop real-time simulation validate the effectiveness of the proposed algorithm against various fault and normal switching events.

Background and Objective: Continuous wearable electrocardiogram (ECG) monitoring is increasingly used for ambulatory arrhythmia surveillance, yet forecasting impending atrial fibrillation (AF) is challenged by inter-patient ECG variability. This study investigated whether personalizing a global model via fine-tuning on an individual's ECG signals improves short-term forecasting of impending AF. Methods: A global model trained on the ICENTIA11K dataset was compared against personalized models fine-tuned across three cohorts: ICENTIA11K, IRIDIA-AF, and MobiCARE. Following preprocessing, models processed 60-second ECG segments for a five-minute forecast horizon. We evaluated the impact of adaptation data volume and analyzed ECG features, such as heart rate and RMSSD. Results: Personalized models significantly outperformed the global model, achieving AUROCs of 0.711 vs. 0.614 in ICENTIA11K and 0.686 vs. 0.585 in MobiCARE. Personalization benefits increased with the amount of patient-specific fine-tuning data. While the global model's accuracy rose as AF onset approached, personalized models in the two external cohorts exhibited distinct temporal dynamics, which may indicate the capture of patient-specific characteristics less dependent on proximity to the AF event. Pre-AF episodes showed elevated heart rates and RMSSD. Feature attributions highlighted clinically relevant precursors, including frequent premature atrial complexes (PACs) and short supraventricular tachycardias (SVTs). Conclusions: Adapting deep learning models with patient-specific wearable ECG data significantly enhances short-term forecasting of impending AF. This personalized framework supports timely preventive interventions and improved AF management in ambulatory monitoring environments.

The Information Bottleneck (IB) is a well established framework that looks for a latent compact representation of a data source, by trading rate and data-size representation, for information accuracy with respect to another target data. The Gaussian IB (GIB) is its simple closed form solution, when the target is jointly Gaussian with the source. Actually, in many practical problems the latent representation has to be stored or represented by a finite number of bits, while the optimal (G)IB solution has not. First, this manuscript theoretically analyzes the effect of scalar and vector quantization of the GIB latent representation, and its impact on the (dis)informativeness with respect to the target data. Then, task-oriented quantization designs are proposed by (jointly) reformulating the GIB optimization problem under a finite-rate constraint on the latent representation. Simulation results on MMSE regression problems confirm the effectiveness of the proposed quantization designs, which show significant gains with respect to more heuristic, or separate, quantization designs of the standard GIB latent representation. Finally, the paper extends the task-oriented philosophy to non-Gaussian settings, by properly modifying the cost function used in variational auto-encoders (VAEs) of IB-inspired vector quantizers.

Recent research has explored integrating Large Language Models (LLMs) with speech encoders to create speech-augmented LLMs capable of contextualized speech recognition. The main challenge lies in aligning the semantic embeddings of LLMs with the acoustic representations of speech encoders. We propose a novel approach that teaches the LLM to first predict phonemes from the speech features before generating the final transcript. By integrating a phoneme prediction step directly into the LLM, the model develops a fine-grained knowledge of pronunciation, reducing acoustic confusion and improving transcription accuracy and explainability. Our method is cheap and simple, as phoneme targets can be automatically derived from existing transcripts. Through comprehensive experiments, we show that intermediate phoneme prediction can improve speech recognition, particularly in low-resource settings, and yields outputs that are acoustically more faithful to the speech.

Speech-aware large language models (LLMs) can incorporate speech through pre-trained acoustic encoders that project speech features into the LLM embedding space. While the choice of the speech encoder critically influences performance, different encoders often exhibit complementary strengths, motivating their combination. In this work, we investigate whether fusing multiple pre-trained speech encoders can enhance speech-aware LLMs for automatic speech recognition (ASR). We explore several fusion strategies beyond simple feature concatenation, including learned combinations and Transformer-based fusion architectures, and evaluate them across mono- and multilingual ASR settings as well as diarized speech recognition. Our results indicate that carefully fusing multiple parallel speech encoders improves downstream performance in all scenarios with limited computational overhead.

