Computer Vision
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Tuesday 10 November
Predicting gene expression directly from hematoxylin and eosin (H&E) images offers a scalable and cost-effective approach to deriving spatial molecular insights from routine histology. Pathology foundation models, typically pretrained using tile-level objectives, may not adequately capture the fine-grained cellular features required for single-cell molecular prediction. In this work, we systematically evaluate and establish domain-informed and task-specific strategies for adapting pathology foundation models to single-cell spatial gene expression prediction. Using paired H&E and Xenium 5K data, our approach improves mean gene-wise Pearson correlation by 2.67x over the standard zero-shot baseline and achieves state-of-the-art performance on this task. We demonstrate strong gene-detection performance across the full gene panel (AUROC = 0.800) and importantly identify that predictive performance varies substantially across biological contexts, with our model achieving its highest Pearson correlation for spatially variable genes (r = 0.643). Our findings demonstrate that, despite the challenges of sparse single-cell measurements, H&E-derived representations can recover biologically meaningful spatial expression patterns, highlighting both the opportunities and limitations of morphology-based molecular inference at cellular resolution.
Functional Analysis
The Fermat–Torricelli problem under isosceles orthogonality-type conditions
The location problem studied by Weber \cite{Weber} leads to the Fermat–Torricelli problem of minimizing sums of distances. Benítez, Fernández and Soriano \cite{BFS} proved that a real normed space of dimension at least three is an inner-product space if and only if every triple of points has a Fermat–Torricelli median in its convex hull. Pakhrou \cite{Pakhrou} showed that it suffices to test equilateral triangles. We establish a characterization using triangles $\{0,x,y\}$ satisfying \[ \|x+y\|=\lambda\|x-y\|,\qquad \forall\,\lambda>2/3, \] with the parameter fixed in each characterization. The case of parameter one gives an isosceles orthogonality criterion. The proof combines radial rescaling with the transfer of planar minimality to the ambient space, without completeness or regularity assumptions on the norm.
Read the full abstract
Machine Learning
CP-MLPs: A Tensor-Rank Theory of Tied and Untied MLP Blocks
Transformer feed-forward blocks combine several design choices: squared activations, gating, and tied versus untied parameter sharing, whose separate contributions to expressivity remain poorly understood. We introduce CP-MLPs, a tensor framework in which each multiplicative unit contributes a rank-1 interaction with three roles: detecting an input feature, reading a value, and writing to the output. Tied gated blocks such as ReLU$^2$ force the detector and value roles to share the same direction; untied gated blocks such as ReGLU and SwiGLU let them differ. We show that this distinction yields exact, degree-wise tensor-rank separations: untied detector-value units are strictly more width-efficient for indefinite quadratic interactions and role-asymmetric monomials, while tied units are structurally matched to symmetric pure-power targets. The width advantage of untying therefore depends on the target structure. Experiments show loss plateaus below the exact-representation width, but these plateaus alone do not establish a positive approximation gap: arbitrarily accurate approximation is governed by border rank. In trained gated MLPs, untied blocks retain an advantage in selected comparisons at matched input-projection budget. Code is available at https://github.com/rarefin/CP-MLPs.
Read the full abstract
Machine Learning
MouldIt: Data Efficiency of Multi-Task Temporal Representation Learning in Injection Moulding
Smart industrial solutions that can address multiple tasks require models that understand the underlying processes. We examine representation learning for modeling the injection moulding process as an example of such solutions. We consider its data efficiency and multi-modal benefit compared to classical machine learning alternatives. To this end, we synthesize a process-informed dataset based on a real reference cycle of an Erlanger Träger part. We then introduce MouldIt, a temporal convolution and attention based variational autoencoder that reconstructs in-mould signals, while classifying parts as acceptable or as one of our synthetic failure modes. We evaluate MouldIt on two tasks, known-class classification and unknown-class detection, across nine dataset sizes in single-modal and multi-modal settings. We compare it with an accept-all baseline and the classical methods k-nearest neighbors, logistic regression, random forests, and principal component analysis, which we adapt to support both tasks. MouldIt outperforms classical methods across both tasks from 4000 cycles in the single-modal setting and from 2000 cycles in the multi-modal setting. It achieves part sorting losses of 6.0 $\pm$ 0.7% and 1.0 $\pm$ 0.3% in the single-modal and multi-modal settings respectively, ranking first across all baselines and a combination of principal component analysis and random forests.
