Fusion

arXiv:2605.29733v1 Announce Type: new Abstract: Scaling data-driven energy forecasting to district level requires models that can be re-used across buildings with minimal target-domain data and honest uncertainty estimates. We present an uncertainty-aware transfer learning (TL) framework for cross-building energy forecasting based on the Temporal Fusion Transformer (TFT), evaluated on a newly released high-resolution real sub-meter dataset: an educational building at Aalborg University, Denmark (source) and the multi-typology NEST building at EMPA, Switzerland (target). We introduce the Transfer Robustness Index (TRI), an architecture-agnostic metric for quantifying generalization quality across domain gaps. A four-strategy layer-freezing ablation shows that Probe-Only fine-tuning, updating only 455 output-layer parameters out of 806K, achieves the best transfer quality (TRI = 3,097), outperforming full fine-tuning and suggesting that TFT encoders learn transferable temporal

arXiv:2605.29236v1 Announce Type: new Abstract: Alarm fatigue in intensive care units (ICUs) is a well documented patient safety crisis. Clinical monitors generate 350 or more alarms per patient per day, out of which 72-99% are clinically irrelevant. Staff desensitization to non-actionable alarms increases the risk of missed true emergencies. This paper presents SigmaMedStat, a machine learning system that evaluates the trustworthiness of physiological alarm signals before clinical action is taken. Four approaches were evaluated on the PhysioNet/Computing in Cardiology Challenge 2015 dataset of 498 four-channel ICU alarm recordings. Primary contribution is a temporal modeling framework that splits each 60 second recording into six consecutive 10-second chunks, and this in turn generates Continuous Wavelet Transform (CWT) scalograms per chunk, encodes each chunk with a shared EfficientNet-B0 encoder, and passes the resulting feature sequence to a two-layer Long Short-Term Memory (LSTM)

arXiv:2605.29120v1 Announce Type: new Abstract: Globally, 340 million people have blindness or moderate to severe visual impairment (BVI)$^1$ which limits independent outdoor navigation$^2$ and negatively affects their health and quality of life$^{3,4}$. We surveyed 112 people with BVI and found that an ideal outdoor navigation aid must be able to perform turn-by-turn directions, path guidance, and obstacle detection and avoidance. Existing navigation tools such as white canes, guide dogs, and electronic travel aids often lack one or more of these criteria and may be expensive or inaccessible$^{5,6}$. Here we introduce Mobilio, a smartphone application that incorporates machine learning, sensor fusion algorithms, and personalized audio feedback to meet all of the outdoor navigation criteria. The reliability of the smartphone sensors and models used for navigation were assessed with engineering tests in representative navigation scenarios. We performed a series of experiments where

arXiv:2605.28296v1 Announce Type: cross Abstract: In modern nuclear physics experiments, identifying events of interest is challenging for nuclear reaction studies with the active target Time Projection Chamber (TPC). In this work, machine learning techniques are employed to analyze the complex data of the 12C + 12C fusion reaction from a TPC named MATE (multi-purpose active-target time projection chamber for nuclear experiments). Specifically, we successfully applied Residual Neural Network (ResNet-50, ResNet-34 and ResNet-18) and Visual Geometry Group (VGG-19) to classify elastic scattering and fusion reaction events from the 12C + 12C reaction. The classification results of the four models are nearly identical, with accuracies of approximately 97% for the simulated data and 90% for the experimental data. Moreover, these approaches successfully identify some events that are misclassified by traditional methods. These models are also applied to classify events from different fusion

arXiv:2605.27549v1 Announce Type: new Abstract: This work presents the compact experimental negative triangularity reactor (CENTAUR), a low overnight cost, high-field tokamak, breakeven reactor design, achieving a predicted total fusion power of 40MW and scientific energy gain of 1.3. Ballooning stability calculations confirm that the device's pedestal is within the first stability regime, which is consistent with the expected ELM-free operation associated with negative triangularity (NT) plasmas. The geometry of the NT divertor allows for high fraction of radiated power (13.5$\%$) between the separatrix and plasma facing components. Heat transport modeling based on simulations of the edge region show heat loads into plasma facing components well below material limits. The magnet system employs rare-earth barium copper oxide (REBCO) high-temperature superconductors in 18 toroidal field coils, an hourglass-shaped central solenoid, and six poloidal field coils to support high-field

arXiv:2605.27417v1 Announce Type: cross Abstract: With the advent of sixth-generation (6G) mobile communication technology, vehicle-to-everything (V2X) communication faces unprecedented challenges in communication efficiency, system generalization capabilities, and model collaboration. Conventional machine learning struggles with high-dimensional state spaces, slow convergence, and poor generalization under heterogeneous V2X nodes, rapidly varying channels, and multimodal sensing data in V2X systems. To address these issues, we propose a quantum-enhanced framework for V2X communication and model aggregation that targets efficient, robust, and intelligent transportation in 6G, which includes four modules: the channel-adaptive semantic communication module, the multimodal fusion module, the model transfer module, and the federated aggregation module. Specifically, the channel-adaptive semantic communication module leverages quantum convolutional neural networks (CNN) and quantum

arXiv:2605.28315v1 Announce Type: new Abstract: General-purpose machine translation benchmarks such as FLORES-200 have reached a saturation regime on Chinese-English pairs, where modern large language models cluster within a narrow band of high scores. Across 22 systems, FLORES-200 zh-en GEMBA scores fall in a 7.87-point range with a standard deviation of 2.29, which compresses the separation between systems on knowledge-intensive domains such as finance, healthcare, law, and science and technology. We introduce HardMTBench, a difficulty-aware diagnostic benchmark for bidirectional Chinese-English domain translation. HardMTBench covers 12 domains and contains 10,000 hand-curated source sentences with reference translations, packaged as 20,000 directional test items. A three-stage construction pipeline builds a domain-balanced candidate pool of 84{,}566 pairs, applies an LLM-based multi-signal judge over knowledge density, translation difficulty, terminology load and reference

arXiv:2605.28296v1 Announce Type: new Abstract: In modern nuclear physics experiments, identifying events of interest is challenging for nuclear reaction studies with the active target Time Projection Chamber (TPC). In this work, machine learning techniques are employed to analyze the complex data of the 12C + 12C fusion reaction from a TPC named MATE (multi-purpose active-target time projection chamber for nuclear experiments). Specifically, we successfully applied Residual Neural Network (ResNet-50, ResNet-34 and ResNet-18) and Visual Geometry Group (VGG-19) to classify elastic scattering and fusion reaction events from the 12C + 12C reaction. The classification results of the four models are nearly identical, with accuracies of approximately 97% for the simulated data and 90% for the experimental data. Moreover, these approaches successfully identify some events that are misclassified by traditional methods. These models are also applied to classify events from different fusion

arXiv:2605.26432v1 Announce Type: cross Abstract: Muon-catalyzed nuclear fusion (\mucf) replaces atomic electrons with negative muons, compressing atomic orbitals by about two orders of magnitude and enabling deuterium--tritium (D--T) fusion under near-room-temperature conditions. This paper reviews the physical principles of \mucf{} and formulates its essential dynamics as a four-step cycle: muonic-atom formation, muon transfer, resonant \dtmu{} molecular formation, and D--T fusion with muon release and recycling. A kinetic model is used to quantify the number of catalysis cycles per muon and the corresponding energy gain. We focus on the central limitation of catalytic efficiency, namely the alpha-sticking effect, and discuss possible breakthrough routes including nuclear-spin and muon dual polarization, in-flight muon-catalyzed fusion, and heavy-ion-driven magneto-inertial fusion. Within the idealized assumptions of the present model, a four-dimensional synergistic scheme combining

arXiv:2605.27089v1 Announce Type: new Abstract: The electron cyclotron emission (ECE) diagnostics suite at ITER utilizes a front-end quasi-optical (QO) system whose design is fundamentally constrained by a field-stop concept. The field-stop defines the Gaussian beam variation throughout the optical system and within the plasma, thereby setting the ECE sampling volume and spatial resolution. An in-situ hot calibration source, optimized using Gaussian beam transmission criteria, provides independent and absolute electron temperature measurements. The QO system extends beyond the front-end to include the polarization splitter unit (PSU), transmission lines, and switchyard, forming an integrated optical path to the ECE instruments. Misalignment between the front-end and PSU reduces the effective field-stop size, degrading spatial resolution and measurement fidelity. The oblique ECE view, a key feature of the ITER design, enhances sensitivity to non-thermal electron populations and

arXiv:2605.26701v1 Announce Type: new Abstract: Ablation with mixed $2\omega$--$3\omega$ lasers is investigated as a possible drive strategy for balancing drive efficiency and ablative stabilization in direct-drive inertial confinement fusion. One-dimensional radiation-hydrodynamic simulations are performed for planar CH targets using the FLASH code [B. Fryxell et al, The Astrophysical Journal Supplement Series \textbf{131}, 273 (2000)]. The total target-incident laser intensity is varied from 100 to $1600~\mathrm{TW}/\mathrm{cm}^{2}$, and the $3\omega$ laser intensity fraction is scanned from 0 to 100\%. Thick-target simulations are used to determine quasi-steady ablation-pressure scalings, while thin-foil simulations are used to characterize the acceleration stage and to evaluate the linear ablative Rayleigh--Taylor instability (RTI) gain using a Takabe-type model. The simulations show that adding a $3\omega$ component to a $2\omega$-dominated drive increases the effective ablation

arXiv:2605.26496v1 Announce Type: new Abstract: The Mixture of Experts MoE architecture is highly promising for resource constrained on device deployments yet training these models from scratch incurs prohibitive costs Current methods attempt to alleviate this by upcycling dense models into MoEs however they often introduce parameter redundancy that degrades inference efficiency Alternatively standard layer pruning mitigates redundancy but inevitably compromises model accuracy To resolve this dilemma we propose Dense2MoE a novel framework that unifies pruning and upcycling through Layer Fusion UpCycling LF UC Guided by hardware Roofline theory Dense2MoE systematically overcomes the inference memory wall by pruning bandwidth heavy attention modules from redundant layers while repurposing their Multi Layer Perceptrons MLPs into MoE experts This structural innovation preserves the models core capabilities and strictly limits active parameters via selective token routing With a modest

