Artificial Intelligence
KV$^2$: A Self-Refining KV Cache
Johannes Wesch, Danni Liu and Jan Niehues
The
memory footprint of the key-value (KV) cache constrains the practical use of long-context models, and it dominates cost when one prefilled context must later serve many different queries. In this reusable setting, query-agnostic compression trades cost against quality: lightweight
estimators are cheap but less accurate, whereas full-context reconstruction scoring is more accurate yet reprocesses the entire
prompt. We introduce KV$^2$, a query-agnostic KV-cache compression method based on selective reconstruction. KV$^2$ first uses a lightweight proxy scorer to identify informative in-context
tokens, then reprocesses only this subset to compute final eviction scores. On RULER, Needle-in-a-Haystack, and LongBench, KV$^2$'s margin over baselines widens as the budget tightens: on RULER 16K at a 2% KV-cache budget it improves the average score over the next-best baseline by more than 40 percentage points, and on LongBench it attains the highest average across 2%-10% budgets at lower compression-stage runtime and peak
memory than full-context reconstruction. Reusable KV-cache compression thus does not require reprocessing the full context. Our code is available at https://anonymous.4open.science/r/KVsquared-0B97.
Read the full abstract ▾
Distributed Computing
Closing the Prediction Gap: Completing Machine Shape So That Predicted Time, Power, Energy, and Mapping Match What Real Hardware Does
Lenore Mullin and Gaetan Hains
A companion paper derives a weighted, communication free partition for heterogeneous, multi institution device ensembles from each device measured shape, rho_machine(d), validated on real hardware. Its experiments show rho_machine is incomplete: fp32 to fp16 speedup and power shifts are not fully indexed, and
memory capacity predictions rely on an assumed, not measured, reservation overhead. This paper closes that gap, indexing five unmeasured fields in the psi selection vocabulary:
memory capacity $M_i^{cap}$, power draw $P_i$, execution unit capability $X_d$, compiler directive configuration $C_d$, and
memory reservation overhead $R_d$. We introduce reached(d), a Boolean for whether compiled code dispatches to specialized hardware, explaining the NVIDIA fp16 power shift that hardware availability alone cannot predict. MoA fixes a computation independently of its target, so compiler directives get a two part admissibility criterion: select a real hardware quantity without discretionary reordering, and be deterministic, not compiler overridable. Across Intel, AMD, NVIDIA, OpenMP, OpenACC, and Open MPI, this admits only cache placement controls, deterministic vector or tile widths, and occupancy caps; heuristics like autotuning are excluded and absorbed into a residual epsilon$(k)_d$. We validate on five devices: A100, H100, V100, MI100, and Max 1550, measuring or establishing all five fields on each. Reachability proves real but partial, occupancy limits are device specific, and reservation overhead ranges about 0.004 to 0.030. A residual of negative 9.8 percent for the excluded Triton autotuner shows an unvalidated heuristic can act opposite a presumed penalty. These results close all five fields for every device in the companion ensemble, converting assumed behavior into measured, target specific quantities.
Read the full abstract ▾
Robotics
Beyond Reward Hacking: Proxy Divergence Across Four Layers of a Staged Humanoid Learning Pipeline
Arunabh Bora
A reinforcement-learning (RL) pipeline for a legged
robot is assembled from proxies. A reward stands in for intended behaviour, a curriculum gate stands in for competence, an evaluation statistic stands in for robustness, and a reference motion stands in for an achievable skill. The traditional view treats only the first of these as optimised against, and so locates specification failure (reward hacking) in the reward alone. I argue that all four are proxies in the same formal sense, that each has a characteristic divergence mechanism, and that each admits a reformulation that closes it. For every layer I state the traditional formulation, derive the condition under which it diverges from its target, and give the alternative: first-order (L1) costs where quadratic kernels are flat, peak and outcome statistics where curriculum gates average, gate reachability and information checks, deterministic and phase-desynchronised evaluation, curriculum state treated as part of the model, feasibility-first reference design with residual feed-forward, and function-preserving input widening that lets one policy grow instead of being retrained. The arguments are illustrated by measurements from one continuous lineage of a PPO policy for a simulated 1.91 m
humanoid, grown over four stages and 13,500 iterations on a single laptop GPU. Among them, a curriculum gate built on averaged error advanced at its rate limit on every check while the skill it gated was absent, and a batched push test whose synchronised resets aliased the gait phase ranked a 0.5 m/s push as more dangerous than a 2.0 m/s one.
Read the full abstract ▾
Computation and Language
Source Preference in the Wild: How LLM Agents Favor Items by Source, and How to Reduce It
Jonghyun Song, Haewon Park, Jeonghoon Shim et al. (+2 more)
As LLM
agents decide on users' behalf which product to buy, which hotel to book, or which paper to cite, a preference for items from certain sources (the sites or services they come from) shapes what users receive and which sources are selected. We study source preference in end-to-end search with 12
agent models across three domains. Comparing items from different sources that satisfy the same requirements at the same position, we find that each model prefers some sources and avoids others in every domain, largely agreeing on which. This preference can outweigh how well items satisfy the request: an item satisfying one requirement fewer is selected about two-thirds of the time when it comes from a preferred source and the better one from a dispreferred source, but almost never in the reverse case. The information identifying an item's source affects selection by itself: hiding it weakens the preference, and relabeling an item with a preferred source raises its selection rate. We test two routes to this preference: training that rewards better items can make a source a shortcut for requirement satisfaction, and missing information can trigger preconceptions about the source. Supplying missing information or a
prompt countering these preconceptions reduces source preference.