Read the full abstract
Optimization and Control
Robustness of controllability scores for stable linear network systems
Controllability scores quantify node importance in linear network systems by optimizing the allocation of input strength. Although the volumetric controllability score (VCS) and average-energy controllability score (AECS) are uniquely defined for Hurwitz systems, their sensitivity to perturbations of the system matrix remains unclear. In this paper, we investigate the robustness of these scores to such perturbations. We first prove that both scores are locally Lipschitz continuous in the system matrix and establish a robustness bound for Hurwitz comparison matrices. We then characterize the directional derivative of the optimal allocation along a differentiable matrix path as the unique solution of a critical-cone quadratic program, whose linear term is computed using differentiated Lyapunov equations. For single-edge weakening, the norm of the resulting response vector provides a robustness-based measure of edge importance. Numerical experiments show that the response, and hence the edge importance, depends on both the criterion and the dynamics, not only on the network topology.
Read the full abstract
Hardware Architecture
Tri-Stack-PIM: A Heterogeneous Multi-Die 3D-Integrated Processing-in-Memory Architecture for Efficient LLM Inference
Large language model (LLM) inference during autoregressive decoding is fundamentally memory-bandwidth-bound. Existing three-dimensional (3D) stacked processing-in-memory (PIM) architectures alleviate the off-chip bandwidth wall, yet remain constrained by several limitations: physical layer (PHY), controllers, and direct memory access (DMA) engines compete with arithmetic units for advanced-node base-die area; hybrid-bonding keep-out zones fragment the base-die floorplan and impede physical design; compute tiles built in a dynamic random access memory (DRAM) process are restricted to low logic density, capping achievable throughput; and nonlinear operations offloaded to the host reintroduce a communication bottleneck. We present Tri-Stack-PIM, a heterogeneous three-die architecture that resolves these tensions by dedicating a top DRAM die to storage, a middle buffer die on a mature process node to PHYs, DMA engines, and near-memory compute units, and a bottom base die on an advanced node exclusively to single-instruction-multiple-data (SIMD) multiply-accumulate (MAC) arrays. We further introduce a Unified Buffer-Aware Ready-based (UBAR) hardware out-of-order micro-operation scheduler that enforces cross-die dependency correctness at runtime while maximizing issue-slot utilization through a multi-term priority function, recovering up to 90% of the cycles that an in-order policy would forfeit. A joint architecture-dataflow co-exploration identifies Pareto-optimal configurations that reduce the energy-delay product (EDP) by 33-66x over naive parameter selection. Evaluated across multi-head, grouped-query, and multi-query attention (MHA, GQA, MQA) mechanisms with batch sizes 1-32, Tri-Stack-PIM achieves geometric-mean speedups of 4.95x, 2.57x, and 1.68x across all three attention mechanisms and batch sizes over a graphics processing unit (GPU) baseline, AttAcc, and H2-LLM, respectively.
Read the full abstract
High Energy Physics (Experiment)
Correlations between $\ensuremath{\mit{\Upsilon}(n\text{S})}\xspace$ production and the underlying event in $\ensuremath{p\!p}\xspace$ collisions with the ATLAS detector
Recent studies from various experiments at the LHC and RHIC suggest that small collision systems exhibit quark-gluon plasma (QGP) like signatures that were typically observed in large systems. While QGP like signatures have been observed in the soft sector, strong constraints on possible partonic energy loss were established in the hard sector. At the same time, there are multiple pieces of evidence pointing to correlations between hard and soft particle production in proton-proton collisions. The correlations between hard and soft particle production in $pp$ collisions are studied in events with $\Upsilon$ mesons, which provide a valuable hard probe to study such effects. The multiplicity and distributions of kinematic variables of charged particles produced in association with an $\Upsilon$ meson are measured using 139 fb$^{-1}$ of data at $\ensuremath{\sqrt{s}}\xspace = 13$~TeV collected by the ATLAS experiment at the LHC. The measurement is performed separately for the first three $\Upsilon$(nS) states in intervals of their transverse momentum. At low transverse momentum, the associated charged-particle multiplicity is observed to be systematically lower in collisions where $\Upsilon$(2S) or $\Upsilon$(3S) is produced compared to $\Upsilon$(1S), with the difference decreasing as the $\Upsilon$ transverse momentum increases.
