Papers

With the advent of Large Language Models, single-task and token-based multi-task models have evolved into instruction-based systems that infer task and target language implicitly from natural language prompts. This trend is reflected in IWSLT's Instruction Following Track, which this year introduced new tasks including an unknown surprise task, posing a genuine challenge against overfitting to known tasks. We present KIT's submission to the Long and Short Instruction Following tracks in the unconstrained setting. Our approach combines a general data augmentation pipeline that converts short-form corpora into long-form training data through segment concatenation, LLM-based label generation, and cross-lingual translation, yielding over 1M instances across six tasks and four languages. We further show that likelihood-based re-ranking, while highly effective for ASR, systematically degrades semantic tasks by spuriously selecting candidates generated from segmented audio processing rather than holistic long-form inference, a failure mode resolved by combining likelihood with Minimum Bayes Risk decoding.

In this paper, we present a GPU-accelerated framework for nonlinear model predictive control (NMPC) based on direct transcription and second-order interior-point methods. Many real-world systems exhibit nonlinear dynamics that cannot be accurately captured by linear models, motivating the use of NMPC. However, NMPC requires the repeated real-time solution of optimal control problems (OCP), which become computationally demanding large-scale nonlinear programs (NLPs) after transcription. Although GPU acceleration has emerged as a promising approach for nonlinear optimization, existing GPU-based NMPC workflows reconstruct structurally identical OCPs at each solve. This introduces substantial overhead even though successive solves differ only through updated system measurements or reference trajectories. To address this limitation, we introduce a parametric interior-point formulation that exploits the fixed structure of transcribed OCPs, enabling reuse of structure-dependent computations (e.g., symbolic factorization in sparse Cholesky) across re-solves. We evaluate the proposed framework on distillation column and 2D heated plate benchmarks against state-of-the-art CPU and GPU configurations. The results show that the framework achieves over an order-of-magnitude speedup in total NMPC run times. These improvements are primarily driven by reduced per-iteration solve times, with GPU execution achieving up to a 94% reduction compared to the baseline. Overall, the results demonstrate the effectiveness of exploiting repeated problem structure in GPU-accelerated NMPC and highlight the potential of the proposed framework to expand the envelope of real-time NMPC applications.

Affine frequency division multiplexing (AFDM) is a promising waveform for integrated sensing and communication (ISAC) systems owing to its superior performance in time–frequency doubly dispersive channels. However, AFDM still faces a pair of challenges: high PAPR and random data symbols produce imperfect autocorrelation sidelobes. To address these challenges, this paper proposes a real-time data-driven framework that optimizes the pre-chirp parameter $c_2$ to enhance the AFDM-ISAC performance. Specifically, a side-information-free optimization problem is formulated to reduce PAPR and the weighted integrated sidelobe levels of both aperiodic and periodic autocorrelation functions, with complexity comparable to that of the conventional AFDM receiver. Furthermore, an efficient non-monotone line-search spectral projected-gradient algorithm is developed by exploiting closed-form gradients. Simulation results demonstrate that the proposed method achieves a superior sensing vs. communications trade-off and is capable of striking a promoted bit error rate performance in the presence of severe power amplifier nonlinearity.

Automatic speech recognition systems commonly rely on reference transcriptions for evaluation, while reference-free approaches often depend on internal confidence estimation or auxiliary language models. We propose READ (Reference-free Hypothesis Evaluation with Acoustic Discrepancy), a novel metric that evaluates ASR hypotheses directly from the speech signal. READ emphasizes the acoustic grounding of hypotheses. It uses a pretrained auto-regressive TTS model to compute the conditional likelihood of speech tokens given a text hypothesis, to measure fine-grained acoustic discrepancy between speech and text. Without additional training, READ can be applied for hypothesis refinement. Experiments show that READ correlates with specific recognition errors and improves ASR outputs, achieving up to 20% relative error rate reduction, with particularly strong gains under noisy conditions.