When hydrogen gas interacts with uranium metal, the combination creates a chemically reactive powder and a runaway reaction that is difficult to stop. The result can impact the safety and lifespan of technology critical for fusion energy, hydrogen storage and nuclear fuels.

arXiv:2605.24201v1 Announce Type: cross Abstract: Medical image computing software is essential for identifying imaging biomarkers that can support diagnosis, prognosis, treatment planning, and clinical research. However, the lack of standardized, user-friendly, and reproducible software environments has limited the broader adoption of advanced medical image analysis workflows. We present Radiuma, a freely available modular platform designed to support reliable and reproducible medical image analysis across multiple modalities and file formats. Radiuma integrates image reading, visualization, registration, fusion, processing, segmentation, radiomics feature extraction, and machine learning modules for classification, regression, and clustering. Its modular design allows users to execute each component independently or connect modules through a visual workflow system, where the output of one step can be graphically passed to the next. This enables the creation of custom, executable,

arXiv:2605.25697v1 Announce Type: new Abstract: The importance of investigating magnetized plasmas/solids in extreme conditions has grown over the last decades, particularly in the field of high energy density physics (HEDP), such as laboratory astrophysics and inertial confinement fusion. However, up to now, the unique capabilities of an X-ray free-electron laser (XFEL), such as high brilliance and low divergence have never been exploited for this type of research. In this paper, we present the first platform developed at SACLA, Japan, that combines a high-power optical laser for generating matter under extreme conditions of pressure and temperature, an XFEL probe, and an external magnetic field. The high current is produced using a 2 kV, 4.8 kJ pulsed power system giving a maximum current of 10 kA which is synchronized with the optical laser and XFEL in a vacuum environment. It flows through a split-pair coil to generate a high magnetic field (10 T at 6 kA) which has 1 cm access

arXiv:2605.25913v1 Announce Type: cross Abstract: Recent demonstrations of non-Abelian braiding of graph vertices on noisy intermediate-scale quantum (NISQ) superconducting processor, and the experimental realization of topological order in general on various quantum hardware platforms necessitate an important question: when does a native (topological) fusion readout genuinely help for topological anyonic Hamiltonians implemented on NISQ hardware? We use the Fibonacci anyons chain as a concrete model for understanding the trade-off between measurement cost and compilation cost in that setting. The comparison is made against a simple grouped-Pauli baseline, and is scored by a covariance-aware mean-squared-error (MSE) of the full energy estimator. We based our benchmark on two different important classes of quantum circuits, namely Floquet time-evolved and variational quantum eigensolver quantum circuits, with the underlying Hamiltonian consisting of both braiding and fusion

arXiv:2605.25239v1 Announce Type: new Abstract: We present FusionCore, an open-source ROS 2 sensor fusion package that fuses IMU, wheel encoder odometry, GPS, and Visual SLAM pose into a single 100 Hz odometry stream using a 23-state Unscented Kalman Filter (UKF). The 23rd state is an online estimate of the wheel encoder's systematic yaw rate bias, identified through GPS heading cross-covariance and subtracted during GPS blackouts to reduce heading drift in coast mode. FusionCore also estimates gyroscope and accelerometer biases as explicit filter states, handles GPS natively in ECEF without a separate coordinate projection node, applies per-sensor Mahalanobis chi-squared outlier gating calibrated to measurement degrees of freedom, and adapts sensor noise covariance automatically from the innovation sequence. VSLAM pose fusion enables GPS-denied operation with any visual odometry or SLAM system, including automatic recovery from map reinitialization. We evaluate against

arXiv:2605.23121v1 Announce Type: new Abstract: Extended magnetohydrodynamic (MHD) simulations of tokamak plasmas regularly produce outputs in multi-dimensional, multiple-field formats; these code-specific formats make it difficult to do cross-code validation/coupling and analyze at a database scale. In this paper, a workflow that converts NIMROD code inputs and outputs to records compatible with version 4 of the ITER IMAS Data Dictionary is presented. The scope of the workflow includes preprocessing of NIMROD code inputs, conversion of hierarchical NIMROD code HDF5 dumps, COCOS-consistent treatment of the coordinate system and sign convention, and encoding finite-element poloidal meshes and toroidal Fourier components through IMAS General Grid Description. Furthermore, the workflow allows for provenance and integrity metadata to be included while providing optimal I/O operations for large array structures. An example conversion based on an NIMROD code simulation of edge harmonic

arXiv:2605.23015v1 Announce Type: new Abstract: This work presents a stepwise validation of the evolution of the radial electric field (Er) and heat transport during the pre L-H transition phase using full-f gyrokinetic simulations of the edge and scrape-off layer in the ASDEX Upgrade (AUG) tokamak, including X-point geometry. Several L-mode time slices up to the L-H transition from a dedicated hydrogen discharge, featuring stepwise increases in ECRH input power, are selected [N. Bonanomi \textit{et al.}, Phys. Plasmas 31, 072302 (2024)] and simulated with the \texttt{GENE-X} code. As the edge boundary conditions are progressively increased between the time slices, particle and heat fluxes rise, and the radial electric field Er well deepens. A detailed validation of the Er profiles and of the Er well depth shows excellent agreement with experimental measurements at the successive time slices approaching the L-H transition. A force balance decomposition identifies turbulence-driven

arXiv:2605.22960v1 Announce Type: new Abstract: Nuclear fusion is an attractive source of energy because the fuel is abundant and it produces low levels of carbon emissions. The tokamak, which confines a plasma using magnetic fields, is the most mature nuclear fusion reactor concept. Maximizing energy confinement by minimizing turbulent heat loss while also minimizing damage to the reactor is essential for producing efficient, commercially viable fusion reactors. Heat exhaust methods used in the scrape-off layer (SOL) of the tokamak greatly influence performance. Conventional heat exhaust methods focus on minimizing reactor damage rather than maximizing confinement. The low-recycling regime, a newer approach, focuses on maximizing energy confinement. Studying the low-recycling regime, which features a high temperature and low density SOL, requires new modeling tools. We have developed the gyrokinetic code Gkeyll into an appropriate tool, and we use it to demonstrate the viability of

arXiv:2605.23188v1 Announce Type: new Abstract: Spiking Neural Networks (SNNs) provide an energy-efficient paradigm for visual recognition. We present SpikingMoE, which integrates a spike-driven Transformer with a Mixture-of-Experts (MoE) framework for dynamic computation. Inspired by the lateral geniculate nucleus (LGN), a spike-driven prompt (SDprompt) enables input-dependent expert routing in a biologically plausible manner. By replacing standard MLPs with spike-compatible expert modules and enforcing binary spike communication, SpikingMoE is designed for neuromorphic hardware. Experiments on CIFAR-10 and CIFAR-100 achieve 94.09% and 74.54% top-1 accuracy, showing that modular expert routing can be incorporated while retaining reasonable performance. To our knowledge, SpikingMoE is the first open-source SNN framework that integrates MoE into a spike-driven Transformer with LGN-inspired routing.

arXiv:2605.22868v1 Announce Type: new Abstract: Autonomous systems and smart-industry deployments increasingly split computation across near-sensor, edge, and cloud resources, where tight energy, latency, and reliability budgets demand run-time adaptivity. In practice, deciding what to compute and transmit at each point is pivotal; yet as multimodal sensor suites (cameras, LiDAR/depth, etc.) proliferate at the edge, most prior approaches either (i) fuse modalities on powerful servers or (ii) apply uni-modal near-sensor filters that ignore cross-modal dependencies, leading to redundant transmissions or missed events. We present FusionSense, a fusion-aware intelligent sensing framework for energy-constrained autonomous edge systems. Lightweight near-sensor classifiers are trained via a three-step procedure: (i) a server-side fusion model learns the downstream task, (ii) filter-out-safe (FoS) labels quantify each modality's necessity relative to the fused decision, and (iii) an edge-side

Tungsten’s superior performance in extreme environments makes it a leading candidate for plasma-facing components (PFCs) in fusion reactors,

arXiv:2605.22680v1 Announce Type: new Abstract: Using a simple circular tokamak geometry, we show the well-known `second stability region' of MHD-ballooning modes exists for linear gyrokinetics too -- whether electrostatic or electromagnetic -- and we suggest that the plasma enters this region in H-mode as a consequence of the bootstrap current and Shafranov shift altering the magnetic field, which may occur if the normalised pressure gradient is $\alpha_{\rm MHD} \simgt 1$ and collisionality is low. By performing simulations in more realistic magnetic geometries, we demonstrate a large reduction in collisionless, electrostatic turbulent transport when going from density and temperature profiles typical of L- and H-mode, respectively. This reduction is shown to be a consequence of both the bootstrap current lowering the global magnetic shear, and the pressure gradient altering the local magnetic shear, pushing the plasma towards the second-stability region. A path connecting the L-

arXiv:2605.22105v1 Announce Type: new Abstract: As first shown by Shafranov, toroidal plasmas in magnetohydrodynamic equilibrium tend to expand in major radius when the pressure is increased. Here, an average measure of the resulting Shafranov shift is introduced, and its properties are discussed for various classes of optimised stellarator configurations. It is shown to be particularly small in quasi-helical and quasi-isodynamic stellarators with a large number of field periods, which are thus particularly robust to variations in the plasma pressure.