Read the full abstract
Instrumentation and Detectors
Investigation of the timing performance of double-sided $p$-type 3D silicon sensors produced by CNM
We investigate the timing performance and spatial uniformity of double-sided $p$-type 3D silicon sensors produced by the Centro Nacional de Microelectrónica (CNM). Three sensor layouts are studied: hexagonal and square geometries with an inter-electrode spacing of \SI{30}{\micro\meter}, and a hexagonal geometry with an inter-electrode spacing of \SI{50}{\micro\meter}. The sensors are characterized using a \SI{1064}{\nano\meter} infrared laser and a custom two-stage wide-band readout. Measurements with a $^{90}$Sr source are used to establish a minimum-ionizing-particle-like reference for the laser signal. The timing resolution improves with increasing bias voltage for all three layouts. Two-dimensional laser scans are used to study the position dependence of the timing response and to quantify its uniformity. Among the investigated devices, the square \SI{30}{\micro\meter} geometry shows the best overall timing performance at the highest measured bias voltage.
Read the full abstract
Cryptography and Security
MOAI: Module-Optimizing Architecture for non-interactive secure transformer Inference
Privacy concerns have been raised in Large Language Models (LLM) inference when models are deployed in Cloud Service Providers (CSP). Homomorphic encryption (HE) offers a promising solution by enabling secure inference directly over encrypted inputs. However, the high computational overhead of HE remains a major bottleneck. To address this challenge, we propose MOAI, an efficient HE-based, non-interactive framework for secure transformer inference. MOAI gains significant efficiency improvement from: (1) a novel evaluation flow that combines column and diagonal packing with consistent strategies across all layers, eliminating expensive format conversions. (2) rotation-free algorithms for Softmax and LayerNorm that significantly reduce the number of costly HE rotations, removing 2,448 HE rotations in BERT-base inference. (3) Column packing removes rotations in plaintext-ciphertext matrix multiplications and interleaved batching further reduces the rotations in ciphertext-ciphertext matrix multiplications. MOAI uses at least 1.7x fewer HE rotations compared to the state-of-the-art works across all matrix multiplications of BERT-base. As a result, we achieve a $52.8 \%$ reduction in evaluation time compared to the state-of-the-art HE-based non-interactive secure transformer inference, THOR (Moon et al., CCS 25). We then apply MOAI on the Powerformer's framework and achieve a $55.7 \%$ reduction in evaluation time compared to Powerformer (Park et al., ACL 25), which approximates Softmax and LayerNorm with simpler functions in transformer and proposes HE-based non-interactive transformer inference. We report an amortized time of 2.36 minutes per input on a single GPU environment. We show the extendibility by applying MOAI in LLaMA-3-8B. Our implementation is publicly available as open source.
Read the full abstract
Human-Computer Interaction
EyeQ: Exploring Corneal Reflection for Mobile Context Sensing
What if a single face-facing camera could sense your surroundings—through your eyes? We introduce EyeQ, which explores corneal reflection as a modality for context-aware mobile interaction. Traditional vision-based context sensing relies on world-facing cameras that demand explicit aiming or dual-camera setups; EyeQ instead shows that an off-the-shelf VLM can read scene context from corneal reflections captured with a smartphone camera. To recover usable detail from mobile corneal crops, EyeQ pairs on-device iris-guided capture with server-side AI enhancement, then leverages VLMs to extract scene types, names, and objects from warped, unfamiliar eye-image sequences. In an N=10 study, we characterize how capture conditions shape reflection fidelity, evaluate what VLMs detect, misclassify, and hallucinate, and demonstrate interaction opportunities, including detecting a steering wheel to distinguish drivers from passengers. This work reframes the eye in HCI—from a target for gaze estimation to a passive mirror of one's surroundings—and examines the privacy tensions it raises.
Read the full abstract
High Energy Physics (Theory)
Quantum Langlands correspondence for Painlevé II
We construct an irregular KZ-BPZ correspondence for quantum Painlevé II from a rank-three Wakimoto-Whittaker module. The irregular KZ equation and the quantum isomonodromy equation form a joint affine system whose Drinfeld-Sokolov reduction yields scalar BPZ equations. We obtain matrix-integral solutions of this system for one fundamental insertion. An exact quantum Langlands duality exchanges the fundamental insertion with the zero of the dual Whittaker differential and allows affine solutions to be reconstructed from the BPZ equations. This duality interchanges the classical and Nekrasov-Shatashvili limits, which yield the Painlevé II Lax connection and an irregular Gaudin oper, respectively.
