Papers

This state-of-the-art report provides an overview of controllable 3D human avatar creation. We describe current 3D avatar systems, which typically consist of three stages: (i) learning priors of human appearance and motion, (ii) creating a personalized avatar, and (iii) animating the avatar. To limit the scope, we focus on the prior learning and avatar creation stages. We define current avatar representations and introduce a taxonomy that categorizes existing work along multiple axes, including body regions and employed priors. We review methods for full-body and head avatars, as well as layered representations that decompose the body into components such as hands, hair, and garments. Finally, we outline common underlying principles, reference key literature for newcomers, and discuss open challenges and future research directions.

Audio-conditioned generators now produce synthetic sound effects from real recordings, so the real and synthetic versions of an event increasingly coexist in sound libraries and in the corpora used to train audio models – yet no benchmark measures whether a representation can match a synthetic clip to the specific real recording it was generated from. I introduce Doppelganger, a benchmark for matching sound effects across the synthetic-real boundary, pairing 10,420 real clips across 34 everyday sound events each with an audio-conditioned synthetic twin, alongside a controlled 7-class corpus. Off-the-shelf audio encoders do not cross the boundary cleanly. Making the embedding ignore the boundary the standard way – training it on sound-event labels – works on familiar sounds but backfires on new ones, dropping below the untrained encoder. Training on the pairs instead – a clip and its own synthetic twin – generalizes. On sound events held out of training, it recovers the exact real source about 80% of the time (up from 61% untrained; chance 0.03%), whereas no objective meaningfully improves category-level recognition on those unseen events. The learned matching is specific to one generator – it survives changes to that generator's settings but not a switch to a different generator, and collapses for the text-only generators tested. A human annotation baseline (49 listeners) lands well above chance but below the models on the same trials. Synthetic twins fool people into calling them real about 29% of the time, yet a generator-specific detector separates these audio-conditioned twins from real recordings perfectly.

Most spoofing countermeasures now place a light classifier on top of a self-supervised (SSL) speech encoder. A growing line of work adds handcrafted acoustic features and fuses them by cross-attention, partly because the attention weights appear to explain which cues the model relies on. Two things about such fusion remain untested: whether the handcrafted features contribute anything once a strong SSL encoder is already in place, and whether the attention map faithfully reflects what the classifier actually uses. We study both questions with MOSAIC, a single model that handles logical access (LA) and physical access (PA) attacks by projecting a 152-dimensional biophonetic vector into six query tokens attending over thirteen intermediate layers of WavLM-Large. Rather than pursuing state-of-the-art accuracy, we treat the model itself as the object of analysis and apply inference-time interventions that zero out selected components. Removing the handcrafted branch lowers the EER by 0.67 percentage points under LA, so WavLM alone is sufficient there, but raises it by 0.76 points under PA, where the branch supplies replay channel evidence the encoder does not carry. The 6x13 attention map is highly stable under resampling (bootstrap rank correlation 0.99), yet its faithfulness varies by domain: layer attention weights predict the causal importance of each layer under PA (rho = +0.62) but not under LA (rho = -0.28). The handcrafted branch and its attention-based explanation are therefore trustworthy for replay attacks and not for synthetic ones. On standard benchmarks the model is mid-range on LA and ahead of the official baselines on cross-source deepfakes (6.21% EER on ASVspoof 2021 DF). The contribution is not a new architecture but a checking procedure: verify handcrafted cues and attention maps by intervention before trusting either.

We develop a unified, certified lower bound on the time-to-boundary margin M for transient stability of interconnected dissipative systems under slow parameter drift. The companion work establishes M as the first-passage time of the joint state-parameter motion to the synchronism boundary and proves M = CCT exactly on the one-machine-infinite-bus reduction, while leaving the multimachine certified margin open. Here a composite (mixed-region) Lyapunov function, formed by absorbing the restoring intra-group coupling into group energy functions and treating only the residual cross-cut coupling through the comparison principle, yields a positively invariant inner estimate of the region of attraction whenever an associated test matrix is a nonsingular M-matrix. The certified region breathes with the drift: its size is governed by a single critical synchronising stiffness k_c (lambda), and as k_c -> 0 at the boundary the region breathes shut and the certified margin M_low <= M_true vanishes. We give a nonlinear sector form of the construction, a domain-neutral resilience-fragility reading in which the coupling that certifies order is the one whose growth certifies collapse, and a constructive control corollary establishing a sharp dichotomy between damping injection and structural action. The mechanism is demonstrated identically on the WSCC nine-bus power system and on an inertial Kuramoto network, whose normalised breathing curves collapse, to leading order, onto a single profile. We present this collapse as numerical evidence for a conjectured universal form; a normal-form proof is identified as the precise open step.