In two seminal articles published in 1964, Brayton and Moser introduced the concept of a mixed potential as a fundamental theoretic tool to describe and analyze a class RLC of nonlinear circuits containing resistors, capacitors and inductors. In this paper, it is shown for the first time that a mixed potential can be introduced for a class RLCM of RLC circuits containing also memristors. This is possible provided a memristor circuit is analyzed not in the traditional voltage-current domain but rather in the flux-charge domain. The flux-charge analysis method (FCAM) plays a crucial role in the extension, in particular, a key step is an equivalence principle established via FCAM between an RLCM circuit in the flux-charge domain and a nonlinear RLC circuit in the voltage-current domain. Several examples are discussed where the mixed potential is explicitly found. These include basic circuits with memristors, such as Chua's circuit with a memristor and also large-scale memristor arrays with a neural architecture. This paper is mainly devoted to the introduction of a mixed potential for memristor circuits and the study of its main theoretic properties, as the possibility to write the circuit state equations in the flux-charge domain in an effective and compact form via the mixed potential. In a companion paper [1], the mixed potential is used to obtain in a systematic way Lyapunov-like results on convergence of RLCM circuits. Those results will extend existing results on convergence that do not cover the important case where there is the simultaneous presence of capacitors and inductors in a memristor circuit.

This paper investigates joint three-dimensional (3D) trajectory planning and resource allocation for a high-altitude platform (HAPs)-unmanned aerial vehicle (UAV) bistatic integrated synthetic aperture radar (SAR) and communication (ISARAC) system in low-altitude networks. In the proposed architecture, the HAPs provides persistent wide-area connectivity by transmitting ISARAC waveforms for ground-user communications, while a low-altitude UAV exploits its proximity and mobility to passively collect ground-target echoes for high-resolution SAR imaging. We formulate a sum-rate maximization problem for ground users subject to stringent SAR imaging signal-to-noise ratio (SNR) and resolution requirements, a total energy budget for ISARAC transmission, and UAV dynamic constraints. The resulting problem is inherently nonconvex. To tackle it, an alternating optimization (AO) framework is developed, where the power-allocation subproblem with fixed UAV states admits a closed-form water-filling solution, while the UAV trajectory optimization with fixed transmit powers is handled via successive convex approximation (SCA) and difference-of-convex (DC) programming. Simulation results verify the effectiveness of the proposed approach and demonstrate its capability to jointly support persistent communication coverage and high-resolution sensing in low-altitude network scenarios.

While neural video coding (NVC) has achieved remarkable rate-distortion performance, real-time decoding on edge devices has become an important demand but remains limited by high complexity. Knowledge distillation (KD) is widely used for model acceleration, yet its application to NVC faces critical challenges. Specifically, the heterogeneity of NVC sub-modules renders uniform architectural reduction suboptimal, necessitating a per-module design for better rate-distortion-speed trade-off. However, searching for diverse architectures via existing neural architecture search (NAS) algorithms is unaffordable due to the expensive training cost of neural video codecs. Moreover, after the lightweight architecture is determined, existing distillation methods overlook the feature-energy sparsity induced by the rate-constraint, which is essential for maintaining compression performance. To address these issues, we propose a two-stage distillation framework KD-NVC. In the first stage, we introduce an acceleration-efficiency-based neural architecture search (AE-NAS) algorithm. It explores the module-wise Pareto frontier to adaptively allocate the acceleration budget across heterogeneous modules. Also, it introduces the acceleration-efficiency metric to determine the final student architecture without practically training all architecture-level candidates. In the second stage, we design an energy-aware feature distillation (EFD) loss that aligns the spatially-aggregated feature-energy signatures between the teacher and student codecs, transferring the rate-induced sparsity patterns for better compression efficiency. Experimental results demonstrate that the proposed framework consistently outperforms existing codec-oriented distillation methods, and achieves 69 FPS decoding at 1080p on RTX 5060 while maintaining comparable RD performance to VTM-LDB.