Read the full abstract ▾
All authors Jonghyun Song, Haewon Park, Jeonghoon Shim, Woojung Song, Yohan Jo
Computer Vision
Bridging Research and Practice: A Systematic Evaluation of Generalist and Dermatology-Specific Models in Clinical Skin Lesion Classification
Emanoel dos Santos, Kelvin Cunha, Rodrigo Mota et al. (+8 more)
The application of machine learning to dermatology has grown substantially in recent years, moving beyond proof-of-concept studies toward potential applications. However, clinical dermatology remains a challenging and still open problem. Diagnostic assessment is often ambiguous, and skin lesions exhibit high variability, compounded by differences in acquisition modality, device quality, and patient demographics. These factors hinder the development of robust models suitable for safe and equitable clinical use. To support translation into practice, it is essential to systematically evaluate how contemporary models generalize across heterogeneous data sources. In this work, we
benchmark a diverse set of architectures on recent dermatology
datasets, spanning dermoscopic images and smartphone-based clinical photographs. We assess the robustness of recent general-purpose and medical
vision-
language models, as well as foundation models, and compare them against task-specific dermatology classifiers, including
embedding-based approaches and convolutional neural networks. Our study provides an evaluation of model performance under distribution shifts, modality changes, and demographic variability. By quantifying the gap between current state-of-the-art models and the requirements of clinical deployment, we aim to contribute to the development of reliable, accessible, and clinically applicable AI systems for dermatology.
Read the full abstract ▾
All authors Emanoel dos Santos, Kelvin Cunha, Rodrigo Mota, Fabio Papais, Thales Bezerra, Natalia Lopes, Erico Medeiros, Shirley Cruz, Jessica Araujo, Paulo Borba, Tsang Ing Ren
Computer Vision
PocketSplat: Mobile Gaussian Reconstruction via World-Space Latent Allocatio
Wenzhi Guo, Xianda Chen, Dongxuan Chen et al. (+2 more)
Mobile Gaussian reconstruction must satisfy two requirements: the reconstruction model must execute within a device resource envelope, and the resulting Gaussian asset must expose a representation size suited to downstream mobile use. Existing feed-forward Gaussian reconstructors commonly decode dense, image-aligned candidates whose final cardinality is implicitly determined by the input resolution and number of views. We present PocketSplat, a feed-forward framework for budgeted mobile Gaussian asset construction. Given a prescribed output budget, PocketSplat organizes dense geometry-aware latent candidates in predicted world space, allocates exact integer capacity across local latent
cells, and decodes complete Gaussian attributes only for retained candidates.
Cell-conditioned latent fusion aggregates repeated multi-view evidence before decoding, while spatial responsibility decoding adapts Gaussian support after local sparsification. Experiments on DL3DV and out-of-distribution
benchmarks establish a strong quality–budget trade-off against feed-forward Gaussian reconstruction baselines. On Mip-NeRF 360, PocketSplat executes directly on a target iPhone and constructs compact, higher-quality Gaussian assets substantially faster than a deployable streamed MVSplat variant; native MVSplat and DepthSplat exceed the device
memory budget.
Read the full abstract ▾
All authors Wenzhi Guo, Xianda Chen, Dongxuan Chen, Guangchi Fang, Bing Wang
Numerical Analysis
Reversible Stratonovich perturbations for accelerating convergence to equilibrium of Langevin dynamics
Guillaume Louis, Grigorios A. Pavliotis and Gilles Vilmart
In this paper, we study the problem of sampling from a probability distribution using Langevin-type dynamics. We consider overdamped Langevin dynamics perturbed by adding a Stratonovich perturbation that preserves reversibility. We prove an optimal scaling for the Stratonovich perturbation applied to Gaussian target distributions and derive an
algorithm to construct such perturbation. Our theoretical results are supplemented by numerical experiments in which we compare our proposed method with nonreversible samplers and overdamped Langevin dynamics. Our numerical experiments demonstrate the efficiency of the approach beyond the quadratic case.
Read the full abstract ▾
Computation and Language
Not Until the Evidence Says So: Teaching LLM Investigators When to Close a Case
Tingzhu Bi, Ping Wang and Meng Ma
Accident, defect and outage investigations end with a decision that ordinary question answering never faces: whether the evidence gathered so far is enough to close the case. We study this decision for LLM investigators, which request evidence from a case file, revise their hypotheses, and either close the case with a conclusion grounded in what they read or leave it open and name what is missing. This judgment does not come with capability: an untrained 9B model overstates its evidence in 97% of its answers, and a frontier model that identifies the right cause in 84% of cases still overstates in 91% and closes 17 of the 41 cases whose official finding is "cause undetermined". Measuring it is also non-trivial: the source of a case largely predicts its label, and a rule that reads only the source reaches 83.0 balanced accuracy on our test cases. We therefore evaluate closure with three tests: closure accuracy, reported against this rule and within each source; evidence dependence, which removes the grounds of a conclusion and checks whether the model stops closing; and conclusion and gap quality, a judged checklist of what the model asserts and what it says is missing. We build Nautil, 731 audited cases from aviation, rail, maritime, chemical-safety and vehicle-defect reports and production server incidents, with teacher trajectories, an out-of-distribution test set and counterfactual evidence versions. Fine-tuning a 9B model on these trajectories makes its closures follow the evidence: removing the grounds lowers its closure rate by 26 points relative to a matched control, overstatement falls from 97% to 35%, and correct, non-overstated conclusions rise from 3% to 43%. Reinforcement learning that rewards only the closure decision then raises balanced accuracy from 69.2 to 83.3, on par with the teacher, and within-source accuracy from 60.4 to 74.1, at some cost in evidence dependence.