Artificial intelligence (AI) and machine learning (ML)-based channel estimators silently degrade when propagation conditions drift from their training distributions. This letter proposes a model-agnostic cognitive digital twin (CDT) framework that combines a variational autoencoder (VAE) with latent activation monitoring to detect distribution drift and autonomously execute continue, update, or retire lifecycle actions without requiring ground-truth channel knowledge. The proposed framework is fully compatible with the AI-native lifecycle management envisioned in 3rd Generation Partnership Project (3GPP). Simulations over various channels demonstrate accurate drift detection and robust channel estimation, consistently outperforming conventional offline-trained deep learning estimators under moderate and severe channel drift.

Tensor probabilistic independent component analysis (TPICA) is a popular approach to analyzing functional magnetic resonance imaging (fMRI) data, which draws its popularity from its ability to enrich the advantages of the statistics-based ICA with the awareness of the multi-way nature of these data, brought about and exploited via a deterministic 3-way (time $\times$ space $\times$ subjects) tensor decomposition (Canonical Polyadic Decomposition (CPD)) model. It has, however, received critique concerning its robustness in realistic fMRI unmixing scenarios, notably those involving sources that are strongly overlapped in space. Such cases may not meet the assumption of statistical independence required in ICA. They can instead be better described as independent vectors (or subspaces) of dependent components, pointing to the adoption of alternative statistical approaches, notably independent vector analysis (IVA). On the other hand, on the deterministic side, CPD is often restrictive and is outperformed by the more flexible block-term decomposition (BTD) model, also in the fMRI source unmixing context. Given the above, plus strong evidence of links between IVA and BTD, it is deemed worthwhile to consider the possibilities of generalizing TPICA to a BTD-based “TPIVA" extension, which would more successfully combine the power of statistics and tensor decomposition. This could also entail a generalization of the BTD model, where (non)collinearity would be replaced by statistical (in)dependence. This note aims to outline the state-of-the-art and the above ideas in more detail, serving as a preliminary, motivating step in this research direction.

With the development of low altitude intelligent systems, multiple unmanned aerial vehicles (UAVs) can collaboratively execute more complex tasks. Conventional task allocation methods usually regard tasks and UAVs as isolated entities, making it difficult to capture task dependencies and UAV communication relationships. To address this issue, this paper proposes a dual heterogeneous graph learning based UAV task allocation method. A directed task graph is constructed to represent task dependencies and encode task resource requirements, while an undirected UAV communication graph is built to model communication relationships and encode UAV resource states. The task allocation problem is formulated as a structural matching problem between the task graph and the UAV communication graph. A graph attention network based feature extraction method is introduced to learn structural representations from both graphs through message passing. A cross attention mechanism is further integrated with proximal policy optimization to optimize the matching between task nodes and UAV nodes for task allocation. Simulation results demonstrate that the proposed method achieves a higher task completion rate and shorter task completion time than benchmark methods under different evaluation settings. Furthermore, a UAV sensing and computing application is developed on the AirSim simulation platform. A large language model is employed to convert natural language task requirements into a structured task graph for autonomous UAV task execution, demonstrating the potential of the proposed framework for natural language driven UAV mission planning and execution.