The efficiency of reactive near-field wireless power transfer (WPT) systems degrades rapidly with increasing separation distance and is highly sensitive to misalignment between transmitting and receiving coils. These limitations restrict the mobility of powered devices and confine many near-field WPT applications to static scenarios. To address these challenges, a multiple-input multiple-output (MIMO) WPT configuration is investigated due to its capability to shape the magnetic field distribution between the transmitter and receiver. Maximum power transfer efficiency can be achieved by appropriately setting the amplitude and phase of each transmitting coil; however, determining these optimal settings requires accurate knowledge of the system's S-parameters. This paper presents the use of the Nelder-Mead iterative optimization algorithm to estimate the input amplitude and phase settings that maximize transfer efficiency in a near-field WPT system. The implementation comprises a four-element transmitter and a two-element receiver. Based on measured S-parameters, the proposed approach significantly improves WPT efficiency under both aligned and misaligned conditions.

With the development of wireless sensor networks, distributed networks have received widespread attention. According to the different ways of connecting the nodes in the distributed network can be divided into different structures, of which the diffusion type structure is the most commonly used one due to its simple, stable and reliable. In order to improve the robustness of the diffusion subband algorithm in distributed networks, the median absolute deviation (MAD) theorem is applied to the error boundary selection, and this paper proposes a diffusion subband algorithm with a robust boundary. Through simulations, it is verified that the proposed algorithm can effectively reduce the update step size in the face of outlier interference, so that the algorithm has a good convergence performance and also has good robustness to impulsive noise.

This paper investigates whether component-level studies can capture additional interactions through Small Signal Analysis (SSA) when the network connected to the Voltage Source Converter (VSC), typically modeled as a simple Thevenin Equivalent, is a more complex IBR-based network. The research investigates cases ranging from basic analytical to an IEEE 9-Bus EMT model, with and without Inverter-Based Resources (IBRs), synthesized as State-Space elements. The study identified that spurious poles at 50Hz related to dq-frame conversion can hinder the accuracy of participation factor analysis. A potential approach involves a two-step process: first, applying Henkel reduction to remove most spurious poles, followed by manual elimination of any remaining ones.

Microwave linear analog computers (MiLACs) have recently emerged as a hardware-efficient solution for implementing multi-antenna communication systems. Unlike existing MiLAC designs based on the ideal assumption of perfect impedance matching (PIM) with reflection-free transmission, this paper investigates MiLAC-aided precoding optimization under a general impedance matching (GIM) model, which enables more flexible precoder design at the cost of a potential reduction in radiated power. Specifically, we consider a downlink multiuser multiple-input single-output (MISO) communication system and aim to maximize the system sum rate by optimizing the MiLAC-enabled transmit precoding subject to physical circuit constraints. The formulated problem is challenging to solve due to the intricate coupling between the precoding and impedance parameters. To address this challenge, we first develop a singular value decomposition (SVD)-based parametric search framework for small or medium size systems. This framework exploits the feasible precoder structure and explicitly captures the tradeoff between power radiation efficiency and precoder design flexibility. We then propose a unified algorithm for solving the optimization problem based on the projected weighted minimum mean-square error (WMMSE) principle for arbitrary size systems with GIM- or PIM-based MiLAC precoding. Simulation results demonstrate that the GIM-based MiLAC design consistently outperforms its PIM counterpart as a special realization, especially in interference-limited scenarios, by allowing a moderate reduction in radiated power in exchange for additional precoder design flexibility and more effective interference mitigation. It is also shown that GIM-based MiLAC design achieves performance close to that of the baseline fully digital precoding system.