Speech recognition often fails on rare, domain-specific terms and context-related named entities. Existing contextualization techniques typically bias decoding with keywords or phrase lists, which does not scale well or exploit deeper knowledge. We propose a training method that teaches a speech-LLM to use broad descriptions (e.g. from videos) as weak semantic priors to perform contextual reasoning grounded in the audio. We build 400 hours of reasoning-augmented speech data by pairing erroneous hypotheses with video metadata and LLM-generated reasoning explanations that justify context-driven corrections. We finetune the speech-LLM to perform chain-of-thought reasoning: generate an initial transcript, then reason over the context, and finally return a corrected transcript. On held-out YouTube-derived test sets, our approach reduces errors, with specific improvements on rare words and named entities, and lays groundwork for deeper contextual reasoning in speech recognition.

The proliferation of text-to-speech (TTS) systems capable of generating realistic synthetic speech poses growing challenges for audio forensics. While binary deepfake detection has received considerable attention, source tracing (i.e., identifying which TTS system produced a given audio sample) remains underexplored, particularly in open-set scenarios where unknown systems may be encountered. We propose a metric learning framework based on the Proxy-Anchor loss function that operates on Wav2Vec2-BERT embeddings to learn a discriminative embedding space for TTS source attribution and out-of-distribution (OOD) detection of unseen systems. We evaluate it on the MLAAD v9 dataset spanning 140 TTS systems across 51 languages, and introduce an architecture merging strategy that groups TTS system versions into unified classes, reducing inter-class confusion. Our system achieves 99.76% accuracy on 110 in-distribution classes and a False Positive Rate (FPR@95) as low as 2.04% for OOD detection. Also, for a fair comparison against the current state of the art, we further evaluate it on the MLAAD v5 official dataset splits, improving the OOD accuracy by almost doubling it. These results demonstrate that Proxy-Anchor metric learning, combined with architecture-aware class design and post-hoc OOD scoring, provides an effective framework for forensic TTS source tracing in both closed-set and open-set settings.

To address the heavy-tailed, spike-prone nature of sea clutter and the scarcity of labeled target data, an unsupervised complex-valued variational autoencoder (VAE) for maritime radar target detection is proposed. In implementation, each complex baseband slow-time sequence is represented by its in-phase and quadrature components, and the model learns their joint reconstruction from clutter-only data. A Student-\(t\) negative log-likelihood is adopted to capture heavy-tailed reconstruction errors while reducing sensitivity to outliers during clutter learning. In addition, a time-domain amplitude error constraint is introduced to penalize slow-time magnitude mismatch in the reconstruction. At inference, reconstruction deviation is used as the detection statistic, and the decision threshold is set via an empirical quantile estimated from a clutter-only validation set to enforce a constant false-alarm rate (CFAR). Experiments on measured sea-clutter data show that detection performance is consistently improved over MF, AMF, and a real-valued \(β\)-VAE under CFAR constraints.

Transformer-based Speech Foundation Models excel in most Automatic Speech Recognition tasks but often suffer performance degradation when applied to domains with mismatched acoustic characteristics. While Parameter Efficient Fine-Tuning (PEFT) methods, such as Low-Rank Adaptation (LoRA), adjust global attention, they lack the local context modeling crucial for capturing domain-specific variations. We propose GC-LoRA, a novel adapter architecture that injects Conformer-style local convolutional processing into pretrained Transformer encoders. By integrating a lightweight adapter to encoder attention output projections, our method efficiently captures local acoustic dependencies without disrupting pretrained global representations. Experiments across diverse datasets (acoustically-degraded, bandlimited, dialectal, child) demonstrate the efficacy of our approach, achieving Word Error Rate (WER) reductions of up to 10.9% compared to baselines while adding minimal trainable parameters.