arXiv:2605.21814v1 Announce Type: new Abstract: A common feature of most numerically optimized stellarator geometries is the presence of sharp ridges on outer flux surfaces, irrespective of the rotational transform. Despite their importance, an analytical theory for their existence has been lacking. In this work, we demonstrate that ridges are not artifacts but mathematical necessities. We develop such a theory for devices with quasisymmetry (QS). We demonstrate that QS exhibits close connections with the theory of geometrical optics, following Parker's ``optical analogy" (E.N. Parker, Geophys. Astrophys. Fluid Dyn, 1989). By mapping vacuum QS to the eikonal equation of geometrical optics, we derive the conditions for ridge formation, identified as field line caustics where magnetic field lines focus. Furthermore, we prove a geometric theorem for stellarator coil design: both ridges and filamentary coils must lie on the zero-determinant manifold of the magnetic gradient tensor. This

arXiv:2605.21637v1 Announce Type: new Abstract: The Pfirsch-Schl\"uter current is a current that flows along the magnetic field lines in a toroidal plasma equilibrium that is required to make the plasma current density divergence free in the presence of a plasma-pressure gradient. A distortion in the plasma shape is caused by the Pfirsch-Schl\"uter current, and it is desirable to minimize both the strength and the distance this current flows along the magnetic field lines. The Pfirsch-Schl\"uter current is localized within a half period of a stellarator when $d\ell/B$ integrated over the half period is the same for all lines in the magnetic surface. It is shown that within parts in a thousand this is the same condition as the distance $\ell_{p/2}$ required for a field line to cross the half period being the same for all lines in the surface. To make the $\ell_{p/2}$'s the same, the lines started on the small major radius side of the plasma must undergo wiggles to make their

arXiv:2509.17086v2 Announce Type: replace Abstract: Detecting and localizing poultry is essential for advancing smart poultry farming. Despite the progress of detection-centric methods, challenges persist in free-range settings due to multiscale targets, obstructions, and complex or dynamic backgrounds. To tackle these challenges, we introduce an innovative poultry detection approach named SFN-YOLO that utilizes scale-aware fusion. This approach combines detailed local features with broader global context to improve detection in intricate environments. Furthermore, we have developed a new expansive dataset (M-SCOPE) tailored for varied free-range conditions. Comprehensive experiments demonstrate our model achieves an mAP of 80.7% with just 7.2M parameters, which is 35.1% fewer than the benchmark, while retaining strong generalization capability across different domains. The efficient and real-time detection capabilities of SFN-YOLO support automated smart poultry farming.

arXiv:2605.22743v1 Announce Type: new Abstract: Parameter-efficient fine-tuning enables fast personalization of text-to-image diffusion models, but composing multiple custom concepts remains challenging due to representation interference. Existing modular methods either rely on expensive post-hoc fusion or freeze adaptation subspaces, which limit expressiveness and concept fidelity. To address this trade-off, we propose Sequential regularized LoRA (SeqLoRA), a constrained continual learning framework that jointly optimizes both LoRA factors via bilevel optimization. Theoretically, we establish strong convergence guarantees for our algorithm and model the residual layer activations as a matrix sub-Gaussian process to derive high-probability bounds on catastrophic forgetting. We further prove that learning the LoRA basis from data minimizes residual interference energy more effectively than frozen-basis methods. Experiments on multi-concept image generation demonstrate that SeqLoRA

arXiv:2605.22635v1 Announce Type: new Abstract: While multi-task learning based automatic radiology report generation (RRG) is widely adopted to ensure clinical consistency, most focus on architectural designs yet remain limited to coarse linear scalarization strategies. These strategies cannot effectively balance the hard constraints of discriminative clinical supervision with the smoothness requirements of report generation. To address these problems, we analyze the failure mechanism of linear scalarization from the perspective of gradient dynamics, utilizing the stochastic differential equation (SDE) framework to characterize it as a "Double Dilemma" of drift term deviation and diffusion term decay. Based on this, we propose a backbone-agnostic optimizer named Conflict-Averse Magnitude-Enhanced Gradient Descent (CAME-Grad). Through conflict-averse direction rectification and magnitude-enhanced energy injection, the algorithm not only ensures geometric validity, but also avoids

arXiv:2605.21106v1 Announce Type: cross Abstract: When multiple electron temperature diagnostics converge on the same value, the standard inference is that the measurement is robust. We show that this convergence is a structural consequence of the shared ionization bottleneck in any plasma where the electron Knudsen number exceeds $\sim 0.01$: all diagnostics downstream of collisional ionization report the effective temperature $T_{\rm eff}$, not the core temperature $T_{\rm core}$. Their agreement is a single measurement reported $N$ times. We introduce a taxonomy: Type A (ionization-gated, $T_{\rm eff}$), Type B (bulk-sampling, $T_{\rm core}$), Type C (distribution-resolving). The ratio $R = T_A/T_B$ yields $\kappa = 3R/[2(R-1)]$ directly. Applied to the solar corona ($R = 2.4$, $\kappa \approx 2.5$) and the tokamak scrape-off layer, single kappa distributions ($\kappa \approx 2$--$10$) reproduce published bi-Maxwellian EEDF decompositions to 3--8\% RMS with one fewer parameter, and

arXiv:2605.20993v1 Announce Type: new Abstract: We present a compact photodiode-based diagnostic array developed to study low-energy neutral beam injection in the LTX-$\beta$ spherical tokamak. The in-vacuum diagnostic combines filtered soft-x-ray (SXR), narrowband Lyman-$\alpha$, and unfiltered AXUV photodiode rows with partly overlapping, nearly coincident tangential views of the plasma, including the neutral beam path. This geometry provides simultaneous sensitivity to beam-induced SXR emission; neutral-hydrogen line radiation associated with recycling, fast neutrals and fueling; and broadband emission that can include direct neutral impacts from fast-ion charge-exchange losses. Initial measurements from 12-20 keV hydrogen beam operation show beam-synchronous detector responses in all three modalities. The unfiltered AXUV signals exhibit millisecond-scale rise and fall times that are much slower than the detector response, that vary across sightlines, and depend on

arXiv:2605.20432v1 Announce Type: new Abstract: We present the Neutron Scattered Spectra Tool (NeSST), an open-source Python package for rapidly constructing primary and singly scattered neutron spectra from inertial confinement fusion (ICF) implosions. NeSST evaluates primary spectra for deuterium-tritium (DT), deuterium-deuterium (DD) and tritium-tritium (TT) reactions. Differential and total nuclear cross sections are read directly from Evaluated Nuclear Data File (ENDF) libraries. This enables elastic ($n$D, $n$T) and inelastic [D$(n,2n)$p, T$(n,2n)$D] scattering from DT fuel, as well as scattering from additional ablator materials such as $^{12}$C, to be treated within a common framework. Relativistic corrections to elastic scattering kinematics are included. Areal density asymmetries are incorporated through a Legendre mode expansion of the neutron-averaged projected areal density, allowing the spectral signatures of implosion non-uniformities to be computed and fitted. The

arXiv:2605.20260v1 Announce Type: new Abstract: Nuclear batteries require radioisotopes with specific combinations of half-life, decay mode, and radiation properties, yet most candidate fuels lack scalable production routes. We show how the future availability of deuterium-tritium (D-T) fusion neutrons could enable manufacturing nuclear battery radioisotopes at many orders of magnitude higher rate than at present. We assess the capability of 14 MeV D-T fusion neutrons to produce nuclear battery radioisotopes by simulating feedstock material irradiation with neutrons. Promising radioisotope candidates include ${}^{147}$Pm, ${}^{63}$Ni, ${}^{39}$Ar, and ${}^{137}$Cs. Some feedstocks allow a radioisotope to be produced at scale while also closing the tritium fuel cycle, resulting in hundreds to over one thousand kilograms of high specific activity radioisotope per gigawatt thermal year of D-T fusion irradiation. We perform OpenMC simulations of an enriched ${}^{148}$Nd blanket for a

arXiv:2605.20713v1 Announce Type: new Abstract: Multimodal IE in social media is difficult because a post may attach multiple images that are weakly related, redundant, or even misleading with respect to the text. In this setting, always-on multimodal fusion wastes computation and can amplify spurious visual cues. The core challenge is to decide, for each candidate span or marked entity pair, whether vision should be consulted at all and, if so, which small subset of images provides trustworthy evidence. We propose SAVER, a selective vision-as-needed framework for multimodal named entity recognition and multimodal relation extraction. SAVER uses a Conformal Groundability Gate (CGG) to estimate span-level visual groundability in MNER, derive pair-level activation in MRE from the two marked entities, and calibrate the activation threshold on a held-out split via a conformal-style procedure with Clopper--Pearson upper bounds. When activated, a submodular relevance--diversity selector

arXiv:2605.20651v1 Announce Type: new Abstract: Existing deep learning frameworks for Optical Coherence Tomography Angiography (OCTA) vessel segmentation are largely derived from the U-Net architecture, which serves as the foundation for most current designs. However, most of these methods focus only on holistic representation, struggling to address the problem of low local contrast unique to OCTA, which leads to vessel discontinuities and loss of detail. To address these problems, we propose LSENet, which builds upon the U-Net architecture by introducing three core innovative modules: To address vessel discontinuities, we introduce the Patch Information Enhance module (PIE), which replaces standard skip connections to execute patch-wise attention. To mitigate detail loss, the Multiscale Feature Fusion module (MFF) is proposed to feed the PIE module rich, multi-scale information by extracting visually interpretable features from both the original input and preceding layers. Finally,

Standardizing calculations of the helium byproducts generated in advanced fission and fusion energy system materials can increase reactor safety and longevity, according to a study led by University of Michigan Engineering with collaborators at Oak Ridge National Laboratory and its management contractor UT-Battelle.