State-evolution analyses of approximate-message-passing and expectation-propagation-type algorithms rely on an effective-channel principle: after a suitable Onsager, orthogonal, or extrinsic correction, the nonlinear module receives a fresh scalar Gaussian observation. This paper studies this principle for diagonal expectation propagation under variance-profile Gaussian sensing matrices. The model preserves Gaussian conditioning, but removes the isotropy that supports the usual scalar decoupling arguments. We prove a finite-time large-system description in which the linear EP module remains Gaussian at the coordinate level, but is generally not a fresh scalar channel. Instead, the residuals form a coordinate-dependent Gaussian process whose covariance is shaped by the variance profile and by the finite linear history of the algorithm. The standard diagonal EP cavity cancels the instantaneous response of the incoming message, but may leave a component predictable from past residuals. We characterize this process through a conditioned matrix-Dyson-equation deterministic equivalent and a Schur-complement representation of the linear module. A Gaussian-regression decomposition then separates the predictable memory from the orthogonal innovation and yields an oracle state-evolution-level correction. Thus, under variance-profile measurements, the limiting object for diagonal EP is a Gaussian-process dynamics with profile-dependent memory rather than the conventional fresh-noise scalar state evolution.

In this paper, we present a class of bearing based distributive nonlinear guidance laws for the cooperative circumnavigation of a stationary target by a heterogeneous team of asymmetric Dubins vehicles. In such a vehicle, the maximal left and right turn capabilities are non uniform. In the given framework, the location of the target is known only to a small subset of the vehicles, called the leaders. The uninformed vehicles, called the followers, use information from their out neighbours in the communication graph, constructed using the nearest neighbour rule. A class of guidance laws is formulated that relies solely on the heading angle and line of sight angles of a designated out neighbour of the vehicle in the graph. Using Zubov theorem, we prove that the proposed guidance laws achieve global asymptotic stability under angular speed only control and ensure the convergence of the trajectories of all the Dubins vehicles to a common centre. The proposed results are validated through numerical simulations.

Speech Large Language Models (SLLMs) underperform their text counterparts on complex reasoning. We reveal that this gap is not a uniform cognitive deficit. Evaluating two architecturally diverse SLLMs, we show speech-to-text (S2T) matches or exceeds text-to-text (T2T) on spatial, syntactic, and factual tasks. Yet on logical tasks requiring entity tracking, S2T accuracy collapses to chance. We diagnose this as an entity binding failure: continuous speech features blur precise entity-property associations during implicit reasoning. To validate this diagnosis, we introduce Entity-Aware Chain-of-Thought (EA-CoT), a lightweight inference-time intervention forcing SLLMs to enumerate entities and bind them to claims before reasoning. EA-CoT bridges the gap, even when spoken names are misrecognized, yielding up to a 24.4 percentage-point accuracy gain. Ablations confirm the gains stem from explicit semantic binding, reframing the gap as an elicitation failure rather than a missing capability.

This paper addresses the Self-Optimizing Control (SOC) problem in industrial continuous processes and proposes a Reinforcement-Learning (RL)-based SOC approach to improve dynamic performance under high-frequency disturbances. In the proposed framework, the SOC controlled variable structure is embedded in the Actor network, and reward functions are designed based on economic indicators. Through interaction with the environment, the RL agent optimizes controlled variables while implicitly considering implementability and steady-state uniqueness. Online fine-tuning is further introduced to alleviate model mismatch. Experiments on a continuous stirred-tank reactor with disturbances compare the proposed RL-based SOC method with the Objective-Guided Controlled Variable Learning Approach based on steady-state data. The results show that the RL method achieves improved dynamic performance under real-time disturbances, generates smooth controlled variable outputs without explicit regularization, reduces hyperparameter-tuning complexity, and enhances adaptability through online adjustment. Overall, the proposed RL-based SOC approach provides an effective solution for nonlinear process control and offers a promising reference for future studies involving multiple disturbances, multiple operating conditions, and model-free scenarios.

Existing datasets cannot support large-scale learning in multi-agent, multi-sensor, or multi-domain autonomy, where diversity and coordination are essential. We present a modular dataset generation pipeline that creates terabyte-scale, ground-truth-labeled data for ground, aerial, and infrastructure-based systems using the AVstack framework and CARLA simulator. Supporting single- and multi-agent configurations with flexible sensor suites, the pipeline enables controllable experimentation across challenging conditions. Representative perception and fusion studies show how generated data can support application-specific training and collaborative autonomy.