Purpose: To develop an automatic dipole inversion method for quantitative susceptibility mapping (QSM) that preserves fine anatomical structures without the need for manual regularization-parameter tuning. Theory: The original approximate message passing with parameter estimation (AMP-PE) framework models image gradients with a single Laplace prior, which does not fully capture the heavy-tailed gradient distribution of brain susceptibility maps. This prior mismatch can lead to over-regularization and blocky reconstructions. We address this limitation by modeling the gradients with a two-component Laplace mixture prior. Methods: We propose a Laplace-Mixture Dipole Inversion (LAMDI) method by incorporating a two-component Laplace mixture prior into the AMP-PE framework with automatic parameter estimation. LAMDI was evaluated on a public in vivo dataset. Its performance was compared with FANSI, MEDI, and AMP-PE with a single-Laplace prior (AMP-PE-L1) under both standard default and reference-tuned settings. Results: On a public multi-orientation QSM dataset, LAMDI achieved NRMSE and SSIM comparable to AMP-PE-L1 while substantially reducing HFEN, suggesting improved preservation of high-frequency anatomical detail. Under reference-based tuning, FANSI and MEDI achieved the best performance for some metrics, but LAMDI remained competitive without requiring reference maps or manual regularization tuning. Conclusion: LAMDI provides an effective and automatic parameter-estimation alternative for QSM dipole inversion by combining competitive reconstruction accuracy with improved preservation of fine anatomical detail.

Reconfigurable intelligent surfaces (RIS) enable programmable control of wireless propagation but remain vulnerable to persistent deep fades in static deployments. This paper introduces a Movable Antenna-enhanced RIS (MA-RIS) architecture where antenna elements physically reposition to sample independent spatial channels, enabling mobility-induced diversity. We model antenna motion using a Stochastic Differential Equation (SDE) framework capturing controlled drift and environmental diffusion. It^o calculus-based analysis characterizes steady-state antenna distributions, spatial decorrelation, and outage probability, revealing fundamental trade-offs between control strength and mobility randomness. To maximize long-term SNR while accounting for control overhead, we propose an overhead-aware Two-timescale framework separating slow antenna trajectory control from fast phase adaptation. The stochastic optimal control problem is solved via predictive approximation of the Hamilton-Jacobi-Bellman (HJB) formulation, enabling real-time implementation. Simulations validate theoretical predictions: the Two-timescale strategy achieves up to 36 dB steady-state SNR with remarkable stability, outperforming position-only control by up to 15 dB and uncontrolled baselines by over 30 dB. Despite experiencing a lower SNR than Active RIS, the proposed approach delivers up to 16 times higher energy efficiency (EE) across varying system scales, establishing a new paradigm of mobility-enabled channel adaptation for resilient wireless systems.

In this paper, the problem of using curved beams to improve wireless communication performance in the presence of a blockage is studied. In particular, a transmitter equipped with a continuous aperture array can generate curved beams to serve multiple receivers by allowing signals to propagate along both straight and curved paths. To optimize the weighted sum-rate, a curved beam model is developed for controlling the beam steering, beam focusing, and beam curving functions, along with a segmented channel model to characterize practical channels induced by the blockage. Based on the introduced curved beam model, an optimization problem is posed with the goal of maximizing the weighted sum-rate of all users under a transmit power budget and physical constraints of curved beams. To solve this problem, the continuous aperture is first converted into finite summations via a discrete sampling of the continuous coordinate. Then, the performance gap between the ideal continuous aperture design and its practical discrete aperture approximation is analyzed. Based on the above discrete approximation, an iterative algorithm is developed to optimize curved beam control parameters. In particular, the original problem is reformulated as a trackable form via fractional programming (FP). Then, the transformed problem is solved by designing an enhanced block coordinate ascent (BCA) method which determines a surrogate-construction point leveraging the local descent from previous iterations, thereby accelerating convergence. Then, a proximal regularization term is included into the surrogate function to control the update magnitude and suppress aggressive update, thereby improving updates stability. Finally, the beam amplitudes are computed based on the effective channel gains. Simulation results show that the proposed method can improve the weighted sum-rate compared to using only straight beam.