Read the full abstract ▾
Distributed Computing
Lightweight and Resource-Efficient Perception for Robotic Guide Dogs
Jinse Kwon, Yoojin Lim, Choonghan Lee et al. (+3 more)
Robotic guide dogs should understand their surroundings, objects, and potential risks. Prior research has focused on raw sensor data from cameras and 2D or 3D
LiDAR, which precisely measure distance points rather than provide a semantic understanding of the scene. While these physical measurements are effective for
robot-centric collision avoidance and
robot safety, they are not suitable for human-centric guidance. The system should recognize the type and relevance of obstacles and explain them, clearly and actionably, in terms of their spatial relation to the user. We present complete on-device perception modules that fuse a 360 camera and a 2D
LiDAR for reliable collision avoidance, with moving-object detection and tracking for human-centric guidance. Finally, in walking-impossible situations, a
vision–
language model delivers pathway explanations as a safety mechanism to reduce user anxiety. In experiments, verification of fused 360 camera–
LiDAR depth shows reliable near-range perception but inherent mid-range bias, while the system as a whole sustained real-time performance under 55 W. On the real-world
egocentric GuideDogQA
benchmark, our system achieved 83.8% accuracy, compared with 67.1% for GPT-4o. These results demonstrate that practical human-centric guidance with real-time on-device inference is feasible even on quadrupeds.
Read the full abstract ▾
All authors Jinse Kwon, Yoojin Lim, Choonghan Lee, Yongseung Yu, Yongin Kwon, Jemin Lee
Artificial Intelligence
Gains and Collapse in On-Policy Distillation:A Reinforcement Learning Perspective
Han Cui, Jianhao Yan, Yun Luo et al. (+3 more)
On-policy
distillation (OPD) has become an important approach to
language model post-training. However, despite its performance gains, OPD can also collapse into excessively long and repetitive generation, and the mechanism underlying these divergent outcomes remains poorly understood. We explain these outcomes through a reinforcement learning perspective: the teacher implicitly rewards student behaviors, even those it rarely exhibits itself. From this perspective, our experiments show that OPD improves performance without expanding the student's capabilities. When the implicit reward model is reliable, OPD makes correct responses easier to sample. In contrast, when the preference misaligns with quality, reward hacking happens: the implicit reward model amplifies overlong, repetitive student rollouts, even though it rarely generates such text itself. Guided by this diagnosis, we find that masking unhealthy responses during training and using SFT initialization can each effectively mitigate the collapse. Together, these findings show that OPD amplifies student behaviors favored by the teacher's implicit feedback, shifting the focus from how well the teacher generates to how reliably it evaluates student rollouts. Our code is available at https://github.com/HancCui/opd_hacking.
Read the full abstract ▾
All authors Han Cui, Jianhao Yan, Yun Luo, Hongbo Zhang, Zhizhang Fu, Yue Zhang
Numerical Analysis
Recursive Multilevel Decomposition of Pairwise Comparison Matrices: Algorithms and Inconsistency Diagnostics
Mert Çarboğa, Waldemar W. Koczkodaj and Yusuf Yayl{\i}
Positive reciprocal pairwise comparison
matrices are studied in entrywise logarithmic coordinates. Starting from the established consistent-residual split, we show that, for any fixed ordering, the residual
matrix admits a unique top-down decomposition into mutually Frobenius-orthogonal layers. The construction applies to every positive reciprocal PC
matrix of size $n \geq 3$, without additional spectral, sparsity, or genericity assumptions. An explicit active-
matrix recursion and an equivalent cumulative parameter-extraction procedure have computational complexities $O(n^3)$ and $O(n^2)$, respectively. The squared Frobenius norms of the layers define an ordering-conditioned residual-energy profile that provides structural information complementary to the
geometric-mean priorities and the Koczkodaj inconsistency indicator (Kii). Entrywise exponentiation yields an exact Hadamard-product reconstruction of the original reciprocal
matrix. Numerical experiments confirm agreement between the two computational procedures, exact reconstruction, mutual orthogonality, and numerical accuracy to machine precision.
Read the full abstract ▾
Cryptography and Security
Reversing the Clock: Layout-Aware Recovery of Design Intent from Clock Distribution Networks
Sascha Tommasone, Zehra Karadağ, Christopher Pawlowicz et al. (+6 more)
Hardware reverse engineering supports competitive analysis and hardware assurance by recovering information about an integrated
circuit (IC) from its physical implementation. While existing techniques primarily recover the gate-level netlist, which represents logical functionality, they often overlook physical design decisions such as placement, routing, delay insertion, and interconnect optimization. The clock distribution network encapsulates many of these decisions; however, no prior published work has recovered this network from a fabricated IC to infer design intent. We present a layout-aware methodology that recovers and analyzes the clock distribution network by integrating a recovered gate-level netlist with layout information extracted from scanning electron microscope imagery. Our four-phase pipeline recovers clock-tree
topology, buffering, gating and switching, interconnect delay, and crosstalk-mitigation measures. We demonstrate our methodology on a commercial 450 nm IC, recovering a global H-tree backbone with local X-tree-like branching, identifying an independent clock tree and global clock gating, quantifying latency, skew, and routing lengths, and confirming the absence of dedicated crosstalk mitigation. Together, these findings let us reason about the designer's intent. To encourage further research and support reproducibility, we release our clock tree recovery
algorithm as open source.