Future sixth generation (6G) communications are expected to support robotic control tasks in applications such as industrial automation and emergency response, where sensors, computing units, and robots are interconnected via nervous system-like networks to form sensing-communication-computing-control (SC3) closed loops. However, the limited battery capacities of devices within these SC3 loops constrain operational duration and degrade control efficiency, particularly in remote or post-disaster scenarios. To address this challenge, wireless power transfer (WPT) can be leveraged to provide continuous energy supply for SC3 closed loops. In this paper, we investigate a wireless-powered SC3 system, where a satellite transfers energy via radio frequency (RF) signals to support the communication and computing processes of multiple SC3 closed loops. By accounting for the intricate coupling among computing, communication, and energy transfer, we propose a holistic design framework to enhance overall control performance. Specifically, we adopt the linear quadratic regulator (LQR) cost as the performance metric and formulate a sum LQR cost minimization problem. The uplink/downlink transmit power, bandwidth allocation, computing capability, communication/computing time allocation, and WPT power allocation are jointly optimized. We recast the problem into a more tractable form and develop an iterative algorithm to solve it. For the special case of a single loop, we further analyze the properties of optimal solutions in energy-limited scenarios to provide insights for practical parameter configuration. Simulation results demonstrate the performance gains of the proposed scheme.

AI-RAN aims to unify artificial intelligence and radio access network workloads on a shared compute substrate. While this paradigm has so far been demonstrated primarily on Graphics Processing Units (GPUs), it remains unclear whether Neural Processing Units (NPUs), which are AI accelerators optimized for inference, can also support wireless baseband processing. Here, we provide the first affirmative answer by resolving the fundamental mismatch between baseband workloads and NPU architecture. A computational isomorphism exists: matrix and vector engines NPUs dedicate to inference inherently cover physical-layer operations. Yet NPU architectures are natively shaped for dense-tensor AI inference, not baseband. This architectural mismatch surfaces as opposing optimization objectives: traditional baseband minimizes arithmetic operations, whereas NPU performance demands maximizing engine utilization. We close this gap by reconstructing communication algorithms onto AI compute primitives, prioritizing engine utilization over arithmetic count. We validate this with a complete OFDM transceiver on an Ascend 310B1 edge NPU, demonstrating end-to-end over-the-air transmission via USRP X300 at 3.0 GHz.

We study a receding horizon game in which multiple agents drive linear systems subject to additive disturbances, private state and input constraints, and shared coupling constraints. We propose a robust game-theoretic control framework that combines tube-based constraint tightening with a finite-horizon generalized Nash equilibrium problem (GNEP), equipped with a discrete algebraic Riccati equation (DARE)-based terminal cost and a decoupled positively invariant terminal set. The framework guarantees recursive feasibility for every bounded disturbance realization. Exploiting the potential-game structure induced by tracking costs, we further establish asymptotic convergence of each agent's nominal state to a steady-state variational generalized Nash equilibrium (vGNE), and show that each agent's actual state converges to a neighborhood of the vGNE determined by the minimal robust positively invariant set.

In MRI, dense receiver coil arrays with a high number of coil elements are used to efficiently detect and encode the signal. Further increasing the number of coils is hampered by electrical cabling and massive electronics that introduce electromagnetic coupling, integration complexity and even safety constraints. Here we introduce the novel Light Coils concept, a fully optical MRI receive architecture in which data transmission, front-end power delivery, and coil detuning are all implemented optically, thereby reducing the massive galvanic cabling to a few optical fibers. For signal encoding, Mach-Zehnder modulators (MZM) are used to convert the MR signal from each coil onto a C-band optical carrier. The preamplifiers are driven via a power-over-fiber (PoF) system that uses a high-efficiency photovoltaic (PV) cell for optical-to-electrical power conversion. A pulse-sequence-triggered optical path controls active detuning. Jointly optimizing modulator bias, optical power and front-end gain under realistic receiver chain conditions, Light Coils can match the signal-to-noise ratio (SNR) of conventional RF coil systems with galvanic cables at MZM input powers of 5-10mW and photonic power converter inputs of 80-100mW. At a clinical 3T MRI system, we show in vivo human brain imaging with a single-channel Light Coil element with an image quality and SNR comparable to a conventional coaxial readout using the identical coil element. Extending the concept to a four-channel array using dense wavelength-division multiplexing over a single fiber, we demonstrate wavelength-selective routing with inter-channel optical isolation exceeding 28dB, reduced noise correlation compared with the galvanic reference, and parallel imaging. These results establish a scalable route towards lightweight, modular, and potentially ultra-dense MRI receive arrays based on integrated photonics and power-over-fiber.