This paper investigates human pose estimation from Orthogonal Frequency-Division Multiplexing (OFDM) signals in an indoor multistatic wireless network operating in the 6 GHz band. We design and validate a processing pipeline that exploits both the amplitude and phase of the Channel State Information (CSI) from multiple radio links to estimate the human body pose. Four deep learning architectures from the literature, namely DT-Pose, MetaFi++, HPE-Li, and VST-Pose, are adapted to the OFDM CSI structure and extended to jointly exploit the amplitude and phase information. The models estimate the pose of a human walking within the network coverage area. Performance evaluation is conducted on an open-access dataset using standard pose-estimation metrics such as Procrustes-aligned Mean Per-Joint Position Error (PA-MPJPE) and Bone Length Loss (BLL). Results indicate that reliable human pose reconstruction can be achieved from 6 GHz OFDM CSI measurements, with DT-Pose providing the best overall accuracy. On average, amplitude-only CSI yields performance comparable to joint amplitude-phase processing, whereas phase information is more beneficial as a complementary feature rather than as a standalone input.

Blood pressure (BP) is a key marker for cardiovascular risk assessment and therapeutic decision-making, and Photoplethysmography (PPG) enables low-cost, wearable-friendly cuffless BP estimation. However, even with recent progress, many PPG-based models are trained with BP regression alone and may rely on amplitude-dominated shortcuts. In addition, demographic covariates that systematically modulate vascular compliance are often incorporated only via late fusion, limiting subject-specific representation learning. We propose a Transformer-based network for cuffless BP estimation from PPG signal, leveraging self-attention to capture long-range dependencies across multiple cardiac cycles. To account for subject-specific vascular differences, the model is conditioned on demographics via FiLM-style feature modulation applied through the attention and feed-forward sublayers of Transformer blocks. In addition, we add an auxiliary morphology head to guide the model to attend to BP-relevant waveform morphology associated with arterial stiffness and wave reflection. Under calibration-based evaluation protocols on the large-scale PulseDB dataset, the proposed method achieves MAE of 4.56 mmHg for systolic BP and 2.62 mmHg for diastolic BP, reducing errors by 47% and 50% compared with prior demographic-enhanced PPG baselines. The resulting lightweight, single-sensor model supports scalable and clinically grounded cuffless BP estimation in calibration-enabled deployment settings.

This study aims to explore the performance of the VAR model in comparison with mel-frequency cepstral coefficient (MFCC) matrices and log-mel spectrograms using deep learning. In pulmonary sound classification, spectrogram-based representations suffer from inconsistent temporal dimensions due to varying respiratory cycle durations. Along with traditional trimming/zero-padding, adaptive-length windowing was presented to fix their temporal dimensions. Their spectral and temporal dimensions were optimized by testing a range of parameters. Different convolutional neural network (CNN) architectures were employed to extract features from the two-dimensional representations obtained over the sub-phases. The extracted sub-phase features were then fused using various strategies including direct concatenation, gated recurrent unit (GRU) network and GRU with attention mechanism. Model performances were assessed through respiratory cycle-based evaluation and subject-based evaluation comprising multiple respiratory cycles. Several data augmentation techniques were also studied to cope with limitations in data size. The best cycle-based F1-score (0.877) was obtained using the MFCC matrices with thirteen coefficients and 64-point time resolution per sub-phase representation followed by direct feature concatenation, and the best subject-based F1-score (0.855) was obtained using the MFCC matrices with thirteen coefficients and 256-point time resolution per full-cycle representation, both obtained by adaptive-length windowing. Augmentation degraded the performance of models overall, yet mixup augmentation was the best among the methods tested. MFCC outperformed log-mel spectrogram and VAR model in differentiation of asthma and COPD. Sophisticated fusion strategies did not improve the diagnosis. Augmentation did not contribute, demonstrating the significance of authentic data in pulmonary sound studies.

Unsupervised term discovery involves segmenting unlabelled speech into word- or syllable-like units and clustering these into a lexicon of candidate types. True lexicons follow a Zipfian distribution, yet the dominant centre-based clustering approach -- K-means -- produces a more uniform distribution due to an inductive bias toward spherical clusters. In this paper we revisit graph-based clustering as a bottom-up alternative, where segment embeddings are connected by pairwise similarity and partitioned using the Leiden algorithm. We show that graph clustering substantially outperforms centre-based approaches (K-means, GMM, BIRCH) in both word- and syllable-level lexicon discovery across three languages, producing more Zipf-like distributions. Another bottom-up approach, agglomerative clustering with average linkage, also performs well, although it is computationally less efficient and allows for less control over the resulting distribution. Our work calls into question the dominance of centre-based clustering for term discovery, and promotes graph clustering as an attractive alternative.