Read the full abstract
Computer Vision
picoEmergencyNet : Attention-Enhanced Ultra-Small Deep Learning Model for Aerial Image Classification in Emergency Monitoring
Real-time disaster recognition from aerial imagery requires deep learning models that operate under strict memory, compute, and latency constraints. While recent approaches achieve high accuracy, their computational cost limits deployment on resource-constrained UAV platforms. This work introduces picoEmergencyNet, an ultra-lightweight convolutional neural network designed for aerial image disaster classification. The proposed architecture combines depthwise dilated convolutions with efficient channel attention and channel shuffling to enable multi-scale feature extraction at minimal computational cost. picoEmergencyNet has 23k parameters and 7M FLOPs, requiring only 0.1 MB of model storage. Evaluated on the AIDER and AIDERv2 disaster datasets, it achieves up to 96.1% accuracy, matching or exceeding state-of-the-art lightweight models that are orders of magnitude larger. Deployment experiments on the NVIDIA Jetson AGX Orin show that, without hardware-specific optimizations, picoEmergencyNet achieves 258 FPS on GPU and 46 FPS on CPU with a 7x lower activation memory footprint compared to competing models. These results demonstrate that careful architectural specialization, rather than model scaling, enables highly efficient and accurate TinyML-ready disaster recognition for real-time UAV-based emergency monitoring.
Read the full abstract
Neurons and Cognition
Human and Machine Annotations Predict Brain Responses to Naturalistic Movies Across Development
Naturalistic movie watching provides a powerful framework for studying dynamic brain function, but linking movie content to neural responses typically relies on labor-intensive human annotation. We investigated whether multimodal large language models (MLLMs) can provide scalable socio-affective and semantic annotations for naturalistic fMRI. We compared MLLM-generated theory-of-mind (ToM) and valence ratings with human ratings and with regional BOLD responses across three animated movies. For two of these movies, we combined machine-derived visual features, natural-language scene descriptions, and socio-affective ratings in an ROI-wise encoding model of fMRI data from 66 participants aged 5–22 years. MLLM ratings partially tracked human judgments, but their neural correspondence differed by construct and model: ToM showed the most consistent human–machine overlap in visual cortex, whereas valence correspondence was largely movie dependent. In the encoding model, visual features yielded the strongest and most widespread prediction, whereas scene descriptions were more spatially selective, concentrated in dorsal-attention and visual regions and extending into right lateral temporal cortex. Age-group comparisons suggested feature-specific developmental differences: scene-description encoding in dorsal-attention cortex was higher in the oldest group, whereas visual-feature encoding in salience/ventral-attention, limbic, and control networks peaked in the middle group. Together, these findings indicate that MLLM-generated annotations capture meaningful aspects of human judgments and neural responses, supporting MLLMs as a scalable complement to, rather than a replacement for, human annotation in naturalistic neuroimaging.
Read the full abstract
Computation and Language
An Interpretable Stylometric Framework for Early Chinese Buddhist Translator Styles: A Case Study on the Attribution of Disputed Works of An Shigao
Attributing early Chinese Buddhist translations requires evidence that scholars can trace back to the texts. We develop an interpretable stylometric framework that uses features defined or reviewed by Buddhist philologists, including function characters, sentence formulas, sentence-final characters, sentence length, and part-of-speech n-grams. We construct stylistic profiles from 31 authenticated translations by seven translators from the Eastern Han through the Western Jin, then compare eight disputed texts with An Shigao's profile. Two texts, T1557 and T0101, show the strongest stylistic similarity to the authenticated An Shigao corpus, with scores of 0.783 and 0.687, respectively. The attributions agree with those reported by a published BERT-based approach on the same set of disputed texts. The framework requires no task-specific training and allows scholars to inspect which linguistic features support each proposed attribution.
Read the full abstract
Data Structures and Algorithms
Min-CSPs on Complete Instances III: Nearly Optimal Characterizations of Approximation for Symmetric CSPs
This paper studies complete Min-$k$-CSPs: an instance on $n$ variables has one nontrivial constraint for every $k$-set of variables (so $\binom{n}{k}$ constraints in total), and the goal is to find an assignment that minimizes the number of unsatisfied constraints. They are motivated by the connections to machine learning and data science, as well as the possibility of being testbeds for new algorithmic techniques. Previous recent results presented polynomial-time algorithms for the decision version of every complete Boolean $k$-CSP, which is a necessary condition for any multiplicative approximations for the minimization version, but nontrivial approximation results have been limited to fragile CSPs, 2-SAT, and NAE-3-SAT, leaving the approximability of most complete Boolean Min-CSPs open, including complete Min-$k$-SAT. In this paper, we resolve this question, up to nearly matching factors, for the natural class of symmetric Boolean predicates. Given $A \subseteq \{0, \dots, k\}$, Min-CSP($A$) is a Min-$k$-CSP where each constraint is specified by $k$ literals and is satisfied if the number of true literals belongs to $A$. Let $w(A)$ be the maximum number of consecutive integers in $A$. Our main results for complete Min-CSP($A$) are (1) a QPTAS for $w(A) \leq 2$ and (2) an $\widetilde{O}(n^{w(A)-3})$-approximation for $w(A) \geq 3$, as well as almost matching hardness results for the latter. Our QPTAS for $w(A)=2$ significantly improves the previous $O(1)$-approximation for Min-2-SAT and the $\operatorname{polylog}(n)$-approximation for Min-NAE-3-SAT, as well as extends the PTAS for fragile CSPs (i.e., $w(A) = 1$).