Neural speech codec has attracted extensive attention for high-quality reconstruction at low-bitrate. However, real-world noise severely degrades its performance and hinders high-quality clean speech reconstruction. To tackle this problem, we propose FocalSE, a novel speech enhancement method that performs feature denoising, noise feature separation and noise recognition in the continuous embedding space of neural speech codecs. Specifically, we develop focal modulation-based compression and decompression to capture global context and local mutual information, and generate focal masks to recover clean feature embeddings. We then separate noise embeddings from noisy embeddings to improve denoising performance. Finally, we use ResNet1D-18 to recognize noise categories for better separation effectiveness. Extensive experiments on two standard datasets, LibriTTS and ESC50, demonstrate that our method outperforms state-of-the-art approaches under low-bitrate and low-SNR conditions.

Non-line-of-sight (NLOS) imaging is an emerging technique with transformative potential, enabling the visualization of hidden objects through indirect light reflection. This paper presents a NLOS imaging method operating in the near-infrared (NIR) wavelengths, specifically employing a raster scanning technique with a pan-tilt device. The NIR laser, operating at a wavelength of 808 nm and an output power of 500 mW, illuminates a hidden target occluded by an obstacle. The imaging process involves three bounces: the laser beam first strikes a relay wall, then reflects off the hidden target, returns to the relay wall, and subsequently reaches the NIR camera. This study systematically evaluates the effectiveness of the proposed method across three distinct targets, demonstrating the capability to recover high-quality images from non-line-of-sight scenarios. The obtained images of the hidden targets are compared with their ground truth images, and the error in the obtained images is assessed based on the criteria of Mean Squared Error (MSE) and Root Mean Square Error (RMSE).

This paper investigates the causal rate-distortion function for networked control systems with multiple encoders and a single decoder, a longstanding open problem in information and control theory. While previous work has explored the causal rate-distortion function for single-encoder and feedback-enabled networked settings, the case of networks without feedback remains unaddressed. We establish a novel directed information lower bound, the first derived for the networked control setting. We further demonstrate the optimality of linear, independent encoders and linear decoders for optimizing this lower bound for Linear Quadratic Gaussian (LQG) plant and quadratic cost, with the condition that the full plant state is observed when sensors are sitting together. By reducing the original infinite-dimensional optimization problem to a finite-dimensional one, our approach simplifies the analysis. Additionally, our directed information lower bound provides an alternate proof for the sufficiency of linear encoders in the single encoder and single decoder setting with side information, extending prior results in the literature. We present Semidefinite Programming formulations for the causal rate distortion function of Gaussian-Markov sources with linear side information and the singular noise matrix.

This paper proposes Scenario-Based Data-Enabled Predictive Control (Scenario-DeePC), which integrates the scenario optimization framework into Data-enabled Predictive Control (DeePC) to provide probabilistic guarantees on constraint satisfaction under uncertainty. In contrast to existing methods, the uncertainty is characterized directly from data by constructing empirical disturbance scenarios from observed prediction errors, keeping the method fully consistent with the data-driven philosophy of DeePC and free of distributional assumptions. We establish the supporting theory, including a distribution-free probabilistic guarantee on constraint satisfaction and recursive feasibility of the receding-horizon scheme. An adaptive extension collects scenarios online, enabling the controller to adjust to changing noise characteristics, disturbances, and operating-point-dependent model mismatches. The approach is demonstrated on a linear Boeing 747 model and a nonlinear two-tank system, showing a significant reduction in constraint violations compared to standard DeePC, while maintaining comparable tracking performance in nominal conditions and improving tracking accuracy in the nonlinear setting.