Recent multimodal large language models mainly process audio as monaural signals, thereby discarding the spatial cues contained in spatial audio for sound localization, spatial relation reasoning, and spatial scene understanding. We propose Spatial-Omni, a lightweight method that implements SO-Encoder to inject First-Order Ambisonics (FOA) spatial audio into existing Omni LLMs as an independent modality, without modifying their original audio encoders. SO-Encoder provides spatial tokens with limited additional context cost and improves spatial audio understanding through efficient staged training. To support training and evaluation, we construct SO-Dataset, SO-QA, and SO-Bench from open-source data, real recordings, and simulations, containing 400K FOA spatial audio clips and 2.1M spatial question answering pairs. SO-Bench covers 16 spatial audio understanding subtasks, including basic detection and location estimation, spatial relation understanding, and complex spatial reasoning. Experiments show that Spatial-Omni outperforms existing open-source Large Audio-Language Models (LALMs) and Omni LLM models on spatial audio understanding tasks while retaining a reasonable level of general audio understanding. Code and data are available at https://github.com/dieKarotte/Spatial-Omni.

The nnU-Net has demonstrated continuous success in medical segmentation tasks, which heavily rely on the availability and diversity of annotated biomedical data. However, assembling medical imaging cohorts remains challenging due to numerous factors such as privacy regulations and annotation costs. As a result, data augmentation plays a crucial role in increasing data availability while maintaining anatomical feasibility. Hence, we propose the ++nnU-Net, a novel data augmentation module based on image registration that operates prior to preprocessing and training take place. Our framework was evaluated across five different 2D datasets. In this workflow, image data go through a two-stage registration process, generating new warped images. The transformations are then applied to the respective segmentation. In addition, the pipeline computes available disk space, generates supplementary binary synthetic masks and generates checkpoints. We demonstrate that the ++nnU-Net outperforms the nnU-Net baseline, yielding improvements in Dice Similarity Coefficient scores. In the most prominent cases, we observe performance gains of approximately 22\%. These findings highlight the effectiveness of registration-based data augmentation, particularly for 2D medical imaging datasets and suggest that the ++nnU-Net provides a practical and scalable approach for enhancing segmentation performance in data-limited settings. The source code for the ++nnU-Net is available at: https://github.com/sofia-adelie/plusplusnnunet.git

While Speech Large Language Models (Speech-LLMs) have achieved strong performance on adult Automatic Speech Recognition (ASR), their effectiveness on child speech remains under-explored, and single models often struggle to handle diverse adult and child age groups simultaneously. This paper proposes a Mixture-of-Experts (MoE) Speech-LLM for unified ASR across adult and child speech spanning diverse environments and age groups. The framework employs a Classifier-based Domain Router (C-DR) with a coarse-to-fine strategy and integrates both a Mixture-of-Projectors (MoP) and a Mixture-of-LoRAs (MoL) to model domain-specific variations. To address routing uncertainty near domain boundaries, an Entropy-Aware Routing (EAR) mechanism is introduced to dynamically incorporate a shared expert. Experiments on public child corpora demonstrate consistent improvements over baselines while preserving adult ASR performance. To our knowledge, this is the first work leveraging Speech-LLMs for unified, multi-domain ASR encompassing both children and adults.

We introduce SSL-GMMVC, an interpretable voice conversion method in self-supervised speech space. The method models paired source-target features with a Gaussian mixture model and performs conversion as a posterior-weighted sum of affine transforms. This yields locally linear transformations that adapt to heterogeneous feature-space structure while remaining analytically tractable. Through objective and subjective evaluations, we show that SSL-GMMVC improves speaker similarity with comparable intelligibility and naturalness, and that even a constrained covariance variant surpasses a deep learning baseline as the number of mixture components increases. Further analyses link component selection to phonetic structure and reveal interpretable scaling and rotation in the learned transforms. These findings highlight SSL-GMMVC as an effective, analyzable framework for voice conversion.