Read the full abstract ▾
All authors Sascha Tommasone, Zehra Karadağ, Christopher Pawlowicz, Michael Green, Bruno Machado Trindade, Eunsung Seo, Christof Paar, René Walendy, Steffen Becker
Computation and Language
Hindsight-Guided Rationale Distillation for Rare Disease Diagnosis
Aarav Singh, Animesh Pathak and Navyansh Singh
We study hindsight-guided
distillation for rare disease diagnosis on ZebraMap: a 1.5B student is fine-tuned on chain-of-thought traces from a 8B teacher that observes the ground-truth diagnosis during generation. Absolute accuracy remains low for all models - the task is hard at this scale - but within this ceiling a filtered variant (StudentF) achieves a small, statistically significant accuracy advantage over the teacher (p < 0.001), concentrated in better-represented diseases. The unfiltered student does not significantly outperform the teacher (p = 0.129), establishing that contamination filtering - not hindsight
distillation alone - drives the gain. The gap traces to an artifact we term GT
hallucination. Label-visible generation causes the teacher to embed "ground truth is X" phrases in its
reasoning chain; SFT copies the pattern. At inference, the unfiltered student reproduces the phrase in 33.9% of cases, with severe accuracy degradation when the hallucinated label is wrong. A regex filter removing these slots reduces contamination to near-zero, producing the observed gain - though the effect remains small. We precisely quantify this gain-cost tradeoff, document frequency-dependent knowledge transfer absent from the RL-trained teacher, and characterize a calibration gap that SFT does not close - identifying both as directions for future work.
Read the full abstract ▾
Multiagent Systems
FinNextAssist: Towards Professional Financial Deep Research Assistant
Xiangyu Li, Fengbin Zhu, Xuan Yao et al. (+9 more)
Deep Research (DR)
agents have demonstrated strong capabilities in complex, research-oriented tasks through autonomous planning, iterative
retrieval, multi-step
reasoning, and structured reporting. However, adapting DR
agents to finance introduces unique challenges: financial analysis demands the joint completion of heterogeneous sub-tasks spanning diverse data types, tools, and analytical workflows. We identify three key requirements for a professional financial DR
agent: integration of authoritative, heterogeneous financial data sources; specialized analytical tools and skills; and dedicated sub-
agents for domain-specific sub-tasks. Building on these principles, we propose FinNextAssist, an end-to-end deep research framework designed for professional financial analysis. FinNextAssist decomposes the research process into four stages: Task Planner, Evidence Compiler,
Reasoning Engine, and Report Assembler, and introduces two novel lightweight sub-
agents: TabAgent, for cross-
market financial table understanding, and HeteroAgent, for cross-modality heterogeneous financial data interpretation. Extensive experiments on FinDeepResearch, the Finance
Agent Benchmark, and FinTMMBench-Web show that FinNextAssist substantially outperforms both strong proprietary and open-source DR
agents, with ablation studies confirming the contribution of each component across diverse
markets and
languages.
Read the full abstract ▾
All authors Xiangyu Li, Fengbin Zhu, Xuan Yao, Siyu Liu, Xiaoluan Liu, Chao Wang, Huanbo Luan, Xiaofen Xing, Xiangmin Xu, Ke-Wei Huang, Richang Hong, Tat-Seng Chua
Systems and Control
SIS Epidemic Containment under Game-theoretic Rational Learning
Urmee Maitra, Keith Paarporn and Ceyhun Eksin
We consider a susceptible-infected-susceptible (SIS) epidemiological model in which individuals dynamically decide whether to follow social distancing or not. Driven by game-theoretic payoffs, their actions influence the spread of infection. Much of the existing work in this area considers individuals that are imitative learners, leading to the well-studied replicator equation. In this paper, we investigate how the dynamical outcomes may fundamentally shift when individuals are instead equipped with logit learning rules, i.e., when individuals possess bounded rationality. We analyze the local asymptotic stability of
equilibria that exist in different parameter regimes of the coupled dynamical system. We show that in regimes where the disease-free
equilibrium is unstable under imitative learning and sufficiently high-rationality logit learning, sufficiently low rationality can avoid such an undesirable outcome by stabilizing the disease-free
equilibrium. We also identify parameter regimes in which an unstable endemic
equilibrium stabilizes when rationality falls below a threshold. We numerically illustrate the convergence of infection state and response trajectories to a stable disease-free
equilibrium at a lower degree of rationality, whereas the same trajectories otherwise converge to a stable endemic
equilibrium as the rationality level rises. We also compare a specific case in which trajectories under replicator dynamics fail to converge, and illustrate convergence when rationality is below an analytically derived threshold.
Read the full abstract ▾
Computational Engineering
Domain-Adaptive Data Assimilation for Global AI Weather Forecasting
Minseok Seo, Noah Brenowitz, Doyi Kim et al. (+2 more)
AI weather
forecasting models are commonly trained on the ERA5 reanalysis, which is unavailable in real time. Operational deployment therefore relies on initial conditions produced by numerical or AI analysis systems that differ from those encountered during training. This mismatch can degrade forecast skill, while retraining for every analysis system is costly. Here, we present Domain-Adaptive Data Assimilation (DADA), an observation-guided framework that adapts external analyses to pretrained AI weather models. Starting from a background state, DADA optimizes only an initial-state perturbation while keeping the forecast model frozen. The perturbed state is propagated through the model, and its short-range trajectory is constrained by real-world observations through a learned observation operator. The resulting initial condition is shaped jointly by observational constraints and the dynamics learned by the target model. We evaluate DADA across five global AI weather models using backgrounds from the Global Forecast System and the AI-based HealDA. Across deterministic and probabilistic forecasts, DADA substantially reduces short-range skill loss caused by changes in the initial-condition source. More broadly, DADA turns observations into a common interface between independently developed analysis and
forecasting systems, enabling pretrained AI weather models to accommodate evolving operational initial conditions without reconstructing ERA5 or retraining the forecast model.