arXiv:2605.19057v1 Announce Type: cross Abstract: Magnetohydrodynamics (MHD) couples the Navier--Stokes and Maxwell equations into a nonlinear system of partial differential equations governing stellar interiors, astrophysical jets, fusion plasmas, and space weather. Numerical advances, including finite-volume Godunov schemes, constrained-transport algorithms, high-order spectral-element and discontinuous-Galerkin discretisations, and adaptive mesh refinement, have made MHD a predictive tool for solar eruptions, tokamak confinement, and magnetised turbulence. A fundamental barrier nevertheless remains. In three-dimensional MHD turbulence, the degrees of freedom required to resolve all active scales grow as $\mathcal{O}(\mathrm{Re}^{9/4})$ or faster, where $\mathrm{Re}$ is the Reynolds number. Direct numerical simulation is therefore intractable at astrophysical and fusion-relevant parameters, particularly when the Lundquist number $S$ exceeds $10^{10}$ and both viscous and resistive

arXiv:2605.19666v1 Announce Type: new Abstract: Accurate cardiac output (CO) estimation from photoplethysmography (PPG) is promising for unobtrusive hemodynamic monitoring, but remains difficult since CO is jointly determined by cardiac function and vascular tone. Conventional feature-based models use physiologically meaningful PPG descriptors, yet depend on accurate pulse detection and may miss latent temporal relationships. In contrast, fully end-to-end deep learning models learn directly from raw PPG but often underuse established PPG-derived prior information. Here, we introduce the Cross-View Attention Fusion Network (CVAF-Net), a prior-guided dual-view deep learning model for CO estimation from short, fixed-length PPG segments. CVAF-Net processes raw PPG as a temporal view and a feature sequence map (FSM) as a structured prior-guided view, and fuses the two representations through cross-view attention. The model was independently evaluated using 5-, 15-, and 30-s segments from

arXiv:2605.18968v1 Announce Type: new Abstract: Laser-driven neutron sources offer ultrashort pulse durations and extreme peak fluxes inaccessible to conventional facilities, enabling novel time-of-flight(TOF) spectroscopy and nuclear astrophysics measurements. We present the first complete start-to-end simulation comparison of deuterium-deuterium (DD) bulk fusion and laser wakefield acceleration-driven photonuclear neutron sources, evaluated for fast neutron capture relevant to the r-process. The simulation chain couples particle-in-cell modeling of the laser-plasma interaction, Geant4 Monte Carlo neutron transport with shielding and background characterization, and a NON-SMOKER-based event generator for multi-neutron capture on Au197 and Rh103. We derive scaling laws for neutron yield, pulse duration, and peak flux from 1J terawatt to 250J petawatt-class systems, including DD bulk fusion scaling laws specific to the short-pulse regime where volumetric ion heating via plasma

arXiv:2605.19666v1 Announce Type: cross Abstract: Accurate cardiac output (CO) estimation from photoplethysmography (PPG) is promising for unobtrusive hemodynamic monitoring, but remains difficult since CO is jointly determined by cardiac function and vascular tone. Conventional feature-based models use physiologically meaningful PPG descriptors, yet depend on accurate pulse detection and may miss latent temporal relationships. In contrast, fully end-to-end deep learning models learn directly from raw PPG but often underuse established PPG-derived prior information. Here, we introduce the Cross-View Attention Fusion Network (CVAF-Net), a prior-guided dual-view deep learning model for CO estimation from short, fixed-length PPG segments. CVAF-Net processes raw PPG as a temporal view and a feature sequence map (FSM) as a structured prior-guided view, and fuses the two representations through cross-view attention. The model was independently evaluated using 5-, 15-, and 30-s segments from

arXiv:2605.18878v1 Announce Type: cross Abstract: Hospital readmission within 30 days of discharge is a leading driver of morbidity, mortality, and avoidable healthcare expenditure in congestive heart failure (CHF). Current clinical risk stratification tools rely primarily on non-imaging data and exhibit limited predictive performance. Point-of-care lung ultrasound (LUS) offers a sensitive, noninvasive window into the pulmonary congestion that characterizes CHF decompensation, yet its prognostic utility for readmission prediction remains largely unexplored. We present a pilot feasibility study, the first systematic machine learning study using B-mode LUS acquired during hospitalization to predict 30-day CHF readmission. Quantitative spatiotemporal embeddings are extracted from a pretrained Temporal Shift Module (TSM) ResNet-18 encoder, and interpretable biomarker features are separately evaluated. Through structured ablations over lung view, temporal representation, multi-view

arXiv:2605.19233v1 Announce Type: new Abstract: Unmanned aerial vehicles (UAVs) are cyber-physical systems whose attack surface spans networked avionics and on-board sensor fusion: a compromised GPS or battery module can mimic a benign mission segment and evade naive anomaly detectors. We present a leakage-free evaluation of quantum machine learning for UAV anomaly detection on the multi-sensor TLM:UAV benchmark. Three contributions support the study. (i) A group-aware temporal protocol (B2) partitions the dataset into ten contiguous TimeUS blocks and evaluates over ten seeds, eliminating the inflation produced by random stratified splits that mix neighbouring samples. (ii) A three-mode feature audit (full/loose/strict) quantifies how much accuracy stems from instantaneous physical signals versus contextual proxies (cumulative energy, battery state, GPS trajectory). (iii) A hybrid XGBoost + Data Reuploading (DRU) classifier is benchmarked against five paired non-linear controls (raw,

This company says its pulsed plasma machine will deliver electricity to the grid by 2029. Some physicists warn that its promises are outrunning what the technology has proved

arXiv:2605.17334v1 Announce Type: cross Abstract: Reliable early detection of lithium-ion battery degradation requires health indicators that are physically interpretable and computable from routine cycler telemetry without access to the degradation region. We introduce \textsc{CausalHealth}, a framework that applies causal graph discovery and $k$-nearest-neighbour transfer entropy to per-cycle voltage, current, temperature, and resistance time series, and organises twelve resulting anomaly scores into three signal-class bundles (Magnitude-shift, Predictive-residual, Complexity-entropy) -- with Isolation Forest reported separately as it falls below the bundle reliability threshold -- to characterise detection sensitivity across ten commissioning fractions (5--30\,\%). The Magnitude-shift class achieves 100\,\% detection across all seven tested cells spanning LFP (MIT--Stanford MATR) and LCO (NASA PCoE, CALCE CS2) chemistries, with a lead time of up to 402 cycles before conventional

arXiv:2605.17888v1 Announce Type: new Abstract: Long-horizon prediction of three-dimensional (3D) wall-bounded turbulence with machine-learning methods remains a challenging task, due to the rapid accumulation of autoregressive errors and the substantially computational cost. To address these challenges, we present a hybrid machine-learning framework, in which a channel-time-attention Swin-UNet (CTA-Swin-UNet) and a multi-time-scale fusion correction (MTFC) strategy are developed to predict the turbulent flow fields in a wall-parallel plane, with affordable computational cost. Then, 3D flow fields are reconstructed via a resolvent-based spectral linear stochastic estimation (SLSE), rooting from the predicted planar flow. Results show that the CTA-Swin-UNet outperforms the baseline models (LSTM, FNO and traditional Swin-UNet) in both single-step prediction and autoregressive rollouts, indicating the effectiveness of introducing the CTA module into the Swin-UNet architecture. At the

arXiv:2605.17138v1 Announce Type: new Abstract: Future nuclear fusion reactors will have to confine plasma with strong kinetic gradients and small fractions of fusion-born energetic particles (EP) that are ~100 times hotter than the thermal ions. In our analysis, we assume the existence of a stable MHD equilibrium and study the unstable plasma perturbations. In this electromagnetic, kinetic, multi-scale, self-organizing system, all species contribute both to the Shafranov shift (equilibrium effect) and to the plasma $\beta$ (plasma response). Nonetheless, due to the high complexity of the problem, many works neglect these effects. We use the global, gyrokinetic code ORB5 to study the plasma stability. Starting from an electrostatic, thermal plasma with adiabatic electrons in a $\beta = 0$ ideal-MHD equilibrium, we systematically increase the realism of our models. And study the linear stability and nonlinear fluxes of Toroidal Alfv\'en Eigenmodes (TAE), and the Ion Temperature

arXiv:2605.16734v1 Announce Type: new Abstract: The dynamics of energetic particle (EP) species, born from fusion reactions or plasma heating schemes, are critical for predicting the behavior of magnetic confinement fusion experiments and future fusion reactors. Because energetic particles are largely collisionless, the orbits of Monte Carlo samples drawn from a given distribution function can be efficiently integrated in prescribed electromagnetic fields. In addition to the static magneto-hydrodynamic (MHD) equilibrium fields produced by the electromagnetic coils of a fusion device, MHD waves can be excited by -- and resonantly transport -- energetic particle populations. FIRM3D is an open-source Python/C++/CUDA software suite for modeling energetic particle dynamics in 3D magnetic fields, available at https://github.com/ColumbiaStellaratorTheory/firm3d. The core guiding-center integration routines grew out of SIMSOPT (Landreman et al., 2021), but have been extended to include

arXiv:2605.16663v1 Announce Type: new Abstract: It should be possible to generate multi-ampere spin-polarized beams of hydrogen isotopes by repeated charge-transfer collisions in highly spin-polarized Cs vapor. Estimates suggest that off-resonant Raman pumping with kW scale narrowband tunable light at 895 nm should be able to produce a 1 m long, 10 cm diameter volume of 80\% polarized Cs vapor. The charge transfer collisions between the Cs and hydrogen result in a high nuclear spin-polarized negative ion beam that can be subsequently accelerated to high energy, neutralized, and be used to heat fusion plasmas with resulting increases in the fusion conversion efficiency.