Coordinated multi-satellite (CoMS) transmission and non-orthogonal multiple access (NOMA) are envisioned to jointly enhance coverage, capacity, and spectrum efficiency for satellite networks. Their integration into a unified CoMS-NOMA framework will allow more efficient, reliable, and energy-efficient multi-user access. This paper investigates the downlink performance of CoMS-NOMA networks from a system-level perspective, in which multiple satellites cooperatively serve multiple users via NOMA. Leveraging tools from stochastic geometry, related angles and distances in CoMS-NOMA are first derived as intermediate results. Then, we obtain the combined signal power distributions and analyze coverage and spectrum performance under both inter- and intra-satellite interference, accounting for potential imperfect successive interference cancellation (SIC). The analytical model is validated across a range of system parameters, including the number of satellites, service region angle, error-propagation factor, and power allocation coefficients. Numerical results indicate that increasing the number of cooperative satellites does not always improve coverage and spectrum efficiency. Additionally, while a higher main-lobe gain improves coverage, a near-perfect SIC provides only slightly greater benefits than a reasonably good SIC. With properly selected power allocation coefficients, CoMS-NOMA achieves up to a 270% improvement in coverage and a 56% gain in sum spectral efficiency, compared with conventional orthogonal and single-satellite schemes, indicating potential for green, energy-efficient satellite networking.

In this study, we propose an overlapped wavelet diffusion framework for Low-Light Image Enhancement (LLIE), which incorporates two complementary components to achieve blocking artifact-free and detail-preserving enhancement. Although recent diffusion-based LLIE methods have demonstrated remarkable performance compared with traditional approaches, DiffLL still suffers from blocking artifacts caused by the Haar Wavelet Transform (WT) and blurred edges or over-smoothed textures due to the limitations of its High-Frequency Restoration Module (HFRM). To overcome these issues, we introduce an Overlapped WT (OWT) that incorporates correlations across neighboring regions, thereby structurally preventing blocking artifacts. Furthermore, we integrate a low-frequency-guided High-Frequency Enhance Block (HFEBlock) to strengthen detail recovery, yielding sharper edges and more reliable textures. Extensive experiments on the LOLv1 and LOLv2-real datasets demonstrate that our framework, termed OWDiff, consistently outperforms existing LLIE methods both qualitatively and quantitatively, achieving superior visual quality while maintaining computational efficiency. OWDiff effectively addresses the structural limitations of the Haar WT and the HFRM, achieving an average PSNR gain of 0.58 dB, along with a 1.64% relative improvement in SSIM and a 5.9% relative reduction in LPIPS, compared to DiffLL across both the LOLv1 and LOLv2-real datasets.

While speech quality is typically assessed on complete utterances, streaming and generative systems require incremental estimation from partial audio. Existing predictors assume full context, degrading on prefix-constrained inputs. Extending ARECHO, we propose ANCHOR, reformulating incremental assessment as a multi-resolution autoregressive task. It models chunk- and utterance-level quality within a single decoder using dual-resolution tokens and a resolution-aware hierarchy for coarse-to-fine refinement. Experiments show substantial robustness under partial input, including a 48% PLCMOS error reduction on 2-second prefixes. Convergence analysis reveals a 4-6 s effective perceptual context horizon. A stress test further isolates structured extrapolation biases under localized corruption. Results demonstrate that hierarchical supervision improves incremental prediction and elucidates how perceptual quality accumulates over time.

Recent speech-aware large language models (Speech-LLMs) rely on a pre-trained speech encoder to convert audio into semantic-rich representations consumable by LLM. In this work, instead, we explore: can an LLM learn to read Mel spectrogram directly without a dedicated speech encoder? We propose Mel-LLM, an encoder-free Speech-LLM that feeds lightly pre-processed Mel spectrogram patches directly into the LLM through a linear projection, allowing the LLM to learn speech-text alignment purely through its own parameters. We conduct extensive experiments on both automatic speech recognition (ASR) and text-to-speech (TTS) tasks. For ASR, we evaluate on the OpenASR leaderboard public sets and production-level scaling experiments, demonstrating that the encoder-free solution achieves competitive performance with only limited degradation compared to encoder-initialized counterparts. We find that when data is limited, initialization from a multimodal checkpoint (Phi-4-MM) is crucial for maintaining performance. We also present ablation studies revealing which LLM layers are less relevant to speech encoding. For TTS, we show preliminary results with a next-token VAE approach. While TTS performance is not yet optimal, these results establish the feasibility of a fully unified encoder-free architecture for autoregressive speech-text modeling.