MRI provides excellent soft-tissue contrast without ionizing radiation, but long acquisition times increase patient discomfort while also raising exam costs and limiting scanner throughput. A common approach to reduce scan time is to acquire fewer measurements, which yields an ill-posed linear inverse problem; recovering diagnostic-quality images therefore requires incorporating prior knowledge beyond the measured data. In follow-up exams, the most recent prior scan of a patient can provide a highly informative subject-specific context, but practical use is complicated by temporal changes (including pathology progression), misalignment between scans, and protocol drift across acquisitions. In this work, we introduce L-TGVN, a Longitudinal Trust-Guided Variational Network that leverages prior scans as side information to reconstruct the current scan from heavily undersampled measurements. Crucially, L-TGVN constrains the influence of prior scans to be consistent with the acquired measurements. Unlike many existing longitudinal reconstruction methods, it does not require explicit pre-registration between prior and current scans. It further accommodates differences in acquisition protocols across visits (e.g., changes in sequence parameters). We evaluate L-TGVN against matched-capacity baselines, including prior-guided methods and methods that do not use longitudinal priors, and observe consistent improvements in standard quantitative metrics together with better preservation of fine structures at challenging accelerations. Source code is available at github.com/sodicksonlab/L-TGVN.

Neural audio codecs are a key component of speech processing pipelines, compressing audio into discrete tokens for downstream modeling. However, existing codecs struggle to balance reconstruction quality with token efficiency, often encoding perceptually irrelevant information such as background noise and recording artifacts at the expense of linguistically and acoustically meaningful content. We reframe audio tokenization as a selective information bottleneck problem and propose CleanCodec, a denoising audio codec which learns to encode only perceptually important features and discard imperceptible information. At just 12.5 tokens per second, CleanCodec achieves state-of-the-art tokenization efficiency, substantially outperforming existing codecs in speaker similarity and speech intelligibility. Evaluations on downstream text-to-speech and voice conversion tasks further demonstrate improved performance and up to 17x faster inference, highlighting significant efficiency gains.

We propose ISS-BKNO, a unified framework that integrates Koopman operator identification, Neural ordinary differential equations (ODEs), fiber bundle geometry, and input-to-state stability (ISS) certification. Unlike prior approaches that address stability, extrinsic inputs, or environmental constraints in isolation, the proposed framework simultaneously learns controlled nonlinear dynamics while guaranteeing global convergence and a computable ISS gain. The architecture introduces a three-stage lifting pipeline: a bundle-aware encoder that separates environment-specific fibers, an environment-conditioned Koopman backbone whose matrix spectrum is constrained to lie in the left half-plane, and a residual neural ODE correction whose Jacobian satisfies a quadratic sector bound. Lyapunov-based ISS regularization turns the stability requirement into a differentiable penalty that is jointly optimized with the prediction objective. Theoretical results establish fiber invariance, ISS with an explicit gain formula, and an approximation error bound that scales with the EDMD residual. Experiments on a pendulum, cart-pole, a unicycle-based navigation task, and a Franka Emika manipulator demonstrate substantially improved prediction accuracy and robustness under matched disturbances compared with existing Neural ODE and Koopman baselines.

Real-time multi-camera 3D reconstruction remains challenging due to the strong dependency among extrinsic calibration, multi-view fusion, and global optimization, which limits reconstruction stability and scalability. This paper presents SAGE-Flow, a decoupled framework consisting of geometry-aligned multi-view calibration (GMAC) and stateless adaptive geometric representation (SAGE). GMAC estimates camera extrinsics from geometric constraints without calibration targets, dense images, or bundle adjustment. SAGE constructs a compact geometric representation by selecting reliable multi-view observations under a bounded geometric budget, achieving linear time and memory complexity. Experiments show SAGE-Flow achieves precise camera calibration, low 3D reconstruction cost, good scalability, and can generate high-quality point clouds under limited throughput.