arXiv:2605.17888v1 Announce Type: cross Abstract: Long-horizon prediction of three-dimensional (3D) wall-bounded turbulence with machine-learning methods remains a challenging task, due to the rapid accumulation of autoregressive errors and the substantially computational cost. To address these challenges, we present a hybrid machine-learning framework, in which a channel-time-attention Swin-UNet (CTA-Swin-UNet) and a multi-time-scale fusion correction (MTFC) strategy are developed to predict the turbulent flow fields in a wall-parallel plane, with affordable computational cost. Then, 3D flow fields are reconstructed via a resolvent-based spectral linear stochastic estimation (SLSE), rooting from the predicted planar flow. Results show that the CTA-Swin-UNet outperforms the baseline models (LSTM, FNO and traditional Swin-UNet) in both single-step prediction and autoregressive rollouts, indicating the effectiveness of introducing the CTA module into the Swin-UNet architecture. At the

arXiv:2605.17334v1 Announce Type: cross Abstract: Reliable early detection of lithium-ion battery degradation requires health indicators that are physically interpretable and computable from routine cycler telemetry without access to the degradation region. We introduce \textsc{CausalHealth}, a framework that applies causal graph discovery and $k$-nearest-neighbour transfer entropy to per-cycle voltage, current, temperature, and resistance time series, and organises twelve resulting anomaly scores into three signal-class bundles (Magnitude-shift, Predictive-residual, Complexity-entropy) -- with Isolation Forest reported separately as it falls below the bundle reliability threshold -- to characterise detection sensitivity across ten commissioning fractions (5--30\,\%). The Magnitude-shift class achieves 100\,\% detection across all seven tested cells spanning LFP (MIT--Stanford MATR) and LCO (NASA PCoE, CALCE CS2) chemistries, with a lead time of up to 402 cycles before conventional

arXiv:2605.17675v1 Announce Type: new Abstract: The widespread adoption of AI-assisted development in scientific software is not a future concern -- it is a present reality. Researchers are already using large language models to write code, generate test cases, and draft documentation, yet this practice remains largely unacknowledged and unguided in formal workflows and published work. This ad hoc, ungoverned use of AI represents a systemic risk to scientific software quality, particularly in safety-relevant modeling and simulation tools subject to strict Software Quality Assurance (SQA), or even Nuclear Quality Assurance Level 1 (NQA-1) standards, for which traceability, independent verification, and documented procedures are paramount. The question facing the scientific software community is, therefore, not whether to permit AI-assisted development, but how to govern it responsibly. This paper proposes guidance for AI-assisted code development in the context of strict software

arXiv:2605.17640v1 Announce Type: new Abstract: Retrieval-augmented generation from videos requires systems to retrieve relevant audiovisual evidence from large corpora and synthesize it into coherent, attributed text. Current approaches struggle at both ends: retrieval methods fail on complex, multi-faceted queries that cannot be captured by a single embedding, while generation methods lack the high-level reasoning needed to synthesize across multiple videos and face memory constraints over long, multi-video contexts. We present MARQUIS: a three-stage pipeline that addresses these limitations through (1) query expansion, fusion, and reranking, (2) calibrated structured evidence extraction, and (3) article generation from extracted evidence, optionally controlled by an RLM. On the MAGMaR2026 shared task, we improve retrieval performance from 0.195 to 0.759 (nDCG@10). For article generation, ITER-QA-BASE improves average human score from 3.09 to 3.83 over the CAG baseline, while

arXiv:2605.17591v1 Announce Type: new Abstract: Aggregate object detection metrics inherently mask catastrophic and repeatable failures in operationally critical, long-tail minority classes. This paper formally defines this pervasive vulnerability as the Hard-Category Reliability Problem (HCRP): the fundamental architectural challenge of strictly rectifying vulnerable categories without compromising the performance boundaries of stable classes under stringent protocols. To systematically dismantle this limitation, we propose Error-Decomposed Class-Conditional Fusion (ED-CCF), an elegant decision-layer inference framework. Diverging from heuristic global post-processing, ED-CCF projects predictions into a sophisticated quad-state error taxonomy, dynamically activating calibration pathways exclusively upon rigorous empirical justification. On a highly constrained 600-image validation benchmark, isolating cz as the critical vulnerability (HCEC=0.86, BSR=0.14), our framework achieves a

arXiv:2605.15935v1 Announce Type: cross Abstract: Plasma shape control in tokamaks requires a real-time controller that tracks dynamically changing shape targets while tolerating diagnostic failures. Classical approaches decompose the problem into equilibrium reconstruction followed by a linear controller, and assume a fixed, fully operational sensor set. We present a reinforcement learning agent that addresses both limitations simultaneously. The agent is trained in NSFsim, a high-fidelity tokamak simulator configured for DIII-D, on a curated dataset of 120 experimental plasma shapes. The shape targets are resampled as random step changes every 0.25 s, exposing the agent to diverse transitions across the full shape envelope. At test time the agent zero-shot tracks dynamic shape sequences; on a held-out static configuration in simulation it achieves a mean shape error of 2.01 cm, and dynamic trajectory following is demonstrated qualitatively in simulation and on the physical device.

arXiv:2605.15844v1 Announce Type: new Abstract: In this study, we develop an interface-contact simulation framework based on physical criteria and machine-learning-assisted classification to describe coalescence and bouncing within a unified formulation. The framework realizes interfacial coalescence and bouncing through the fusion and generation of multiple volume-of-fluid fields. When adjacent interfaces are predicted to coalesce, multiple VOF fields are collapsed into a single VoF field. When approaching interfaces are predicted to bounce, a single VOF field is regenerated into multiple VOF fields, allowing the interfaces to continue evolving independently. With this treatment, the difficulties associated with topological transition, regime-map identification, increasing computational demand, and stochastic behavior during interfacial approach are separated from the interface-tracking procedure. These decisions are instead assigned to a physics-guided machine-learning model with

arXiv:2605.15485v1 Announce Type: new Abstract: We report the first observation on the EXL-50U spherical torus that energetic particles injected by neutral beam injection (NBI) can be stably accelerated to significantly higher energies - reaching up to 2.5 times the injection energy, occurring without significant large-scale magnetohydrodynamic (MHD) bursts. Simulations based on EXL-50U parameters indicate that small-scale magnetic reconnection, mediated by multiple magnetic islands, fails to accelerate bulk thermal ions but efficiently energizes seed fast ions. Unlike global MHD events, such small-scale reconnection is ubiquitous in magnetic confinement devices and does not degrade core confinement. This mechanism offers a novel and potentially universal channel for auxiliary ion heating in future fusion reactors.

arXiv:2605.15935v1 Announce Type: new Abstract: Plasma shape control in tokamaks requires a real-time controller that tracks dynamically changing shape targets while tolerating diagnostic failures. Classical approaches decompose the problem into equilibrium reconstruction followed by a linear controller, and assume a fixed, fully operational sensor set. We present a reinforcement learning agent that addresses both limitations simultaneously. The agent is trained in NSFsim, a high-fidelity tokamak simulator configured for DIII-D, on a curated dataset of 120 experimental plasma shapes. The shape targets are resampled as random step changes every 0.25 s, exposing the agent to diverse transitions across the full shape envelope. At test time the agent zero-shot tracks dynamic shape sequences; on a held-out static configuration in simulation it achieves a mean shape error of 2.01 cm, and dynamic trajectory following is demonstrated qualitatively in simulation and on the physical device.

arXiv:2605.15889v1 Announce Type: new Abstract: Security in cloud computing has become a major concern due to several factors such as layered cloud architectures, dynamic environments, and exposure to unseen or zero-day attacks. Moreover, intrusion detection systems (IDS) typically operate at specific layers and rely heavily on machine learning models, which often perform well in experimental settings but fail to sustain performance in real cloud deployments. In this work, we implement a confidence-aware multilevel intrusion detection system using reinforcement learning tailored for cloud environments. The system secures three distinct layers: network, host, and hypervisor. Machine learning models at each layer detect known attack patterns, while prediction confidence distinguishes reliable decisions from uncertain outcomes. Within the multi-gate flow, low-confidence events pass through a learned-threshold confidence gate (Gate-1), followed by a Chroma memory-matching gate (Gate-2),

arXiv:2605.15549v1 Announce Type: new Abstract: The demand for clean energy is ever increasing, with new nuclear technologies presenting a complementary solution to renewable energies. However, designing and operating these systems is exceptionally difficult, given the complexity of the physical phenomena that interact to form the system dynamics. While high-fidelity simulations help to understand the non-linear, multi-physics interactions within a reactor, they are computationally expensive and rarely suitable for real-time applications. Furthermore, model-based approaches are inherently sensitive to simplifying assumptions required to derive their governing equations and parameters, leading to inevitable discrepancies with real-world measurements. In contrast, Machine Learning (ML) methods have the potential to generate reliable surrogate models which may be able to quickly predict the system's behaviour. However, the number of data-driven methods that can potentially be used for

arXiv:2605.15252v1 Announce Type: new Abstract: Modern pedestrian dead reckoning (PDR) systems rely on fusing noisy and biased estimates of position, velocity, and calibrated orientation derived from loosely coupled sensors to determine the current pose of a localized object. However, discrepancies in the sampling rates of sensor-specific estimation methods and unreliable transmission pose significant challenges. And traditional methods often fail to effectively fuse multimodal sensor data during dynamic movements characterized by high accelerations, velocities, and rapidly varying orientations. To address these limitations, we propose a simple recurrent neural network (RNN) architecture capable of implicitly forecasting asynchronous sensor data streams from diverse estimation methods along reference trajectories. The proposed approach introduces PDRNN, a modular hybrid AI-assisted PDR system that handles each component as an independent ensemble of machine learning (ML) models to

arXiv:2605.14520v1 Announce Type: new Abstract: We consider the Landau-Coulomb equation for a (hydrogen) plasma heated by an external electric field. In this setting, theoretical and experimental results in plasma physics show the emergence of so-called \emph{runaway electrons} which are linearly accelerating but only lead to a minimal increase of the plasma temperature. Runaway electrons are a major obstacle in nuclear fusion since they can overcome the confinement and damage the structure of the reactor. We rigorously prove the well-posedness of the underlying nonlinear \emph{open} Landau-Coulomb system in a perturbative setting and the conjectured growth bounds for the mean velocity and plasma temperature. We show that the mean velocity is linearly increasing in time, and capture the sharp logarithmic growth of the temperature. Furthermore, we prove that the electron distribution can be asymptotically described by a scattering-type Maxwellian. Due to the different nature of the

arXiv:2605.14939v1 Announce Type: new Abstract: Reliable position and shape control in tokamak plasmas requires accurate real-time regulation of several strongly coupled shape parameters. The control vectors that disentangle these couplings, referred to as \textit{virtual circuits} (VCs), enable independent shape parameter control for a specific Grad--Shafranov (GS) equilibrium. Numerical calculation of VCs is not currently feasible in real time, therefore VCs are usually computed prior to each experiment, using a small number of reference GS equilibria sampled along the desired scenario trajectory, with each VC used to control the plasma within a preset time interval. While effective near the reference equilibrium, this approach can lead to degraded performance as the plasma departs from the reference equilibrium and/or from the desired trajectory, and it complicates the design of robust control strategies for rapidly evolving plasma configurations. In this paper, we construct