An uplink multiuser pinching-antenna system (PASS) is considered, where multiple dielectric waveguides are deployed at the base station and one pinching antenna (PA) is activated on each waveguide. For practical implementation, each PA is restricted to a finite number of preconfigured locations. The resulting uplink sum-rate maximization problem is represented as a layered tree search. Three algorithms are then developed: a greedy search (GS), a beam search (BeS), and an optimal branch-and-bound (BnB) search. In GS, the locally best branch is selected through efficient matrix-inverse updates. In BeS, several promising partial paths are retained to provide a tunable performance-complexity tradeoff. In BnB, noncompetitive subtrees are pruned through a monotonic transformed objective without loss of optimality. Substantial gains over a conventional fixed array are demonstrated by numerical results. Near-optimal performance is also achieved by GS and moderate-width BeS at a lower computational co t than BnB.

Federated learning (FL) is severely hindered by statistical heterogeneity due to variations in scanners, acquisition protocols, and patient populations. Such non-IID data induces client drift during local optimization, leading to unstable convergence and suboptimal global models when parameter-based aggregation is applied. We propose a prototype-based, influence-aware federated learning framework (FedProIn) that uses multiple learnable class prototypes to capture shared semantic structures across heterogeneous clients. We introduce feature divergence loss and prototype contrastive loss to mitigate client drift by decomposing it into feature drift and prototype drift. In addition, we propose a normalized influence aggregation strategy that adaptively weights client prototypes according to their contribution to the global representation, reducing the impact of biased or low-quality updates. Experimental results on two publicly available medical datasets, HAM10000 and Matek-19, demonstrate that FedProIn achieves accuracies of (83.5% IID, 81.1% non-IID) on HAM10000 and (96.2% IID, 95.8% non-IID) on Matek-19, respectively, outperforming existing baselines in both conditions. Our code is available at https://github.com/harsh-kmr/FedProIn.

Autoregressive (AR) text-to-speech (TTS) models generate discrete speech tokens sequentially, which makes inference slow and can degrade robustness, since local errors propagate to later positions and can escalate into hallucination. This limitation stems from their left-to-right AR commitment: each token must be determined before future speech-token context is available. However, such ordering is not an inherent requirement for TTS, since the model receives the full input text before synthesis. In this paper, we introduce DELTA-TTS, a lightweight LoRA-based adaptation framework that converts a pretrained AR TTS model into a discrete diffusion language model (dLLM) for confidence-ordered speech-token decoding. To better capture the local structure of speech, DELTA-TTS incorporates a convolution module that injects local acoustic context, together with a 1/t-weighted training objective and a time-shifted inference schedule that together defer low-confidence positions to later steps. Trained on only 585 hours of LibriTTS, DELTA-TTS achieves a 1.75% WER on Seed-TTS test-en, outperforming its AR backbone while generating tokens 3.3 times faster. Further analysis shows that DELTA-TTS produces sharper text-speech alignment, increases overall decoding confidence, and mitigates the hallucinations observed in AR generation. Audio samples, inference code, and model weights are available at https://slpr-lab.github.io/DELTA-TTS.

Unsupervised syllabic tokenization aims to learn discrete syllabic tokens that capture latent linguistic content-related structure from raw speech. Recent syllabic tokenization methods employ teacher-student distillation of the pretrained HuBERT to organize latent speech frame representations into syllabic segments. However, when trained with an utterance-level cross-entropy objective, the model predicts speaker identity rather than linguistic content, thereby compromising the purity of syllabic tokens. To address this problem, we propose a speaker-disentangled syllabic tokenizer that regresses speaker-perturbed student representations toward clean teacher targets within fixed-length chunks. Experimental results demonstrate that our proposed method achieves state-of-the-art performance in syllable boundary detection and syllabic segment clustering. Moreover, a speech language model trained on our syllabic tokens achieves a 7% relative improvement in syntactic and semantic understanding over the phone-level SpiRit-LM.