Read the full abstract
Robotics
An IoT Based Approach for Platform Independent Positioning Service
KAIST Indoor Localization System (KAILOS) was introduced in 2014 and became one of the world-first complete indoor positioning solutions. It is mainly based on the WiFi signal to estimate location and provide positioning service via mobile application that has many restrictions by the Operating System. This study proposes an extended architecture of KAILOS to make use of the IoT device for removing the barriers of the mobile-app-based approach and conveniently providing indoor positioning service. Design of the IoT device that contains hardware and software is also presented. Via this study, we prove that the IoT-based approach is promising and can be a crucial feature to widen applications of indoor localization services.
Read the full abstract
Signal Processing
Stability Analysis of Single- and Multi-Snapshot ESPRIT
Spectral estimation seeks to recover unknown frequencies from $T$ snapshots, each consisting of $\Omega+1$ noisy Fourier measurements. Super-resolution refers to the recovery of frequencies whose minimum separation $\Delta$ is smaller than $1/\Omega$. We establish state-of-the-art stability bounds for single- and multi-snapshot ESPRIT when the frequencies form well-separated clumps, with the largest clump containing $n^\star$ frequencies. In particular, for single-snapshot ESPRIT, our error bound scales as $$ \mathcal{O}(\Omega^{-1}\mathrm{SRF}^{2n^\star-2}\cdot\mathrm{Noise}), $$ where $\mathrm{SRF}:=(\Omega\Delta)^{-1}$ denotes the super-resolution factor, provided that the noise level satisfies $\mathrm{Noise}\lesssim(\Omega\Delta)^{2n^\star-2}$. In comparison with the minimax bounds [arXiv:1904.09186] and the computational resolution limits [arXiv:2003.02917, arXiv:2211.15208], our result shows that single-snapshot ESPRIT achieves the optimal stability scaling in the super-resolution regime. For multi-snapshot ESPRIT, we establish error bounds under both deterministic noise and Gaussian white noise, attaining the best-known SRF dependence under weaker conditions. Numerical experiments corroborate the theoretical error bounds.
Read the full abstract
Theoretical Economics
Herding with Multiple States: When the Prior Matters
In sequential observational learning with more than two states, the same information structure and preferences can yield learning for some full-support priors and herding for others. This disproves a conjecture of Kartik, Lee, Liu, and Rappoport (2024, working-paper version). In the counterexample, the cascade region has nonempty interior, yet learning occurs for generic full-support priors outside it. Focusing on information and preferences, we introduce two necessary conditions for learning. Escapability requires that signals can provide arbitrarily strong evidence that any given action is ex post suboptimal; pairwise escapability is its pairwise variant. If escapability fails, herding occurs for some initial prior; if pairwise escapability fails, it occurs for every full-support prior. For a class of environments with interval optimality and monotone likelihood ratios, escapability completely characterizes learning and herding independently of the full-support initial prior. With location-shift signals, this characterization reduces to pairwise distinguishability, a generalization of unbounded beliefs (Smith and Sørensen, 2000).
Read the full abstract
Functional Analysis
On Isomorphic Embeddings of Noncommutative Lorentz Spaces
It is shown that for $0<p_1,p_2,q_1,q_2<\infty$, the Lorentz operator ideal $C_{p_1,q_1}$ embeds isomorphically into $C_{p_2,q_2}$ if and only if (1) $p_1=q_1=2$ or (2) $p_1=p_2$ and $q_1=q_2$, which is in contrast to the classical result that $\ell_{q}$ embeds into $\ell_{p,q}$ as a (complemented) subspace. We also study the isomorphic embeddings between operator ideals generated by the weighted Lorentz sequence spaces $d(w,p)$ ($0<p<\infty$). We characterize those (commutative/noncommutative) Lorentz function spaces $L_{W}^p$ that are isomorphic to subspaces of noncommutative $L_p$-spaces for $1\leq p<2$ over some noncommutative probability spaces, which is a noncommutative version of Schütt's theorem.