Read the full abstract ▾
All authors Minseok Seo, Noah Brenowitz, Doyi Kim, Hyesook Lee, Changick Kim
Computational Engineering
BRIDGE: Bridging Routine Inputs and Digital modelling for Geotechnical Engineering
Roshan Philip Saji, Waleed El-Sekelly, Tarek Abdoun et al. (+1 more)
The mechanical properties needed for geotechnical design require laboratory tests and are thus available for far fewer field samples than routinely measured basic (index) properties which are available for nearly every sample. We term this difference as the characterization gap. Purely data-driven models do not explicitly enforce physical laws and may be unreliable beyond their training domain, while direct constitutive calibration preserves physical consistency but requires advanced laboratory data and expert judgment. We therefore propose BRIDGE (Bridging Routine Inputs and Digital modelling for Geotechnical Engineering) a modular framework that predicts mechanical properties from routinely available basic properties. BRIDGE screens the
dataset for outliers and non-physical responses, bins the retained samples into
Cells, and selects a small calibration
dataset from each
Cell, while the remaining samples are used for validation. A mechanical model is then calibrated for each calibration sample, and the converged calibrations collectively define the mechanical surrogate of the
Cell together with its parameter ranges and the corresponding predictive envelope in the mechanical property space. A
Cell is validated when most validation samples fall within this envelope, whereupon the surrogate can predict the mechanical property ranges of new samples whose basic properties lie within the
Cell. In this work, we adopt a finite element implementation of the Extended Cam-Clay (ECC) constitutive model as the mechanical model, whose four parameters are calibrated in a multi-stage process. Applied to the Tokyo-CLAY/14/67760 database, BRIDGE predicts the mechanical properties of an entire
Cell from calibrations of just 10 to 20% of its samples. Finally, since the mechanical model, the binning criteria, and the sampling method are components of BRIDGE, the framework is easily extensible.
Read the full abstract ▾
All authors Roshan Philip Saji, Waleed El-Sekelly, Tarek Abdoun, Mostafa E. Mobasher
Computer Vision
Geometry-Aligned Semantic Matching for Cross-Modal Planar Image Registration
Zhiwei Wang, Defeng He, Yuxing Li et al. (+2 more)
Cross-modal image matching establishes stable and accurate
geometric correspondences across modalities for planar registration. Existing semantic representations provide cross-modal consistency, but semantic similarity does not necessarily imply
geometric correspondence. Meanwhile, fine-grained CNN features provide accurate local details but lack global cross-modal semantic guidance for stable refinement. To address these issues, we propose CDPM, which first establishes geometrically consistent semantic representations and then preserves their dominant role in correspondence estimation during fine-grained localization. Specifically, we progressively adapt DINOv3 using geometrically consistent cross-modal patch pairs, enabling feature similarity to better reflect true cross-modal spatial correspondences. We then construct a DINO-Centric Feature Pyramid, where multi-scale DINO representations maintain stable cross-modal correspondences, while a lightweight CNN branch provides auxiliary structural details for precise local refinement. Extensive experiments on three cross-modal
datasets demonstrate the superior performance of CDPM. On VIS-IR, compared with the dense matcher RoMa, CDPM improves AUC@3/5/10/20 by 7.36, 13.40, 13.75, and 10.42 percentage points, respectively, and reduces mACE from 5.83 to 2.78 pixels. It also outperforms RoMa v2 across all metrics while requiring 45.6% fewer FLOPs. The online demo and
dataset are available, and the code will be released on our project page at https://warren-wzw.github.io/CDPM/.
Read the full abstract ▾
All authors Zhiwei Wang, Defeng He, Yuxing Li, Meilu Zhu, Edmund Y. Lam
Cryptography and Security
LiBRA: Detection-Aware Image Watermark Removal via Bidirectional Latent Optimization
Saibo Ye, Huajie Chen, Xin Guo et al. (+5 more)
Digital watermarking supports source attribution for AI-generated images, but its reliability depends on resistance to removal attacks. Some attacks attempt to remove watermarks by forcing the decoded watermark to differ from the original. However, this can produce an inverted watermark that remains detectable, causing removal to fail, while further attempts to alter the watermark may unnecessarily degrade image quality. To address these limitations, we present LiBRA (Latent In-band Bidirectional Removal Attack), which aims to make watermarks undetectable while preserving image quality. Instead of continually pushing the watermark toward inversion, LiBRA adjusts the image to conceal the watermark without encouraging further changes that could degrade image quality. Some attacks keep pushing decoded bits away from the original watermark, even when further changes preserve detectability and damage image quality. With access to the watermark key and decoder, LiBRA makes bounded changes in a public autoencoder's latent space. Unlike inversion-driven objectives that cannot correct excessive inversion, LiBRA guides average decoding confidence toward random guessing from either direction. This helps avoid an inverted but detectable watermark. Leaving individual bits flexible allows image-quality constraints to favor less damaging changes, while an optional frequency-guided mask limits their location. We verify removal using an exact two-sided binomial test rather than assuming the confidence target guarantees success.
Read the full abstract ▾
All authors Saibo Ye, Huajie Chen, Xin Guo, Le Yang, Chi Liu, Xiangyu Hu, Jingjing Guo, Tianqing Zhu
Machine Learning
Sample complexity of variance-reduced policy gradient: weaker assumptions and lower bounds
Gabor Paczolay, Matteo Papini, Alberto Maria Metelli et al. (+2 more)
Several variance-reduced versions of REINFORCE based on importance sampling achieve an improved $O(\epsilon^{-3})$ sample complexity to find an $\epsilon$-stationary point, under an unrealistic assumption on the variance of the importance weights. In this paper, we propose the \algo (Defensive Policy Gradient)
algorithm, based on defensive importance sampling, which achieves the same rate without any assumption on the variance of ordinary importance weights. We also establish lower bounds in a generalized black-box policy-optimization model that hides states and actions and permits parameter-dependent rewards. In this model, the optimal rates are $\Theta(\epsilon^{-4})$ with bounded-variance one-policy feedback and $\Theta(\epsilon^{-3})$ with mean-square-smooth coupled two-policy feedback. Under standard policy-regularity conditions, REINFORCE and \algo realize the corresponding oracle conditions and attain the $O(\epsilon^{-4})$ and $O(\epsilon^{-3})$ upper bounds, respectively. Although the lower bounds do not apply directly to the classical MDP interaction model in which these
algorithms operate, this correspondence provides oracle-level evidence that the faster rate of \algo is optimal and genuinely separated from that of vanilla policy gradient.