We present GEPARD (Generative, Prosody-aware, Autoregressive text-to-speech model for Realtime Dialogue), a streaming text-to-speech model for real-time spoken dialogue. GEPARD generates speech autoregressively with an LLM backbone - text and audio embeddings are trained together in a single decoder-only model - and decodes it to a waveform with an FSQ-based neural codec, streaming audio chunk-by-chunk as text arrives. Our central goal is a TTS architecture served by a standard LLM engine (vLLM) without modifying its compute kernels. This defines the overarching design principle: the backbone is a standard full-attention transformer, while all non-trivial auxiliary mechanisms - zero-shot voice cloning, text augmentation, and classifier-free guidance - are moved out of the autoregressive decode loop into prefill, or distilled directly into the weights. On streaming end-to-end inference, a single stream reaches a Real-Time Factor of about 0.067 (roughly 15x faster than real-time); under 256 concurrent streams the system reaches an aggregate speedup of about 204x on a single server-class GPU. We detail: (1) system-level solutions for vLLM-native serving; (2) the "short register" (1-2 word) failure mode of autoregressive speech decoders, with diagnostic probes and a mitigation; and (3) distillation of two-pass classifier-free guidance over text into single-pass weights via Direct Preference Optimization (DPO).

This study analyzes patterns of stress and exhaustion among student pilots throughout flight training using a combination of physiological and self-reported measurements. The Perceived Stress Scale (PSS-10) was used to measure perceived stress and exhaustion before and after each flight, while physiological data, including heart rate (HR), electrodermal activity (EDA), skin temperature, and acceleration, were continuously recorded during flight sessions. To identify recurring patterns in arousal and workload, physiological signals were preprocessed and analyzed across the flight stages. The findings indicate a buildup of workload-related weariness over time, as evidenced by steady increases in EDA and skin temperature across flights, as well as post-flight increases in self-reported exhaustion. Heart rate responses were more event-specific, with brief spikes during high-demand phases of flight. Overall, the findings demonstrate the value of combining physiological signals with subjective reports to identify patterns of stress and fatigue during real-world flight training and highlight the potential of data-driven approaches for monitoring pilot well-being.

This paper presents a machine learning framework for data-driven inverse design of V-beam thermal sensors. The goal is to determine the optimal sensor geometry: beam inclination angle, beam length and beam width that achieves a target displacement under a given temperature. The design should also provide the geometry with minimum structure volume and minimum mechanical stress the sensor must support. This problem is ill-posed as for a given displacement there are multiple possible geometric configurations, causing direct regression methods to fail. We document a series of five exploratory trials that progressively revealed the nature of the problem culminating in a two-phase solution: a neural network forward model trained to map geometry and material constants to sensor responses, a gradient-descent inverse optimization over the frozen forward model, minimizing stress and volume simultaneously. The proposed pipeline utilizes a 3000-sample dataset and achieves a MAPE of 4.76% for predicting the displacement, more than 70% of predictions having MAPE of under 5%.

Extended Reality (XR) wearables require always-on perception within tight power envelopes of a few watts and motion-to-photon latency budgets below 20 ms, leaving only a few milliseconds for neural-network inference. Bit-serial computing is attractive for such energy-efficient neural network acceleration, but many existing architectures still process all bits even when ReLU sets the final output to zero. This paper presents BitFair, a software-hardware co-designed bit-serial CNN accelerator with learnable bit-level early termination and adaptive bit ordering, working under the ultra-low-power and strict latency requirements of XR applications. BitFair exploits dynamic bit-level sparsity by learning per-layer thresholds that trigger early termination when partial sums reliably predict that the final ReLU output will be zero. Furthermore, it searches for layer-wise bit orders that prioritize informative bits, maximizing early termination without sacrificing accuracy. A GlobalFoundries 12-nm FinFET implementation with a core area of 0.34 mm^2, 104 KB on-chip memory, and voltage scaling from 0.55 to 0.70 V achieves sub-millisecond latency, up to 117.0 BTOPS/W, and 0.07 pJ/SOP. On IBM DVS128 Gesture and N-MNIST, BitFair achieves 96.5% and 97.7% accuracy, respectively, while improving effective energy efficiency by 4.0-22.1x and accuracy by up to 9.2% over prior fabricated XR vision accelerators.