Read the full abstract
Quantum Physics
Machine-learning decoders for fusion-based quantum computing in real time
Fault-tolerant quantum computation requires decoders that combine accuracy, throughput, low latency, and scalability to large numbers of physical qubits. In photonic fusion-based architectures, erasures from photon loss and probabilistic fusion measurements are a dominant error mechanism, motivating decoders that remain accurate at high erasure rates. We introduce PRISM, a transformer-based machine-learning decoder designed to exploit the known locations of erasures in protocols subject to both Pauli and erasure errors. We benchmark PRISM across a wide range of fault-tolerant protocols based on surface codes, color codes, and quantum low-density parity-check codes, with distances up to 24. Across these benchmarks, PRISM matches the logical error rates of integer-programming-based decoders while being significantly faster, including at error rates up to 50% of threshold. We further optimize the decoder and its real-time hardware implementation to reduce latency while maintaining accuracy. Our optimized implementation on an AMD VP1902 FPGA can decode logical blocks of distance 8 in both space and time, demonstrating a latency of 6.59 microseconds and an amortized decoding time of 1.56 microseconds per logical block is possible. These results show that machine-learning decoding can provide accurate, low-latency inference for erasure-dominated fault-tolerant quantum architectures.
Read the full abstract
High Energy Physics (Phenomenology)
Where does the neutrino Mass Ordering preference come from?
After the recent JUNO result, the comparison of the effective mass splittings measured in $\bar\nu_e$ disappearance at reactors and in $\nu_\mu$ disappearance has become one of the leading handles on the neutrino mass ordering. In this paper, we study which data sets drive the present preference for the ordering, combining JUNO and Daya Bay with $\nu_\mu$ disappearance data from NOvA, T2K, MINOS, and IceCube/DeepCore. While the combination disfavors the inverted ordering at $1.9\sigma$, this preference comes from a balance between several small preferences. In particular, the combinations of reactor data with T2K or with NOvA point to opposite orderings. The values of the mass splitting preferred by these two experiments differ by about $3.5\%$, comparable to the difference expected between the two orderings, and thus a shift of this size in a single experiment is enough to reverse the preference. Moreover, in the context of this combination of reactor and $\nu_\mu$ disappearance data, further JUNO data would not bring the rejection of the inverted ordering beyond about $3\sigma$, as the comparison is then limited by the precision of the $\nu_\mu$ disappearance experiment. Improving the $\nu_\mu$ disappearance measurements requires control of the neutrino energy reconstruction, and thus the modeling of neutrino-nucleus interaction physics, at the percent level.
Read the full abstract
Computer Vision
DepthGS: Depth-Guided Gaussian Splatting with Adaptive Regularization for High-Quality Surface Reconstruction
Recently, 3D Gaussian Splatting (3DGS) has emerged as an efficient representation for novel view synthesis, offering fast differentiable rendering and high visual fidelity, and has increasingly been extended to surface reconstruction. However, the relationship between depth information and reconstruction error remains insufficiently explored, limiting the principled design of geometric optimization strategies. To fill the above gap, we propose DepthGS, a depth-guided Gaussian splatting framework. We derive theoretical relations between depth discrepancies and reconstruction error under single-view and multi-view settings, and rigorously establish corresponding upper bounds on reconstruction error. The above analysis informs the design of adaptive regularization strategies in both single-view and multi-view optimization. More specifically, for single-view optimization, we propose a depth-difference-guided adaptive normal regularization strategy that modulates the normal constraint strength using a geometric reliability weight computed from local depth differences. For multi-view optimization, we introduce a cross-view depth consistency regularization strategy that promotes consistency between neighboring-view depth maps and the corresponding depth predictions obtained by transforming the current view's depth predictions into those neighboring views. Extensive experiments on DTU, Mip-NeRF360, and Tanks and Temples demonstrate that DepthGS achieves competitive performance in both surface reconstruction and novel view synthesis.