arXiv:2605.14129v1 Announce Type: new Abstract: Merging direct and indirect-drive has long been viewed as an optimal hybrid laser-fusion scheme that combines the uniformity of x rays with the efficiency of direct illumination. We present the first integrated 2D simulations of hybrid shock drive (HSD) targets for the OMEGA laser. The HSD scheme [L. Ceurvorst et al., Phys. Rev. E 101 063207 (2020)] uses x rays from a thin Au-coated x-ray converter outer shell to drive the initial shock into a standard direct-drive capsule. Direct illumination is used to implode the target after the first shock. The design effectively suppresses laser-imprint seeding of hydrodynamic instabilities, maintaining shell integrity during the implosion. This scheme will enable fielding low-adiabat, high-convergence implosions on OMEGA with expected performance greatly exceeding those of current designs. HSD targets are projected to significantly enhance fusion yields, potentially increasing the record Lawson

arXiv:2605.15150v1 Announce Type: cross Abstract: We show that the low-energy states of non-Abelian topological orders possess extensive magic which is long-ranged, and cannot be eliminated by a constant-depth local unitary circuit. This refines conventional notions of complexity beyond the linear circuit depth which is required to prepare any topological phase, and provides a new resource-theoretic characterization of topological orders. A central technical result is a no-go theorem establishing that stabilizer states--even up to constant-depth local unitarie--cannot approximate low-energy states of non-Abelian string-net models which satisfy the entanglement bootstrap axioms. Moreover, we show that stabilizer-realizable Abelian string-net phases have mutual braiding phases quantized by the on-site qudit dimension, and that any violation of this condition necessarily implies extensive long-range magic. Extending to higher spatial dimensions, we argue that any state obeying an

arXiv:2605.14939v1 Announce Type: cross Abstract: Reliable position and shape control in tokamak plasmas requires accurate real-time regulation of several strongly coupled shape parameters. The control vectors that disentangle these couplings, referred to as \textit{virtual circuits} (VCs), enable independent shape parameter control for a specific Grad--Shafranov (GS) equilibrium. Numerical calculation of VCs is not currently feasible in real time, therefore VCs are usually computed prior to each experiment, using a small number of reference GS equilibria sampled along the desired scenario trajectory, with each VC used to control the plasma within a preset time interval. While effective near the reference equilibrium, this approach can lead to degraded performance as the plasma departs from the reference equilibrium and/or from the desired trajectory, and it complicates the design of robust control strategies for rapidly evolving plasma configurations. In this paper, we construct

arXiv:2605.14249v1 Announce Type: new Abstract: We present EnergyLens, an end-to-end framework for energy-aware large language model (LLM) inference optimization. As LLMs scale, predicting and reducing their energy footprint has become critical for sustainability and datacenter operations, yet existing approaches either require production-level code and expensive profiling or fail to accurately capture multi-GPU energy behavior. As a result, practitioners lack tools for deciding which optimizations to prioritize and for selecting among existing deployment configurations when exhaustive profiling is impractical. EnergyLens addresses this gap with an intuitive einsum-based interface that captures LLM specifications including fusion, parallelism, and compute-communication overlap, combined with load-imbalance-aware MoE modeling and an empirically driven communication energy model for multi-GPU settings. We validate EnergyLens on Llama3 and Qwen3-MoE across tensor-parallel and

arXiv:2605.13863v1 Announce Type: new Abstract: Anomaly detection in dynamic networks is critical for applications from cybersecurity to industrial monitoring, yet existing methods face challenges in energy efficiency, temporal precision, and adaptability. This paper introduces ASTDP-GAD, a novel Adaptive Spiking Temporal Dynamics Plasticity framework for Graph Anomaly Detection that integrates spiking graph neural networks with STDP learning for energy-efficient neuromorphic detection in dynamic networks. Our framework unifies spiking neural computation, STDP learning, and graph-based anomaly detection through the following key innovations: temporal spike graph encoding with adaptive Leaky Integrate-and-Fire (LIF) dynamics; LIF-based graph attention with lateral inhibition; event-driven hypergraph memory with STDP-inspired prototype updates; spike rate contrast pooling based on spiking irregularity; adaptive STDP layers capturing causal temporal relationships; and multi-scale

arXiv:2605.13164v1 Announce Type: cross Abstract: The extremely low solubility of helium in liquid metals may lead to rapid supersaturation, promoting spontaneous formation of helium bubbles by nucleation. Once nucleated, the stability of these bubbles is governed by the properties of the helium liquid metal interface. In particular, interfacial tension between the immiscible phases controls bubble interactions and induces local pressure inhomogeneities. This work is motivated by the need of a better understanding of helium bubble formation in liquid Pb Li alloys, which are of particular relevance for the design of breeding blankets in the future nuclear fusion reactors. We employ classical molecular dynamics simulations to investigate helium segregation in a range of lead lithium systems, including the limiting cases of pure lead and pure lithium. Changes in local pressure are evaluated from direct mechanical calculations, enabling the characterization of interfacial properties.

arXiv:2605.12723v1 Announce Type: new Abstract: We examine the structure of Direct Simulation Monte Carlo (DSMC)-resolved internal compression layers in rarefied micro-nozzle flows and show that their apparent parametric complexity is largely a registration and finite-thickness scaling effect. A density-gradient diagnostic identifies the compression-layer station \(x_s\), while a jump-based thickness \(\delta_j=\Delta\rho/\max|\partial\rho/\partial x|\) defines a shock-centered coordinate \(\xi_j=(x-x_s)/\delta_j\). In physical coordinates, the leading proper orthogonal decomposition (POD) mode of the centerline density profiles captures only \(83.33\%\) of the fluctuation energy, whereas the jump-scaled coordinate increases this value to \(98.33\%\). A two-dimensional shock-window POD further confirms that this compactness is not a centerline artifact: in the registered \((\xi_j,\eta)\) frame, the first density mode captures \(94.98\%\) and the first two modes capture \(99.05\%\) of

arXiv:2605.11033v1 Announce Type: cross Abstract: TokaMind is a multi-modal transformer (MMT) foundation model pre-trained on tokamak plasma diagnostics data from MAST, where it was shown to outperform CNN-based approaches on fusion benchmarks. We investigate whether its learned representations generalize to physically distinct but structurally analogous domains. Through systematic experimentation across four domains-industrial bearing degradation, NASA CMAPSS turbofan degradation, and two independent power grid PMU datasets-we identify four transfer-favoring characteristics that help explain where TokaMind's pretrained representations are most effective. Power grid synchrophasor data matches this target-domain profile most directly, while industrial degradation datasets demonstrate that TokaMind can still yield useful performance under partial alignment, especially when task design and feature construction expose physically meaningful degradation structure. On the GESL/PNNL 500-event

arXiv:2605.13433v1 Announce Type: new Abstract: Generative recommendation (GR) has emerged as a promising paradigm that replaces fragmented, scenario-specific architectures with unified Transformer-based models, exhibiting scaling-law behavior where recommendation quality improves systematically with increased model capacity and training data. However, deploying GR at scale on Ascend NPUs faces fundamental system-level challenges. These challenges are further exacerbated on Ascend NPUs due to the absence of high-performance implementations for jagged operators and the architectural mismatch between irregular sparse primitives and NPU's dense-computation-optimized design. In this paper, we present \model, an Ascend-affinity training system for generative recommendation that systematically addresses these bottlenecks through three core innovations: (i) Ascend-affinity jagged acceleration, including fusion operators that eliminate padding redundancy and dynamic load balancing that

arXiv:2605.13297v1 Announce Type: new Abstract: Periodic crystals repeatedly instantiate similar local coordination motifs across translated cells and chemically related structures, but current equivariant atomistic models usually encode these patterns only implicitly in dense edge features. We introduce PaMM, a periodic motif memory that augments the UMA eSCN-MD edge encoder with explicit pair and triplet lookup features. Pair motifs are keyed by $(Z_j, Z_i, b_r)$ and triplet motifs by $(Z_j, Z_i, Z_k, b_\theta)$, hashed into fixed-size tables and fused with the baseline edge representation through lightweight gate-only and affine-equipped variants. We evaluate PaMM in a matched UMA-S OMAT setting and focus on a narrow question: whether explicit motif memory helps at a fixed intermediate training budget. At the 10k-step checkpoint, both PaMM variants improve over the plain baseline; gate-only gives the best energy MAE, while the affine-equipped variant gives the best force MAE. A

arXiv:2605.13218v1 Announce Type: new Abstract: Cancer is one of the leading causes of death worldwide, making the development of rapid, minimally invasive, label-free and scalable diagnostic strategies a major challenge in modern oncology. In this context, spectroscopic liquid biopsy has emerged as a promising alternative, as it enables the holistic characterization of biochemical alterations in biological fluids. In this work, we propose a multimodal spectroscopic liquid biopsy framework for multicancer detection based on the combination of Fourier Transform Infrared (FTIR) spectroscopy, Raman spectroscopy, and Excitation-Emission Matrix (EEM) fluorescence spectroscopy together with Machine Learning (ML) methodologies. Serum samples from breast cancer patients, colorectal cancer patients, and healthy controls were analyzed through the three spectroscopic modalities. After modality-specific preprocessing, low-level data fusion (LLDF) was employed to integrate the complementary

Brown dwarfs are notoriously difficult to find. These “failed stars” aren’t big enough to sustain nuclear fusion, and therefore aren’t as bright as more traditional main sequence stars. In fact, they’re nearly invisible in optical light, and faintly visible in infrared. But thanks to dozens of citizen scientists combing through archival infrared datasets from the Wide-field Infrared Survey Explorer (WISE), and a paper published in the Astronomical Journal detailing their work, we now have an additional set of over 3,000 candidate new brown dwarfs in our stellar neighborhood, more than doubling the total number found so far.