Read the full abstract ▾
All authors Gabor Paczolay, Matteo Papini, Alberto Maria Metelli, Istvan Harmati, Marcello Restelli
Computation and Language
Predicting Steering Vectors and Adapter Weights for Few-Shot Author-Style Transfer
Leonard Popp, Danni Liu, Supriti Sinhamahapatra et al. (+1 more)
Adapting large
language models to an individual author's style from a few examples is challenging, and scientific writing sharpens the difficulty: formal conventions leave little surface variation, and authors write about their own topics, so extracted “style” easily entangles with content. We study style-conditioned abstract generation from a few example abstracts per author and propose three methods: (1) contrastive activation steering, (2) a network that predicts steering vectors, and (3) a hypernetwork that predicts LoRA adapters. We find a consistent trade-off between style imitation and output quality: fine-tuning buys most of the available style signal but forfeits fluency, while the hypernetwork achieves the best trade-off on both seen and unseen authors. Our steering operates at author level, contrasting an author's abstracts against style-neutral generations for the same content. This holds topic fixed, removes the need for a predefined style inventory, and outperforms inventory-based steering. % [EDIT 1a] softened "no single optimal axis" claim Moreover, our analyses demonstrate that manually extracted and predicted steering vectors are near-orthogonal yet score comparably, indicating that style conditioning here can admit at least two unrelated directions rather than requiring one particular axis.
Read the full abstract ▾
All authors Leonard Popp, Danni Liu, Supriti Sinhamahapatra, Jan Niehues
Graphics
Budgeted-GS: Real-Time Large-Scale Gaussian Splatting via Factoring LOD
Haipeng Wang
3D Gaussian Splatting achieves excellent visual quality with real-time rendering, but at the scale of entire cities it does not fit: a trained model carries millions of primitives and gigabytes of
memory, and real-time rendering at high quality on a consumer GPU remains out of reach. We introduce Budgeted-GS, a post-hoc method that turns any trained 3DGS model into a factoring tree, a multi-resolution hierarchy of moment-matched aggregates. After a construction pass of a few seconds, a single quality parameter selects, for each view, the level of detail that fits the
memory of the target device, so the same city-scale model serves GPUs with widely different
memory capacities. When a new scene is to be trained, the same theory applies: instead of growing a full-sized model and compressing it afterwards, budget-centered training first measures how many primitives the scene needs and then trains the model directly at that size, avoiding the wasted effort of optimizing primitives that are later discarded. Both methods are grounded in a measurable capacity floor, a budget-error law derived from optimal transport in phase space; selection rules certified by recent covering
theorems decide which primitives are redundant. The floor answers how many primitives a scene actually needs and how many can safely be given up. We validate the floor on 13 public scenes under a preregistered protocol, and exercise both methods from object scenes to an official city capture, rendering it at native 1920x1080, full SH, in real time on one consumer GPU.
Read the full abstract ▾
Machine Learning · Cross-listed from quant-ph
Landscape-Dependent Performance of Photonic Quantum Solvers in QUBO Feature Selection for Financial Risk Detection
Nirvik Sahoo and Paul Robert Griffin
Feature selection for imbalanced classification tasks such as credit card fraud and consumer default detection requires balancing predictive relevance, inter-feature redundancy, and computational feasibility. We
benchmark three computing paradigms, classical branch-and-bound optimization (Gurobi), photonic
entropy computing (QCI Dirac-3), and simulated photonic
boson sampling (Piquasso), across thirteen feature-selection methods on two
datasets: ULB Credit Card Fraud (30 features) and AmEx consumer default (159 features). Each method is routed to the solver matched to its mathematical structure. On ULB, Dirac-3 MI-Spearman matches the all-features model using 13 of 30 features (mean F1 0.873 +/- 0.023 over five runs, best run 0.896), and Piquasso is the best method at k=5. On AmEx, performance rises steadily with the feature budget and every paradigm approaches F1 = 0.80 only near the full feature set. Most differences between Gurobi and Dirac-3 on identical methods fall within run-to-run variation; the large gaps occur where the certified optimum generalizes poorly, most sharply for distance correlation on AmEx at k=25 (Gurobi F1 = 0.422 vs. a Dirac-3 mean of 0.746). At matched budgets, F1 varies about ten times more across methods on ULB than on AmEx, which we trace to how concentrated the predictive signal is in each feature space.