In this paper, we propose a reconstruction framework that leverages the Wavelet Scattering Transform (WST) as a multi-scale feature extractor to impose statistical priors under sparse observation conditions. The reconstruction problem is formulated as an optimization task and solved using a neural field, with the WST incorporated into the training loss function. As a proof of concept, we validate the proposed method on HRTF upsampling. A masking strategy is applied to the WST coefficients, resulting in a two-phase procedure. The first phase learns a binary mask from a small multi-subject dataset, while the second phase applies the learned mask to the WST coefficients of an individual HRTF to preserve informative statistical structures during reconstruction. Validation against baseline methods, which also serve as an ablation study of the different components of the framework, demonstrates the effectiveness of the proposed approach.

Age of Information (AoI) has become a central metric for the design of wireless update systems, especially in applications where fresh measurements support tracking, estimation, and control. Despite its popularity, the use of mean AoI or peak AoI as a surrogate for closed-loop performance is often motivated by intuition rather than by a control-theoretic derivation. This paper examines whether minimizing the mean AoI is in fact optimal for networked control systems. For scalar linear time-invariant systems with delayed intermittent updates, we show that, under state-independent scheduling policies, the infinite-horizon LQR tracking problem reduces to an optimization over the distribution of inter-scheduling intervals. The resulting objective depends on higher-order statistical moments, and in unstable or correlated regimes on exponential moments, of the inter-scheduling process rather than only on its mean. Consequently, policies with identical mean AoI can induce substantially different tracking costs. We further extend the analysis to disturbances with exponentially decaying autocorrelation and derive equivalent cost formulations that expose the role of the full interval distribution. Finally, we evaluate the theory using real vehicle trajectories from the NGSIM US-101 dataset. The empirical results match the predicted performance trends, demonstrating that mean AoI alone is insufficient for control-oriented network design.

The image-source model (ISM) is a widely adopted method for efficiently simulating acoustic room impulse responses (RIRs) under specular reflection assumptions. Acoustic paths between source and receiver are traced to lattice points computed from successive reflections over bounding planes of the room. Rectangular rooms bound the total number of image-sources to be polynomial in the RIR's duration or distance $k$ equivalent, with degree equal the number of room dimensions $N$. Direct ISM simulations are therefore compute upper-bound by $O \left ( k^N \right )$, and consider only cases of $N \leq 3$ for tractability and real-world applications. This work proposes an alternative computational method that lowers the asymptotic compute bound to $O \left ( N k^2 \log k \right )$ for integer coordinates and room dimensions via reducing ISM lattice point counting to the classic Gauss circle problem (GCP). We extend the lattice counting model to frequency-dependent and reflection weighted image-sources in higher dimensions, relating solutions between successive dimensions via the convolution operator. Two constructions for realizing RIRs are presented, along with time-frequency controls, error and run-time analysis, and RIR statistics.

The framework of robust Markov decision processes (RMDPs) allows the design of reinforcement learning agents that satisfy performance guarantees under worst-case transition dynamics. Traditional RMDPs consider discrete-time dynamics and recently, sample-efficient policy gradient algorithms have been considered in this context. This paper investigates policy gradient algorithms within a continuous-time RMDP framework. Policy gradients and adversarial gradients are derived using pathwise and adjoint-based formulas for stochastic and ordinary differential equations. We propose double-loop optimisers to obtain linear convergence in the oracle-based setting and an $\tilde{\mathcal{O}}(\frac{1}{\epsilon^2})$ sample complexity in the sample-based setting in an analysis which also derives novel tools for the framework of undiscounted total cost MDPs. Additionally, we propose mean-field optimisers as distributional optimisers with an $\tilde{\mathcal{O}}(\frac{1}{K})$ oracle-based convergence rate and an $\tilde{\mathcal{O}}(\frac{N^2}{\epsilon})$ sample complexity under $N$-particle approximation. The effectiveness of continuous-time policy gradient algorithms is confirmed for both optimisers on continuous-time RMDPs with neural ordinary differential equation dynamics.