arXiv:2605.12116v1 Announce Type: new Abstract: This is the six month progress report to Fusion Energy Science (FES) and the American Science Cloud (AmSC) on the MPEX AI Digtial Twins project that was started in October 2025. There are two milestones to demonstrate the Artificial Intelligence (AI) advantage for MPEX operations and scientific discovery, that will be completed by June 2026. The first is a Helicon AI Hot-Spot Controller (Sec. 3.1), which is the helicon heating component of the more comprehensive planned MPEX AI Hot Spot Digital Twin (Sec. 3). The second is an E-beam Damage Assessment Digital Twin (Sec. 4.1), which is a reduced electron beam damage modality prototype for the MPEX AI Damage Assessment Digital Twin (Sec. 4). These two phase I milestones are on track for the June demonstration. In addition to these two milestones, progress on configuring the Galaxy software interface for automation, validation and data analysis is reported (Sec. 5). This interface now

arXiv:2605.11886v1 Announce Type: new Abstract: Qualification of components operating in future fusion power plants will be heavily reliant on simulations of component behaviour. The lack of representative test environments for many aspects of the expected operating environment will necessitate full or partial virtual qualification of components. The cornerstone of virtual qualification is credible validation of the simulation models on which it relies. In this work, we present a probabilistic model validation framework that forms the basis for implementation of virtual qualification in fusion. We demonstrate our framework on a representative component; a high heat flux heat sink subject to a tightly coupled multi-physics loading. We perform data-rich, optimised experiments, in which we implement high fidelity diagnostics and rigorously quantify aleatoric and epistemic uncertainty on all measurements. Our simulation approach efficiently samples input uncertainty distributions to

arXiv:2605.11308v1 Announce Type: new Abstract: Conductivity models for warm dense matter inform simulations of planetary structure and fusion experiments. State-of-the-art conductivity calculations based on density functional theory approximate many-body physics and neglect electron-electron scattering lifetimes. We introduce a many-body framework for electrical conductivity using the GW approximation of the electronic self-energy. For beryllium, improved transition energies yield a surprisingly large reduction in low-temperature DC conductivity, while electron-electron scattering primarily reduces high-temperature DC conductivity.

arXiv:2605.11033v1 Announce Type: new Abstract: TokaMind is a multi-modal transformer (MMT) foundation model pre-trained on tokamak plasma diagnostics data from MAST, where it was shown to outperform CNN-based approaches on fusion benchmarks. We investigate whether its learned representations generalize to physically distinct but structurally analogous domains. Through systematic experimentation across four domains-industrial bearing degradation, NASA CMAPSS turbofan degradation, and two independent power grid PMU datasets-we identify four transfer-favoring characteristics that help explain where TokaMind's pretrained representations are most effective. Power grid synchrophasor data matches this target-domain profile most directly, while industrial degradation datasets demonstrate that TokaMind can still yield useful performance under partial alignment, especially when task design and feature construction expose physically meaningful degradation structure. On the GESL/PNNL 500-event

arXiv:2605.11753v1 Announce Type: new Abstract: Multimodal summarization requires models to jointly understand textual and visual inputs to generate concise, semantically coherent summaries. Existing methods often inject shallow visual features into deep language models, leading to representational mismatches and weak cross-modal grounding. We propose a unified framework that jointly performs text summarization and representative image selection. Our system, SPeCTrA-Sum (Sampler Perceiver with Cross-modal Transformer and gated Attention for Summarization), introduces two key innovations. First, a Deep Visual Processor (DVP) aligns the visual encoder with the language model at corresponding depths, enabling hierarchical, layer-wise fusion that preserves semantic consistency. Second, a lightweight Visual Relevance Predictor (VRP) selects salient and diverse images by distilling soft labels from a Determinantal Point Processes (DPP) teacher. SPeCTrA-Sum is trained using a multi-objective

arXiv:2605.11735v1 Announce Type: new Abstract: The efficient operation of modern cellular networks hinges on the accurate analysis of spatio-temporal traffic data. Mastering these patterns is essential for core network functions, chiefly forecasting future load to pre-empt congestion and imputing missing values caused by sensor failures or transmission errors to ensure data continuity. While deeply connected, forecasting and imputation have historically evolved as separate sub-fields. The dominant paradigm, Spatio-Temporal Graph Neural Networks (STGNNs), while effective, are often specialized, computationally intensive, and exhibit limited generalization. Concurrently, adapting large pre-trained language models (LLMs) offers a powerful alternative for sequence modeling, yet existing approaches provide weak structural guidance, leading to unstable convergence and a narrow focus on forecasting. To bridge these gaps, we propose U-STS-LLM, a unified framework built on a spatio-temporally

arXiv:2605.11705v1 Announce Type: new Abstract: The training of large multimodal models fundamentally relies on massive image-text datasets, which inevitably incur prohibitive computational overhead. Dataset selection offers a promising paradigm by identifying a highly informative coreset. However, existing approaches suffer from two critical limitations: (i) single-modality-dominated sampling methods, which ignore the fine-grained cross-modal information imbalance inherent in multimodal datasets and thus lead to semantic loss in the other modality; and (ii) coarse-grained sample-scoring-based sampling methods, where the selected coreset tends to be biased toward the scoring model, making it difficult to guarantee distributional equivalence between the coreset and the original dataset. Meanwhile, existing distribution matching and discrete sampling strategies often fail to jointly account for global semantic structure, local fine-grained details, and redundancy-aware coverage in dense

arXiv:2605.08597v1 Announce Type: cross Abstract: To effectively describe the plasma with strong magnetic field, the force-free electrodynamics was introduced, within which the Grad-Shafranov equation plays the key role. The Grad-Shafranov equation governs the global structure of a electromagnetic field in equilibrium with symmetries. It is widely applicable in an amount of scenarios, such as the tokamak, the solar corona, the magnetosphere of Earth, neutron star and black hole, etc. However, in different situations, the Grad-Shafranov equation is expressed differently, and the derivations might be complicated. In this work, via the language of differential form, we provide a general expression of Grad-Shafranov equation, from which the expression in any specific situation can be quickly obtained. Additionally, we present a Lagrangian density for a scalar field whose on-shell condition is precisely the Grad-Shafranov equation.

arXiv:2605.10694v1 Announce Type: new Abstract: A central unresolved question in fusion energy research is whether energetic alpha particles, the primary products of deuterium-tritium fusion reactions, enhance or degrade plasma confinement. In burning plasmas, the operating regime of future devices such as ITER and SPARC, alpha particles become the dominant heating source, yet their impact on confinement has remained uncertain. Here, we present self-consistent simulations of burning plasmas that simultaneously evolve microturbulence, alpha-particle heating, and macroscopic plasma profiles to steady state, and find that alpha particles can substantially improve confinement. Fusion-born alpha particles weakly destabilize toroidal Alfven eigenmodes (TAEs), which nonlinearly enhance zonal flows that shear apart and suppress ion-scale turbulence. The resulting reduction in turbulent heat transport drives stronger core profile peaking, increasing alpha heating by up to 25% and establishing

arXiv:2605.10465v1 Announce Type: new Abstract: Reliable modeling and control of core density is essential for reactor-relevant magnetic confinement fusion operation, motivating cryogenic pellet injection as a primary fueling actuator and the need for predictive pellet source models in integrated modeling. Here we present an upgrade of the physics-based pellet code HPI2 in which the plasmoid release spatial step is determined self-consistently from ablation physics, $dx_{var}=v_{\mathrm{pel}}\,t_{\mathrm{exit}}$ (optionally rescaled to trade accuracy for computational cost), removing an ad-hoc discretization parameter and improving numerical robustness across injection conditions. The upgraded model is first validated in stand-alone against a high-field-side pellet-fueled, ohmic, WEST discharge (#58656) by comparing synthetic and measured interferometry line-integrated density increments, obtaining a mean error of $\sim 10\%$. We then perform full-radius, time-dependent integrated

arXiv:2605.09913v1 Announce Type: new Abstract: Tokamak design is inherently challenging due to several cross-competing effects which require a careful and calibrated treatment to obtain an optimal operational envelope. Incorporating physics across varied fidelities is crucial in this exercise. Jenga is developed as a unified design and modeling framework for tokamaks, seamlessly coupling systems-level studies to high-fidelity models based on first principles. In this work, static Grad-Shafranov (GS) equilibrium for an entire pulse and the neutronics study of the Mega Ampere Spherical Tokamak Upgrade (MAST-U) tokamak are carried out in Jenga. Coil currents and plasma profiles from the EFIT++ reconstruction of MAST-U shots are used to reproduce the plasma poloidal flux and shape targets at different time slices. The results from Jenga are also in good agreement with FreeGSNKE and Fiesta codes. Neutronics analysis is performed for a hypothetical 50-50 mixture of deuterium-tritium (DT)

arXiv:2605.09720v1 Announce Type: new Abstract: This paper expands on the TRANSP description given in Computer Physics Communications 312 (2025) 109611 by describing recent progress in TRANSP's predictive functionality and emphasizing the development of the PT_SOLVER module and integration of the high-fidelity T3D/GX framework for plasma profile prediction using a high-fidelity gyrokinetic model for turbulent transport. PT_SOLVER is a modular, multi-region, parallel solver for coupled transport equations of particle density, electron and ion energy, and toroidal angular momentum that uses an implicit Newton method to advance the solution of these equations. The numerical formulation includes source coupling, moving-geometry terms, and nonlinear stabilization based on modified Peclet numbers, thereby enabling the PT_SOLVER to handle the stiffness associated with gradient-dependent transport models. Stabilization occurs via a nonlinear function controlling discretization in zones of