Read the full abstract ▾
Artificial Intelligence · Cross-listed from q-bio.QM
Multimodal reasoning for broadly neutralizing antibody discovery from label-free human B cell repertoires across virus families
Hantao Lou, Jianqing Zheng, Can Yue et al. (+21 more)
Discovering broadly neutralizing antibodies (bnAbs) from human natural immune repertoires remains a fundamental challenge in immunology, hindered by: the extreme rarity of bnAb, incomplete understanding of their cellular origins across pathogens, and the inability of existing computational tools to generalize across emerging viral threats. Here we present ImmuneAgent, a closed-loop AI system that integrates
multimodal reasoning with continual meta-learning and wet-lab feedback to overcome these barriers. Applied to screen the natural BCR repertoires from vaccinated or infected cohorts, the system achieves a ~55% neutralization antibody discovery rate (60 of 110 cloned candidates) and a ~11% bnAb yield (12 of 110), substantially outperforming a state-of-the-art sequence-based neutralization predictor or cofolding models evaluated at the same cloning budget. Five ImmuneAgent-discovered antibodies conferred 100% in vivo protection against lethal influenza challenge, comparable to the clinical-stage therapeutic MEDI8852. The system recovered the cellular and structural determinants of bnAb activity and identified FCRL5+CD27+ atypical
memory B
cells as a conserved bnAb reservoir and hydrophobic interface enrichment as a cross-viral structural signature, which generalized to unseen antigens, discovering human metapneumovirus (hMPV) cross-neutralizing and human papillomavirus (HPV)-neutralizing antibodies without antigen-specific sorting. These results validate that ImmuneAgent is a generalizable framework for rapid therapeutic antibody discovery against emerging viral threats.
Read the full abstract ▾
All authors Hantao Lou, Jianqing Zheng, Can Yue, Meihan Zhang, Yuanchao Bao, Yu Chen, Mengting Huang, Yupeng Yang, Qianyu Pan, Nana Fu, Yansong Shi, Hongli Li, Yangyang Chai, Ruyi Chen, Wansheng Li, Zhu Liang, Rongmei Yao, Yuanhan Mo, Lei Wang, Chunmei Wang, Yun Quan, Qiong Zhang, Xiangxi Wang, Xuetao Cao
Systems and Control
A New Index for Quantifying Stability Adaptation Effort in Power Electronics-Dominated Power Systems Based on Frequency-Domain Impedance Identification
Luis A. Garcia-Reyes, Macarena Martin-Almenta, Edgar Nuño-Martínez et al. (+4 more)
The increasing use of converter based resources has made stability assessment in modern power systems more challenging, particularly because many devices are available only as black box models. Conventional
eigenvalue based methods require full dynamic representations, while frequency domain impedance techniques based on the Generalized Nyquist Criterion (GNC) provide stability insight but do not quantify how much a converter must be retuned to restore stability. This paper introduces the Adaptation Effort Index (AEI), a metric that measures the sensitivity of dominant modes to parameter variations and quantifies the effort required for a converter to reconfigure itself under unstable operating conditions. The AEI combines the Nyquist Stability Margin obtained from frequency domain screening with a modal sensitivity index derived from fitted state space models synthesized from black box impedance data. The method is applied to a system comprising a grid forming and a grid following converter. Results show that, although both units can be tuned to stabilize the system, the grid following converter requires significantly less adaptation effort. The AEI therefore provides a practical and model agnostic tool for identifying which converter should be reconfigured in converter dominated grids, offering valuable guidance for system operators and manufacturers.
Read the full abstract ▾
All authors Luis A. Garcia-Reyes, Macarena Martin-Almenta, Edgar Nuño-Martínez, Javier Renedo, Antonio Cordón-Rodríguez, Oriol Gomis-Bellmunt, Eduardo Prieto-Araujo
Computational Engineering
Data-Free Weak-Form Staggered Neural Operators for Magneto-Mechanical Coupling in Finite-Strain Elastomers
Alireza Yazdandousthamedani, Ahmad Moeineddin, Reza Najian Asl et al. (+2 more)
Magneto-active elastomers, as a class of smart materials, exhibit strongly coupled
magnetic and mechanical behavior at finite strains. Considering variations in microstructure, material properties, and geometry can lead to computationally expensive analyses. Building on the finite operator learning (FOL) framework, this study develops a data-free, physics-informed operator-learning framework for families of coupled finite-strain magneto-mechanical boundary-value problems. The governing magnetostatic and mechanical equations are enforced through finite-element weak-form residuals, allowing the neural operators to be trained without labeled finite-element solution data. The main contribution of this study is the development of a weak-form staggered neural operator (WSNO) framework for strongly coupled magneto-mechanical saddle-point problems. Separate
magnetic and mechanical neural operators are trained alternately using a staggered optimization strategy, while their physical coupling is retained through the constitutive relations and residual evaluations. The resulting framework learns mappings from parameterized material and
geometric descriptions to the corresponding coupled
magnetic and mechanical solution fields. The proposed framework is investigated across several settings, including heterogeneous random-inclusion microstructures, varying
magnetic phase contrast, area-fraction-dependent geometries, strongly out-of-distribution material morphologies, and three-dimensional geometry-parametric problems. In addition, the learned operator is combined with a neural-initialized Newton strategy, in which the nonlinear finite-element solver is initialized using the neural prediction. The results demonstrate that the proposed operator-learning framework can accurately capture coupled magneto-mechanical responses across a broad range of parametric problem settings.