Physical sensing and actuation noise floors should inform how much belief resolution a decision-making system can reliably use. We introduce Finite Reliability Representations (FRR), a framework for covering belief spaces by reliability cells: regions within which the optimal action-value function Q*(b,u) varies by at most a tolerance epsilon, uniformly over actions. The framework is formulated on beliefs rather than states and uses a cover rather than an equivalence quotient, because approximate decision-closeness is not transitive in general. A central technical point is that noisy Bayesian updates should not be treated as globally contractive on arbitrary beliefs. We therefore separate three objects: the fixed-observation filter map, the predictive observation law, and the controlled belief-transition kernel. For nonlinear continuous-state systems, FRR is obtained under a reachable-set Lipschitz modulus for the belief-transition kernel. For finite-state POMDPs, the same construction becomes exact on the belief simplex: prediction is linear, Bayesian correction is a normalized positive linear map, sensor noise enters through observation-distribution distinguishability, and actuation uncertainty enters through an action-execution channel. Under the corresponding action-value Lipschitz condition, an FRR cover supports a cell-constant policy whose suboptimality is bounded by 2 epsilon/(1 - gamma). We also introduce reliability entropy, the logarithm of the minimal number of reliability cells, as a measure of certified decision-relevant belief complexity. The framework distinguishes representation sufficiency from fundamental performance floors imposed by sensing, process, and actuation noise. It applies to finite POMDPs, linear-Gaussian filters, locally linearized nonlinear filters, and particle-filter implementations through analytic or empirical certification of reliability cells.

We establish finite-sample closed-loop stability guarantees for Model Predictive Path Integral (MPPI) control applied to discrete-time Linear Time-Invariant (LTI) systems with additive Gaussian process disturbances. The key observation is that, for unconstrained LTI/quadratic systems with the DARE terminal cost, the exact finite-horizon MPC law has the same first control action as the infinite-horizon LQR law for every planning horizon. Thus, finite-sample MPPI can be analyzed as a stochastic perturbation of LQR. First, we show that the MPPI control law approximates the LQR feedback with high probability. The approximation error decomposes into a Monte Carlo term that decreases with the sample count and an infinite-sample temperature bias that persists at finite temperature but vanishes as the temperature is reduced. The resulting constants are written in terms of the horizon-dependent stacked cost matrices, making explicit that the finite-sample certificate is parametrized by the selected planning horizon. Second, we use a Lyapunov perturbation argument to prove practical exponential stability in expectation. On sample paths that remain in a compact Lyapunov sublevel set over a finite operating horizon, the expected state norm decays exponentially up to three residual floors: a process-noise floor, an MPPI approximation floor, and a confidence floor from the per-step sampling failure probability. The sufficient sample threshold is explicit and computable from the DARE solution, LQR stability margin, MPPI sampling parameters, temperature, and planning horizon. In the joint limit of infinite samples and vanishing temperature bias, the result recovers the stochastic LQR stability bound.

Advanced neural technologies in speech synthesis and voice conversion (VC) have introduced severe risks to personal privacy, necessitating robust Speaker Anonymization Systems (SAS). Existing SAS approaches modify voice characteristics in the hand-crafted feature space or speaker embedding space, often struggling to provide sufficient identity variance across generated voices. In this paper, we propose NouveauVoice, a novel pseudo-speaker generation framework based on a Hierarchical Deep Variational Autoencoder (NVAE). Integrated as a standalone plug-in module on top of state-of-the-art architectures (FACodec and CosyVoice2), our approach leverages tractable sampling and the Evidence Lower Bound (ELBO) objective to synthesize highly expressive pseudo-speaker embeddings with significantly enhanced speaker diversity. Evaluating our framework under a protocol similar to the VoicePrivacy Challenge alongside Maximum Mean Discrepancy (MMD) analysis, we demonstrate that NouveauVoice achieves strong identity concealment, yielding an Equal Error Rate (EER) exceeding 38% against an automatic speaker verification attacker model. Our system shows a reasonable trade-off between strict anonymity, rich pseudo-speaker diversity, and downstream speech utility, such as intelligibility and emotional expressiveness.