Fixed convolutional representations can induce structured geometry in time-frequency data even without task-specific or source-domain training. We audit a previously reported morphology-space analysis of two granite acoustic-emission experiments. The original cumulative angular-path endpoint is reproduced to within 2.1e-7 in the Archive-B/Archive-A ratio. However, the same directional contrast is obtained with ImageNet features and with three randomly initialized EfficientNet-B0 encoders: the preserved Aether representation gives a ratio of 1.411, ImageNet gives 1.361, and the random encoders give 1.480–1.503. We therefore withdraw the attribution of this contrast to interferometric pretraining. We also correct a reversal between archived OG1/OG3 file identifiers and the physical experiments. Phase randomization and temporal shuffling alter the preserved trajectory, but these responses are reinterpreted as pipeline-dependent diagnostics rather than evidence of a learned universal elastic morphology. The corrected result is narrower: multiple fixed convolutional maps expose a reproducible contrast between two acoustic-emission datasets. The signal statistics and architectural biases responsible for that contrast remain open questions.

The deployment of modern machine learning (ML) solutions on resource-constrained edge devices highlights implementation challenges. This is especially true for extreme edge applications that include safety-critical components, such as autonomous navigation tasks. This paper demonstrates an artificial neural network (ANN) design leveraging Metal-Oxide Resistive RAM (RRAM) -based Analogue Content Addressable Memory (ACAM) as an efficient hardware substrate for performing metric-based classification and online adaptation on the edge. The proposed design is based on a custom Template piXeL (TXL) cell used for building the ACAM module, where each TXL cell acts as a configurable receptive field neuron. These cells employ a Radial Basis activation function to calculate the distance of an input from the programmed receptive field. The TXL can be organised into dense arrays for calculating the distance of a high-dimensional input against all stored prototypes, effectively performing fast and energy efficient similarity search. This hardware engine enables on-the-fly learning, where the receptive field parameters can be tuned to track domain shift. Through simulation of the proposed TXL-RBF classifier we can achieve 89.1% accuracy on the MNIST dataset while consuming 185fJ per cell per operation when operating at 10MHz.

We present Safecloud, a distributed, encrypted, self-pricing storage and streaming network whose storage and routing nodes never see plaintext and never hold keys. Each file is split into chunks, encrypted on the owner's device, and distributed across Drops (browser tabs storing ciphertext in IndexedDB) and Jets (federated routing servers). Only the owner, or an authorised grantee, can decrypt. We make five contributions: (1) A one-root key hierarchy: every key derives deterministically from a single root via HKDF, and owner and range-scoped grantee derive identical chunk keys (derivation agreement); a subtree key derives its range and nothing else (delegation containment). (2) Convergent content addressing: identical content yields identical ciphertext and identifiers, enabling deduplication without plaintext exposure, with identifiers binding authenticated ciphertext so a keyless Drop verifies integrity (blind verifiability). (3) Three parallel trees over one navigation path (Merkle for integrity, key-derivation for confidentiality, access for authorisation), with sound Merkle-verified retrieval. (4) The key tree doubles as a streaming index: a player derives each segment key in O(1), seeking by derivation, while parallel tracks (video, audio, captions) are independent subtrees unlockable per-track and per-segment, a combination we believe no prior encrypted-storage network offers. (5) Jets and Drops earn Safebux verifiably, kept honest by a one-signature proof-of-storage challenge under chilling-effect Proof-of-Corruption, a zero-sum economy that is significantly cheaper than Filecoin's proof-of-replication sealing (which is slow and provides no confidentiality). We give the architecture, cryptographic construction, a threat model, and an open-source reference implementation, stating precisely what is implemented versus designed.