Evaluating text-to-music (TTM) systems remains expensive because music impression (MI) and text alignment (TA) scores rely on human mean opinion scores (MOS). Most automatic MOS estimators are trained with point-wise regression or distributional classification. These objectives do not directly optimize rank-based metrics and provide weak geometric constraints for cross-modal coherence. To address these gaps, we propose DeRA-MOS, a decoupled optimization framework for TTM evaluation. For MI, we introduce a batch-aware listwise ranking loss that models relative order within each mini-batch and better aligns with evaluation based on Spearman's rank correlation coefficient (SRCC). For TA, we introduce a score-anchored modality alignment loss that maps human scores to target audio-text similarity and regularizes the latent space before fusion. By effectively mitigating the point-wise training mismatch and modality drift, experiments on MusicEval demonstrate that our decoupled framework yields substantial improvements in both MI and TA ranking metrics, establishing a robust paradigm for large-scale TTM evaluation.

Head computed tomography (CT) typically uses sub-millimeter in-plane resolution but 2-5 mm through-plane spacing, creating substantial anisotropy that degrades multiplanar reconstructions, volumetric measurements such as hematoma volume estimation, and downstream algorithms that assume near-isotropic voxels. We present a deep learning system that synthesizes intermediate CT slices from pairs of neighboring axial slices, halving the effective through-plane spacing. The system improves three-dimensional visualization while simultaneously producing inherently denoised outputs, yielding two complementary benefits from a single inference pass. To build a reliable system, we systematically evaluate pixel-wise losses, namely mean squared error (MSE) and mean absolute error (L1); structural-similarity losses, namely the structural similarity index (SSIM) and its multi-scale variant (MS-SSIM); and hybrid combinations. On a held-out test set, all converged models outperform classical interpolation baselines and pretrained video frame interpolation methods (RIFE, FILM) on all structural measures, with MS-SSIM+L1 offering the strongest balanced profile. We also document training instability in SSIM-family losses and identify partial remedies: the standard numerical fixes eliminate the dominant failure mode but leave residual divergence at smaller batch sizes. All results are reported with patient-level bootstrap confidence intervals and paired statistical tests. As an illustration, we apply the system to an out-of-distribution head CT series from Hospital Universitario Virgen del Rocío: the model synthesizes intermediate slices and exhibits on the real slices the implicit-denoising signature predicted by our theoretical analysis, supporting in a single external case that interpolation quality and implicit denoising are not confined to the training distribution.

Diffusion MRI (dMRI) tractography is the only noninvasive approach for mapping white-matter pathways in the living human brain. It represents each brain as a tractogram: a large, unordered set of three-dimensional streamlines that includes information about both local streamline geometry and whole-brain anatomical organization. This structure makes tractograms a natural but challenging target for representation learning. Existing methods treat streamline classification and subject-level prediction as separate problems: streamline classifiers focus on geometric patterns, whereas subject-level prediction often depends on hand-crafted features. As a result, current methods do not learn reusable representations that connect streamline anatomy with whole-brain inter-subject variation. Here we introduce TractFM, a tractogram foundation model that learns reusable representations directly from whole-brain streamline sets. TractFM combines a local streamline encoder with a permutation-equivariant tractogram encoder, allowing all streamlines from a subject to be contextualized jointly in a single forward pass. Pretraining on dense anatomical tract parcellation, i.e., assigning anatomical labels to individual streamlines, yields two complementary representations: contextualized streamline-level embeddings for tract parcellation and compact subject-level descriptors for downstream prediction of subject phenotypes. Across three tractography algorithms and five dMRI datasets, TractFM transfers to both streamline-level and subject-level tasks. Its frozen representations achieve accurate tract parcellation and predict age and sex across independent datasets. These results show that whole-brain geometric context, learned once, can generalize across tractography pipelines, datasets, and prediction tasks.