Diabetic retinopathy (DR) is a leading cause of vision impairment worldwide, highlighting the need for accurate and accessible screening tools. Optical Coherence Tomography (OCT) provides high-resolution structural information of the retina, whereas OCT angiography (OCTA) offers complementary vascular information that is highly relevant for DR diagnosis. In this study, we propose a cross-modal fusion of OCT B-scans with single-channel en face OCTA using a bidirectional cross-modal attention network for automated DR classification. Two independent datasets, OCT500 and UIC, comprising 730 subjects in total, were utilized to evaluate performance under within-dataset, combined-dataset, and cross-dataset generalization settings. A ConvNeXt V2 model trained solely on OCT images served as the unimodal baseline. In addition to ground-truth (GT) OCTA, we explored the use of translated (TR) OCTA generated from OCT scans, eliminating the requirement for dedicated OCTA hardware. Experimental results demonstrate that cross-modal fusion consistently outperforms unimodal OCT classification across all evaluation scenarios. Fusion with GT OCTA improved classification accuracy and discriminative performance, while TR OCTA achieved comparable or superior results in most settings. Furthermore, TR OCTA improved sensitivity and cross-dataset generalization, indicating enhanced robustness to domain shifts. These findings demonstrate that attention-based OCT-OCTA en face fusion provides clinically meaningful improvements for DR detection and suggest that computationally generated OCTA can serve as a practical, low-cost alternative to hardware-acquired OCTA, enabling broader deployment of high-performance retinal screening systems in resource-limited clinical environments.

We present an L1-optimal control problem class with linear nonnegative costs subject to multiplicative Itô diffusion processes with elementwise linear input constraints. Forward invariance of the positive orthant is established for the considered stochastic dynamics, and a simulation method consistent with this invariance property is proposed. Both finite-horizon and discounted infinite-horizon stochastic L1-optimal control problems are considered. These problems admit explicit solutions characterized by a vector-valued ordinary differential equation in the finite-horizon case and by an algebraic equation in the infinite-horizon case. Notably, the optimal value function and feedback policy coincide with those of the corresponding deterministic problem, demonstrating robustness to multiplicative stochastic uncertainty. A portfolio example illustrates our results.

Model predictive control (MPC) of building HVAC systems needs thermal models that answer a causal question: what indoor temperature, energy use, and comfort will result if a control action is applied? Time-series foundation models (TSFMs) can forecast passive building trajectories with strong zero-shot skill, but high factual accuracy does not imply valid response to control interventions. We show that an observational grey-box model with the best passive accuracy predicts cooling effects with the wrong sign, and that adding control and weather covariates to a TSFM does not fix intervention response. We introduce ThermoForce, a control-ready interventional thermal world model that keeps a TSFM frozen as a passive free-response prior and learns a compact, physics-structured forced-response operator for the causal effect of HVAC actuation. The operator is monotone in the control input by construction, is identified from one to three days of control excitation, and composes with the free response into a counterfactual-capable world model. Across paired EnergyPlus heating and cooling interventions, ThermoForce attains the lowest intervention-effect error and correct effect sign where covariate-TSFM, observational grey-box, and distillation baselines fail. Embedded in MPC on the BOPTEST benchmark, it reduces thermal discomfort by 33–84% relative to the native controller across three two-week windows while simultaneously reducing energy, using a frozen backbone, 195 trainable parameters, and CPU-only computation. ThermoForce reframes foundation models for building control: passive prediction and forced intervention response must be structurally separated for a model to be control-ready.

Whole-body fluorodeoxyglucose positron emission tomography combined with computed tomography is widely used in cancer care, but manual lesion delineation is slow, subjective, and difficult to scale. We present GLOW-FDG, an open-source artificial intelligence model for whole-body cancer lesion segmentation in fluorodeoxyglucose positron emission tomography and computed tomography. The model was trained on 1,563 scans spanning multiple cancer types and evaluated on 185 external scans from independent institutions. Across breast cancer, nonmetastatic and oligometastatic lung cancer, head and neck cancer, and metastatic melanoma, GLOW-FDG consistently outperformed publicly available benchmark models in lesion detection, while reducing false positives and maintaining strong segmentation accuracy. Quantification of total tumor burden and total lesion glycolysis was robust across cohorts, and performance approached the variability observed between expert radiation oncologists. These results support GLOW-FDG as a generalizable tool for automated cancer segmentation and quantitative imaging biomarker extraction in whole-body imaging.