Read the full abstract
Programming Languages
Curriculum proof repair with learned counterexamples
Language models (LMs) have substantially lowered the barrier to applying formal verification for building reliable and secure software. Existing approaches commonly follow the verifier-in-the-loop paradigm, but the verifier often provides only coarse, binary pass/fail feedback, which offers little actionable signal on how the proof should be repaired. Recent work shows that counterexamples provide more actionable feedback by grounding proof failures in concrete program states, but generating useful ones at inference time requires costly iterative refinement. We present TopoVerus, an LM-based framework for Verus proof repair that internalizes counterexample-guided diagnosis as a model capability, producing useful counterexamples in a single pass. Our key insight is to train counterexample generation not as a standalone validation task, but as a data-driven diagnostic objective: a counterexample is useful when conditioning on it helps the model produce a safe, verifier-passing proof. Given a failed Verus proof and verifier error, TopoVerus generates a structured source-level counterexample, repair rationale, and repaired proof in one trajectory, optimized with reinforcement learning from Verus feedback and a proof-code safety gate. To guide the order of invariant repairs, TopoVerus introduces a curriculum with a topological graph reward based on invariant dependencies derived from inductiveness tests, which encourages counterexample generation to follow the same topological order. Across seven Verus benchmarks with 471 problems, TopoVerus achieves 61.6%/72.2% weighted-average Safe-Pass@1/Safe-Pass@3, outperforming the strongest baseline, Claude Sonnet 4.5, by 5.7/6.3 points. We also show that TopoVerus's learned counterexamples can directly augment the reasoning of existing LM-based proof repair approaches, improving their success rate by 64.3% compared to chain-of-thought diagnostics.
Read the full abstract
Computer Vision
FlashMotion: Ultra-Fast Long-Term 3D Motion Generation
3D human motion generation plays a crucial role in game development, extended reality (XR), digital humans, and embodied simulation. Beyond motion generation, practical pipelines include rigging, retargeting, and rendering, making ultra-fast long-term 3D motion generation without compromising quality valuable for real-time interactive applications. Despite promising progress, high-quality long-horizon motion generation under tight inference budgets remains challenging, owing to 2 coupled bottlenecks: (1) generative strategies that require many sampling steps for high-quality 3D motion generation and (2) model architectures with substantial per-step computational and memory costs for long sequences. We present FlashMotion, a co-designed framework for high-quality long-horizon 3D motion generation under a strict inference budget, covering text-to-motion, music-to-dance, and speech-to-gesture tasks. To enhance trajectory-level efficiency, we introduce a 3D motion-space MeanFlow formulation with kinematic and dynamic consistency constraints for few-step generation, together with an Adaptive Soft masking mechanism for temporally coherent motion editing. To improve per-step efficiency, we design a lightweight non-autoregressive BiMamba architecture with condition-guided state selection and AdaIN-based condition fusion mechanism, enhancing cross-modal alignment and generalization to longer motion sequences. Extensive experiments demonstrate that FlashMotion achieves overall state-of-the-art performance in both quality and efficiency across text-to-motion, speech-to-gesture, and music-to-dance generation. The project page is available at https://sun-happy-ykx.github.io/FlashMotion-Project/
Read the full abstract
Quantum Physics
Packet-wise multi-hop quantum routing of hybrid quantum-classical packets
The ability to dynamically route quantum information across multiple nodes is a key requirement for realizing versatile and scalable quantum networks. Quantum routers are devices to achieve packet-level routing beyond static point-to-point connections by handling hybrid packets combining classical headers and quantum payloads. Previously proposed methods have selectively accessed the classical and quantum components of the packets utilizing wavelength splitters and optical switches, and repacked the hybrid packets before transmission through the outgoing link with regenerating classical headers. This work reports a novel optical-fiber-based quantum router that minimizes the disturbance and latency on the hybrid packets through more transparent operation. We experimentally demonstrate double-hop routing via two cascaded routers to three distinct receivers without regeneration of the headers. As a practical application, quantum key distribution packets based on time-bin qubits and a BB84 protocol with a two-decoy-state method are successfully delivered to the three receivers achieving nonzero secret key rates. These results demonstrate the feasibility of extensive quantum networks interconnected by multiple-step router nodes.
Read the full abstract
Machine Learning
Multifidelity Operator Inference for Non-intrusive Model Reduction under Data Scarcity
Many-query analyses, in which a computational model is repeatedly evaluated, play an important role in uncertainty quantification and optimization for engineering design. However, many-query analyses can be prohibitively expensive when considering high-fidelity numerical models, especially those governed by partial differential equations (PDEs). Surrogate modeling replaces costly simulations with cheap approximations, enabling practical uncertainty quantification and design workflows. Machine learning (ML) approaches to surrogate modeling have gained significant traction in the development of surrogate models due to their non-intrusive nature; they do not require access to the underlying simulation code. The Operator Inference (OpInf) approach, in particular, fits reduced operators to state and corresponding time derivative data. However, high-fidelity training data is scarce due to limited computational budgets, and OpInf models trained on such scarce data often exhibit high variance, a critical liability in high-risk engineering design. We propose Multifidelity Operator Inference (MF-OpInf), which reduces this variance by supplementing scarce high-fidelity snapshots with abundant low-fidelity data drawn from coarser discretizations of the same problem. MF-OpInf handles data across varying fidelity levels by utilizing a multifidelity linear regression framework to estimate the reduced operators. Numerical experiments on model PDE problems demonstrate that MF-OpInf predicts PDE solution fields with higher accuracy compared to the high-fidelity-only OpInf with the same computational budget.