Full-duplex spoken dialogue models can listen and speak simultaneously, making them a promising architecture for natural conversation. However, current models are trained solely with supervised learning through token-level likelihood maximization, which does not directly optimize interaction-level behaviors, causing interactivity issues such as excessive silence and ill-timed turn-taking. Recent work has applied reinforcement learning (RL) to improve interactivity, but existing methods address only a limited set of interactive behaviors in their rewards. In this work, we propose a post-training alignment method that comprehensively improves the interactivity of full-duplex spoken dialogue models through RL. We address the four canonical axes of interactivity: pause handling, turn-taking, backchanneling, and user interruption. For each axis, we extract short audio segments from human conversation corpora and optimize the model with axis-specific reward functions. An extra LLM-based reward for response quality prevents semantic degradation. We apply our method to two open-source models, Moshi and PersonaPlex, demonstrating consistent improvements in interactivity on both offline evaluation with pre-recorded audio and real-time multi-turn dialogue evaluation.

Network control theory can be used to model intrinsic and extrinsic strategies to steer neural dynamics. Standard approaches are node-centric, structural, and focused on achieving desired instantaneous states. Here, we develop an edge-centric approach which incorporates both structure and function to achieve extended patterns of neural dynamics characterized by desired synchronization states. Our method, Quantifying Underutilized Influential Edges for Targeted Synchronization (QUIET), is an edge-centric framework that integrates structural controllability of individual white matter connections and mutual information between pairwise functional timeseries to identify energy-efficient synchronization pathways. QUIET identifies quiet highways, edges that are structurally influential but functionally underutilized, to optimize regional synchronization. We validated QUIET across 75 synthetic configurations, where QUIET-ranked edge sets significantly outperformed random selection in 93% of cases (p<0.01). The framework, tested on Human Connectome Project participants, revealed that the control energy required for synchronization of the salience network correlates with fluid intelligence. QUIET, applied to healthy adults undergoing dexmedetomidine-induced unresponsiveness, showed that the frontoparietal and default-mode networks exhibited the largest control energy required for synchronization in both awake and sedated states. QUIET is released as a stand-alone software to be used to study theoretically-defined synchronization pathways, which in turn could inform testable hypotheses in perturbative studies.

Effective demand response in smart microgrids requires prosumers to cooperate voluntarily under strategic self-interest, a coordination problem structurally equivalent to a repeated Prisoner's Dilemma on a social network. This paper presents a multi-agent simulation in which a Large Language Model (LLM) Influence Compiler issues structured demand-response directives to a population of heterogeneous prosumer agents, each governed by a hybrid decision architecture combining game-theoretic base probability (derived from payoff history, neighbour imitation, and exploitation memory) with LLM narrative evaluation of incoming coordination signals. The hybrid architecture resolves a key methodological challenge: LLMs aligned via Reinforcement Learning from Human Feedback (RLHF) exhibit strong cooperation bias when used as direct decision-makers, producing flat dynamics regardless of grid conditions. By separating strategic reasoning from grounded narrative evaluation, the model generates realistic prosumer behaviour across six personality archetypes, with baseline cooperation near 50% and clear differentiation under influence. Compiled structured directives achieve 33.3% demand-curtailment cooperation versus 27.0% for unstructured messaging and 28.0% for a no-intervention baseline ($Δ_\mathrm{comp} = +0.063$), with the advantage preserved across both grounded and idealized agent substrates ($Δ= +0.083$) and across all resistance levels ($R = 0.1$ to $0.7$). Hub-targeted dissemination via high-centrality network nodes outperforms peripheral or random targeting, confirming that grid topology provides mechanistic amplification independent of message content. These results suggest that structured LLM compilation, grounded agent reasoning, and network-aware targeting are complementary design principles for scalable, interpretable demand-response coordination in smart-city energy systems.