Accurate and scalable flood mapping remains challenging due to limited ground observations, heterogeneous terrain conditions, and the difficulty of enforcing hydrodynamic consistency within data-driven models. This work introduces a physics-guided deep learning framework that integrates multi-modal remote sensing (Sentinel-1 SAR, Sentinel-2 optical imagery, and DEM-derived terrain features) with constraints from the depth-averaged shallow water equations (SWE). The proposed hybrid architecture combines a UNet to capture fine-scale spatial details with a Fourier Neural Operator (FNO) to model basin-scale hydraulic interactions, while physics-informed residual losses ensure mass and momentum consistency. Evaluated across diverse floodplain settings, the hybrid model achieves an Intersection over Union of 0.82 and an F1 score of 0.90 for flood extent prediction, outperforming UNet-only and FNO-only baselines. Using hydrodynamic simulations as reference data, the model achieves an RMSE of 0.21 m for water depth and 0.15 m/s for flow velocity. Physics consistency is maintained, with low residuals and mass imbalance below 2.1%. Ablation studies confirm that removing physicsbased regularization significantly degrades performance, underscoring the value of physical constraints for stability and generalization. These results demonstrate that embedding hydrodynamic principles into deep learning yields more accurate, reliable, and physically coherent flood predictions, offering strong potential for operational monitoring and large-scale deployment.

Selecting a clustering algorithm and its hyperparameters without labels is a common difficulty in engineering machine learning pipelines that work with unsupervised analysis of sensor, image, or process data. Clustering validation indices (CVIs) provide internal scores for ranking candidate clusterings, but most popular CVIs are built from Euclidean compactness and separation terms and so tend to favour compact, convex partitions. Their performance is known to degrade on non convex, irregular, or variable density data, where kernel transformations or alternative distance measures are typically used at the cost of additional tuning and computation. This paper introduces the Central Description Length (CDL) clustering validation index. CDL uses the observed within cluster compactness, the estimated cluster centers, and the estimated cluster covariances to compute a probabilistic upper bound on the description length associated with the unobservable true cluster centers. The bound condenses intra cluster compactness and centroid displacement into a single computable quantity and is evaluated on the partition produced by any clustering algorithm. The implementation uses only observable quantities (the data, the partition, the estimated centers, and the estimated covariances) and does not use ground truth labels. On synthetic benchmarks with non convex and arbitrary shape clusters, CDL-CVI selected the reference number of clusters more often and reached higher Adjusted Rand Index (ARI) values than the conventional CVIs we tested, without an additional kernel preprocessing stage. On image benchmarks (MNIST, CIFAR-10, STL-10) clustered from frozen unsupervised embeddings, CDL-CVI returned cluster numbers close to the reference class counts across K-means, DBSCAN, and spectral clustering in the reported trials.

Prospective reconstruction is crucial in many clinical applications such as MRI-guided radiotherapy, which demands accurate image reconstruction and fast motion estimation from currently acquired measurements. However, prospective reconstruction remains challenging due to ultra-sparse sampling and stringent latency requirements. In this work, we propose PDMR, a Prospective Dynamic 3D MRI Reconstruction framework with latent-space motion tracking. Our core idea is to learn an efficient and generalizable latent manifold of motion fields offline, enabling rapid online adaptation for prospective reconstruction. Specifically, we parameterize the deformation vector fields (DVFs) on a low-dimensional manifold, effectively reducing the search space for fast online adaptation, and employ a tri-plane representation to achieve geometry-aware and memory-efficient encoding of 3D motion. Experiments on both XCAT digital phantoms and in-house abdominal MRI datasets demonstrate that PDMR achieves high-fidelity and temporally consistent reconstruction across multiple prospective scenarios (Immediate and After-2min), outperforming state-of-the-art retrospective and online methods. Our results suggest a promising pathway toward ultra-fast, motion-aware prospective MRI reconstruction in clinical practice.