Read the full abstract
Statistics Theory
First-order asymptotic expansions for spectral convergence of normalized graph Laplacians constructed from samples on manifolds
Under i.i.d. uniform sampling from a closed Riemannian manifold embedded in Euclidean space, we derive first-order asymptotic expansions for the eigenvalues and eigenvectors of symmetrically normalized graph Laplacians constructed via Gaussian kernels and provide concentration inequalities for the remainder terms. As a direct consequence, we establish the corresponding weak convergence results. Unlike most related works, our analysis does not impose a multiplicity-one assumption on the eigenvalues, thereby extending the applicability of the framework to general manifolds with potentially nontrivial eigenvalue multiplicities.
Read the full abstract
Statistical Mechanics
Stochastic evolution of interacting directed strings
We study the stochastic dynamics of a system of $N$ polymer rings on an infinite cylinder with only excluded volume interaction, using simulations and also analytically. The strings can diffuse in the transverse direction by local Rouse-like moves where an up-kink changes to a down -kink, or vice versa. The motion is biased if the rates of flip are not equal. In the initial state, we start with a step initial condition: all the sites with a positive $y$ coordinate are empty, and the sites with $y \leq 0$ are occupied with a uniform density. We study the scaling form of the density profile in the hydrodynamic limit. The scaling form is a function of $y/\sqrt{t}$ in the unbiased case, and a function of $y/t$ in the biased case. In the unbiased case, in the steady state on a torus, the marginal probability distribution of the positions of the strings on a line with fixed $x$-coordinate is exactly same as that in the ground state of quantum-mechanical spin-less fermions on a line. The biased problem is equivalent to a growing surface in $2+1$ dimension, which is expected to be in the Kardar-Parisi-Zhang universality class.
Read the full abstract
Computer Vision
PUMA: Probability-Constrained Sequential Interaction for Forensic Localization
Image manipulation localization must jointly exploit semantic structure and noise-sensitive forensic evidence. However, the effects of spatial correspondence, matching quality, and interaction order are often conflated. Existing semantic-forensic fusion methods provide complementary representations, but do not explicitly preserve quality-mismatch information during spatial correspondence or adaptively select the order in which information is propagated across branches. We propose PUMA (Probability-constrained Unbalanced Matching and Aggregation), which combines spatially constrained unbalanced optimal transport with position-wise convex selection between two sequential operators through a shared positional interface. Before normalization, PUMA retains signed transport-mass residuals, constructs row-stochastic cross-branch and within-noise-branch operators, and updates noise features through a gated residual path. PUMA achieves strong localization results on four test sets, with scores on COVERAGE and NIST16 exceeding the published results of the representative methods listed in our comparison. The experiments support the utility of local matching states and adaptive sequential interaction for cross-source localization and several post-processing conditions.
Read the full abstract
Computer Vision
When to Start Denoising? Adaptive Denoising Entry for Generative Video Compression
Generative video compression offers a promising solution for compressing and reconstructing videos under extremely low bitrate constraints, among which the diffusion-based solutions still achieve the best performance by providing the powerful generative latent prior. However, their effectiveness critically depends on when the compressed latent is introduced into the diffusion process, which directly governs the trade-off between reconstruction fidelity and perceptual realism. To address this limitation, we identify denoising start-timestep selection as a key factor governing this trade-off and formulate it as a geometric optimization problem. Based on this formulation, we propose a novel adaptive diffusion entry method for generative video compression (ADE-GVC), the first timestep-aware generative video compression framework, featuring a lightweight predictor that estimates an adaptive denoising start point from the compressed latent. A temporal adapter further extracts compression-aware conditions and injects them into a pretrained video diffusion model, enabling the reconstruction of videos with both high fidelity and realism. More importantly, we observe a strong quantitative relationship between the predicted timesteps and the distortion between the original and compressed latents, providing empirical support for our optimization formulation. Extensive experiments demonstrate that ADE-GVC achieves the state-of-the-art performance on perceptual metrics and substantially improves reconstruction realism over existing generative video compression methods. Notably, our ADE-GVC achieves an average bitrate savings of 40.99% (on DISTS) and 95.24% (on PSNR) compared to the latest open-source baseline GLC-Video.