Read the full abstract ▾
All authors Alireza Yazdandousthamedani, Ahmad Moeineddin, Reza Najian Asl, Shahed Rezaei, Michael Kaliske
Neural and Evolutionary Computing
Aggregate accuracy conceals concentrated temporal vulnerability in a spiking speech classifier
İsmail Can Dikmen
Aggregate accuracy cannot reveal which utterances are locally vulnerable or how internal activity changes when labels remain stable. We retain every prediction for 725,070 adjacent-bin, one-count changes around 100 validation utterances of a frozen SpikeSCR-based classifier. The canonical native-horizon GPU singleton path reaches 86.0836% validation accuracy. Equal-source expected accuracy under a uniformly chosen neighbor rises from 84.00% to 84.54%, although 13 of 84 initially correct sources admit adverse neighbors. Five sources carry 93.21% of adverse moves. Margin-guided and surrogate-gradient searches miss sparse cases at fixed query budgets; a post hoc gradient prefix finds all 13 at 73.45% of the census of initially correct sources. Source-matched traces and clean-state interventions distinguish internal change from harmful direction and recoverability. Two count-readout replicas have similar validation accuracies but a 5.21-fold class-change gap concentrated in four sources. Controlled query/key source isolation eliminates 531 batch-order label changes and restores bitwise order invariance across all 9,981 validation score vectors. Isolated batch predictions reproduce padding-matched singleton labels on 9,980 of 9,981 inputs. Paired CPU/GPU replay of all 25,820 class-changing neighbors gives 99.8993% label agreement and preserves all 13 vulnerable sources and their fixed witnesses. Complete source-conditioned maps distinguish vulnerability incidence, concentration, internal change, and execution dependence that aggregate accuracy leaves unresolved.
Read the full abstract ▾
Computer Vision
Does Physics Live in the Activations? Localizing Physical Quantities in Video Diffusion Models
Jonas Kneifl, Jakub Skalski, Bartłomiej Twardowski et al. (+1 more)
Video generation models produce strikingly realistic sequences and are increasingly proposed as world models, yet recent
benchmarks reveal pronounced deficits in their physical
reasoning. This raises the question of whether these models internalize physical principles or merely reproduce familiar motion patterns. We address this by probing internal representations of
video Diffusion Transformers (DiTs) for simulator-derived ground-truth physical quantities spanning kinematic motion and rigid-body dynamics under
gravity and contact. We find that these quantities are linearly decodable with high accuracy early in the denoising process, substantially outperforming a baseline decoded directly from the model's own noised latents, indicating that the relevant physical information is actively constructed during denoising rather than already present in the input. Additionally, we show that activations at on-object
tokens carry the relevant physical information and that quantities defined over multiple frames are readable from single latent frames. Hence, information is sharply localized within the
token sequence and is computed globally but stored locally. The probes further show partial extrapolation, transferring to scene variations and object configurations outside their training regime, so what they read is not simply a correlate of the scenes they were fit on. When fitted directly in the full-resolution activation space, the probing directions can serve as steering vectors to change the model's output.
Read the full abstract ▾
All authors Jonas Kneifl, Jakub Skalski, Bartłomiej Twardowski, Kamil Deja
Cryptography and Security
EvoRiskBench: An Evolving Benchmark for Runtime Security Risks in Workspace Agents
Shiyi Kuang, Xuemei Luo, Kun Liu et al. (+7 more)
Workspace
agents combine large
language models with execution harnesses to perform stateful, multi-step tasks that access or modify external resources. Existing
benchmarks leave gaps in executable coverage of their runtime security risks, while evolving model capabilities, harnesses, tools, and threats motivate
benchmark evolution. We introduce EvoRiskBench, an evolving
benchmark organized around the EP-Path-EF framework, which links an initial risk entry point to a one-hop technical effect through an
agent-mediated risk path. The framework defines nine entry-point categories and five effect categories; a 20-participant study supports their interpretability and classification consistency on representative cases. Guided by this framework, an automated end-to-end workflow constructs and executes risk cases in isolated environments and independently verifies outcomes using runtime traces and environment states. The
benchmark provides a reproducible
dataset of 450
adversarial tasks across six scenarios. We evaluate nine model-harness configurations spanning three models (GPT-5.6 Sol, DeepSeek-V4-Pro-0813, and Claude Opus 5) and three harnesses (Claude Code, Codex, and OpenClaw). Our results reveal substantial vulnerabilities across systems. The most vulnerable configuration, Codex with DeepSeek-V4-Pro-0813, reaches a 68.44% attack success rate (ASR), indicating that configuration of workspace
agent is insufficient to ensure secure autonomous execution. ASR varies more across models than harnesses, and harness differences depend on the model. The
benchmark cases and evaluation platform will be released after completion of artifact safety and reproducibility checks.
Read the full abstract ▾
All authors Shiyi Kuang, Xuemei Luo, Kun Liu, Junhai Li, Rui Tian, Feng Shi, Bo Shen, Nianyu Li, Dehui Li, Ping Chen
Neural and Evolutionary Computing
Learning While Inferring: Local and Parallel Learning for Edge SNNs across Sensing Modalities
Yanxun Zhang, Yifei Wang, Changze Lv et al. (+6 more)
Edge intelligence requires models to sense continuously in real time and to keep adapting on-device, all under tight compute, energy, and
memory budgets. Although spiking neural networks (SNNs) enable efficient event-driven inference, standard surrogate-gradient backpropagation (BP) serializes updates and blocks ongoing inference. We investigate Bidirectional Spike-Based
Distillation (BSD) as an on-device learning principle that lets edge SNNs learn while inferring. BSD couples a stimulus-driven forward network with an independent target-driven reverse network and aligns their intermediate representations through local objectives. Because the two pathways have disjoint computation graphs until
alignment, forward inference, reverse inference, and stage-wise updates can run concurrently. On 25
benchmarks spanning the five sensing modalities of SOUL, BSD stays within 3.8 percentage points of matched BP baselines on average, while only its forward branch is needed at deployment. The learned representations also transfer well to few-shot class-incremental learning without replaying past data. By removing the dense floating-point backward chain, BSD reduces the projected training latency to $0.72\times$ and the estimated training energy to $0.36\times$ that of BP.
Read the full abstract ▾
All authors Yanxun Zhang, Yifei Wang, Changze Lv, Jingwen Xu, Yiyang Lu, Xiaohua Wang, Di Yu, Xin Du, Xiaoqing Zheng