arXiv:2605.09191v1 Announce Type: new Abstract: The energy distribution of energetic protons inside a solid target is a key quantity governing nuclear reaction yields and energy deposition in high-intensity laser-driven fusion, including nonthermal proton--boron (p--B) schemes and proton fast ignition. Yet it has remained inaccessible to conventional particle diagnostics, which detect only ions escaping the target and are perturbed by intense plasma electromagnetic fields. Here we establish a quantitative diagnostic that uses nuclear activation reactions occurring within the target itself as an internal probe of the in-solid proton energy distribution. Applied to laser-driven p--B fusion experiments on the kJ-class laser, the method reconstructs an exponential-equivalent in-solid proton energy distribution from the absolute yields of $^{11}\mathrm{C}$ and $^{7}\mathrm{Be}$ produced via $\mathrm{^{11}B(p,n)^{11}C}$ and $\mathrm{^{10}B(p,\alpha)^{7}Be}$, and yields the absolute number

arXiv:2605.08103v1 Announce Type: new Abstract: High-entropy alloys (HEAs) have attracted growing attention for their exceptional mechanical and thermal properties arising from complex atomic configurations. In this paper, we propose crystal fractional graph neural network for predicting the energy of high-entropy alloys by explicitly integrating both local atomic environments and global compositional information. The model consists of three components: a crystal graph neural network, which employs graph attention network layers to learn local interactions among 16 on-site atoms within the crystal lattice; fractional neural network, a fully connected network that embeds the global fraction of constituent elements; and feature fusion neural network, which fuses the outputs of the two submodels to predict the total crystal energy. We train the model on a dataset of 1,049 crystal structures and validate it on 198 quaternary structures, optimizing all hyperparameters via Optuna. Our

arXiv:2605.09600v1 Announce Type: cross Abstract: Accurate skin lesion segmentation is vital for dermoscopic Computer-Aided Diagnosis. However, visual ambiguity and morphological irregularity often defeat spatial modeling, necessitating multi-domain architectures. Existing paradigms frequently overlook the active use of prediction uncertainty, leading to deterministic frameworks that suffer from blind cross-domain fusion and overfit to label noise. To address these issues, we propose the Uncertainty-Guided Dual-Domain Network (UGDD-Net). UGDD-Net introduces a novel "Glance-and-Gaze" mechanism to transform uncertainty into an active guiding signal. Specifically, the Uncertainty-Guided Bi-directional Feature Fusion (UGBFF) module uses pixel-level uncertainty to modulate spatial-spectral interactions. The Uncertainty-Guided Graph Refinement (UGGR) module constructs a topology-aware graph to propagate reliable semantic consensus and refine uncertain nodes. Finally, the

arXiv:2605.08103v1 Announce Type: cross Abstract: High-entropy alloys (HEAs) have attracted growing attention for their exceptional mechanical and thermal properties arising from complex atomic configurations. In this paper, we propose crystal fractional graph neural network for predicting the energy of high-entropy alloys by explicitly integrating both local atomic environments and global compositional information. The model consists of three components: a crystal graph neural network, which employs graph attention network layers to learn local interactions among 16 on-site atoms within the crystal lattice; fractional neural network, a fully connected network that embeds the global fraction of constituent elements; and feature fusion neural network, which fuses the outputs of the two submodels to predict the total crystal energy. We train the model on a dataset of 1,049 crystal structures and validate it on 198 quaternary structures, optimizing all hyperparameters via Optuna. Our

arXiv:2605.10780v1 Announce Type: new Abstract: Representation autoencoders that reuse frozen pretrained vision encoders as visual tokenizers have achieved strong reconstruction and generation quality. However, existing methods universally extract features from only the last encoder layer, discarding the rich hierarchical information distributed across intermediate layers. We show that low-level visual details survive in the last layer merely as attenuated residuals after multiple layers of semantic abstraction, and that explicitly fusing multi-layer features can substantially recover this lost information. We propose DRoRAE (Depth-Routed Representation AutoEncoder), a lightweight fusion module that adaptively aggregates all encoder layers via energy-constrained routing and incremental correction, producing an enriched latent compatible with a frozen pretrained decoder. A three-phase decoupled training strategy first learns the fusion under the implicit distributional constraint of

arXiv:2605.10510v1 Announce Type: new Abstract: Biomedical knowledge graphs are increasingly large, dynamic, and multimodal, driven by rapid advances in biotechnology such as high-throughput sequencing. Machine learning models can infer previously unobserved biomedical relationships and characterize biomedical entities in these graphs, but existing knowledge graph embedding methods and their continual learning extensions either assume static graph structure or fail to exploit multimodal information under evolving data distributions. They also apply uniform regularization across all model parameters, ignoring that different modalities may exhibit distinct forgetting dynamics as the graph evolves. We propose the Continual Multimodal Knowledge Graph Learner (CMKL), a CL framework for biomedical KGs that natively encodes structure, text, and molecules, fuses them through a Mixture-of-Experts (MoE) router, and protects previously learned knowledge with standard EWC regularization and a

arXiv:2605.07047v1 Announce Type: new Abstract: This paper introduces the MAESTRO workflow, that enables the coupling of the PORTALS framework [P. Rodriguez-Fernandez et al, Nucl. Fusion 2024] with external solvers for the plasma equilibrium, pedestal physics, divertor constraints and heating. The surrogate-based optimization nature of the transport solver is ideally suited for external coupling, allowing efficient steady-state predictions of plasma profiles with full physics models. Improvements in the surrogate modeling of quasilinear transport models with PORTALS are presented, which enable the efficient handling of discontinuities in the transport fluxes that can arise from numerical issues or physical instabilities with extreme stiffness. The combination of physics-informed methods and advanced numerical techniques allows the MAESTRO workflow to provide accurate and efficient predictions of steady-state plasma profiles, which are critical for fusion reactor design and

arXiv:2605.06745v1 Announce Type: new Abstract: Computational complexity and storage requirements are crucial factors influencing the performance and efficiency of convolutional neural networks (CNNs) in resource-constrained environments. This paper presents a high-performance embedded target detection system based on FPGA and YOLOv3-Tiny, specifically designed for embedded artificial intelligence applications. By integrating lightweight CNN optimization techniques with hardware accelerator design, significant improvements are made in both computational efficiency and resource utilization. Key optimizations, including low-bit quantization, batch normalization fusion, and table lookup mapping, reduce model parameters and computational complexity. Additionally, an FPGA hardware accelerator with a pipelined architecture is developed to enhance the efficiency of convolution operations while minimizing off-chip data transmission through modular design and on-chip cache optimization. On the

arXiv:2605.06745v1 Announce Type: cross Abstract: Computational complexity and storage requirements are crucial factors influencing the performance and efficiency of convolutional neural networks (CNNs) in resource-constrained environments. This paper presents a high-performance embedded target detection system based on FPGA and YOLOv3-Tiny, specifically designed for embedded artificial intelligence applications. By integrating lightweight CNN optimization techniques with hardware accelerator design, significant improvements are made in both computational efficiency and resource utilization. Key optimizations, including low-bit quantization, batch normalization fusion, and table lookup mapping, reduce model parameters and computational complexity. Additionally, an FPGA hardware accelerator with a pipelined architecture is developed to enhance the efficiency of convolution operations while minimizing off-chip data transmission through modular design and on-chip cache optimization. On

arXiv:2605.07314v1 Announce Type: new Abstract: Knowledge Graphs (KGs) have proven highly effective for recommendation systems by capturing latent item relationships, while recent integration of Large Language Models (LLMs) has further enhanced semantic understanding and addressed knowledge sparsity issues. Nevertheless, current KG-and-LLM-based methods still face three main limitations: 1) inadequate modeling of implicit semantic relationships beyond explicit KG links; 2) suboptimal single-channel fusion of ID and LLM embeddings, which often leads to signal interference and blurred representations; and 3) insufficient consideration of user-item interaction frequency variations in recommendation strategies. To address these challenges, we propose the Dual-Channel Graph Learning (DCGL) framework, featuring three key innovations: 1) a dual-channel architecture that structurally decouples rich semantic information from user behavioral patterns, preventing early interference; 2) a

arXiv:2605.06911v1 Announce Type: new Abstract: Subseasonal-to-seasonal (S2S) temperature forecasts, spanning several weeks to a few months, are critically needed in agriculture practice, energy planning, and extreme-weather induced risk management, yet their reliability varies substantially across seasons and regions. Forecast skill is often attributed primarily to lead time, but this perspective does not fully explain the spatiotemporal patterns of predictability. Here we show that S2S predictability is organized across interacting temporal components, spatial heterogeneity, and large-scale pattern coherence, and that this structure can be explicitly characterized and exploited. We develop a dual-scale learning framework that separates calendar-aligned historical climate context from lead-time matched recent weather evolution, combining them through spatially adaptive fusion to enable stable temperature forecasts across the 30 to 90-day window. The learned fusion weights reveal that

arXiv:2605.06894v1 Announce Type: new Abstract: Machine learning (ML) in real-world systems must contend with concept drift, adversarial actors, and a spectrum of potential features with varying costs and benefits. Malware naturally exhibits all of these complexities, but for the same reason, it is challenging to curate and organize data to study these factors. We present McNdroid, to our knowledge the largest longitudinal multimodal Android malware benchmark for malware detection and drift analysis. McNdroid spans 2013--2025, excluding 2015, and represents each application with three aligned modalities--static features from manifests and smali code, dynamic behavioral features from sandbox execution, and graph-based features from function-call graphs. Using temporally separated splits, we evaluate standard ML and deep-learning detectors across increasing train--test time gaps. Results show clear temporal degradation, while multimodal fusion outperforms the best single modality across

arXiv:2605.06509v1 Announce Type: new Abstract: Video diffusion models perform well in short-video synthesis, but their training-free extension to long videos often suffers from content drift, temporal inconsistency, and over-smoothed dynamics. Existing methods improve temporal consistency by combining a global branch with a local branch, but they often further decompose appearance consistency and temporal dynamics within each branch using predefined criteria. This assignment is unreliable when appearance and action progression are tightly coupled, such as in camera motion and sequential motion. We analyze the video temporal extension issue from a singular-spectrum perspective and show that enlarged self-attention windows induce spectral concentration: spectral energy becomes dominated by a few low-rank singular directions, preserving coarse structure but suppressing high-rank spatial details and motion-rich temporal variations. To mitigate this problem, we propose FreeSpec, a