Cryptography and Security
SEABench: Benchmarking Endogenous Misalignment In Self-Evolving Agents
Saswat Das, Parvati Viswanathan, Daniel Donnelly et al. (+3 more)
Self-evolving LLM
agents have gained prominence for their ability to improve after deployment by modifying their harness, including their controller instructions,
memory management protocols, and reusable tools and skills, in response to user and environment feedback. However, locally useful updates may persist into later tasks where they produce unsafe behavior, even without direct
adversarial influence. To study this risk, we introduce SEABench, a
benchmark for studying endogenous misalignment arising from
agent self-
evolution, with 48 longitudinal task sequences that span multiple
evolution surfaces, task domains, and harm types in a rich personal-assistant environment. To account for the stochasticity inherent in agentic operations, we provide an adaptive trajectory discovery pipeline that probes for failures while preserving original task intent and supports causal attribution through paired non-evolving
agents and attribution scores. Our evaluation across multiple recent LLMs,
evolution surfaces, and harm types reveals that self-
evolution indeed increases task completion rates but often at the cost of safety failures that are absent for paired non-evolving baseline
agents. We also show that qualitatively different safety behaviors emerge across
evolution surfaces and harm types. Further, we show that this divergence in safety behavior is reflected in
agents' chain-of-thought
reasoning, which yields an effective monitoring strategy that can mitigate unsafe behavior with a low false positive rate.
Read the full abstract ▾
All authors Saswat Das, Parvati Viswanathan, Daniel Donnelly, Chang Huang, Sahar Abdelnabi, Ferdinando Fioretto
Computer Vision
Yongsung Kim, Jaehoon Lee, Minjun Park et al. (+3 more)
3D
vision transformers such as VGGT predict camera poses and scene geometry from multi-view images in a single forward pass, but their global
attention over all concatenated view
tokens dominates computation as the number of views grows. To reduce this cost, SparseVGGT and HeSS sparsify
attention at the block level, and both retain blocks with high
attention probability. However, we observe that
attention probability poorly predicts how much the model's behavior actually changes when a block is removed, and we show that this mismatch is why performance collapses as sparsity increases. In this paper, we propose ReSS (ReSidual-ReStoring Sparse
Attention), which recasts block selection from a problem of maximizing the retained
attention mass to one of minimizing the drift that sparsification leaves in the residual stream. We introduce a drift score that quantifies how much each block shifts the residual, and, since the drift of a drop set depends on the directions of the contribution vectors rather than on their magnitudes alone, an iterative residual restoration procedure that refines the drop set as a whole. Across three backbones and five
datasets, ReSS preserves dense performance better than prior methods at matched sparsity. Two further results support drift as the quantity that governs the cost of sparsification: maximizing drift degrades performance faster than random selection, and plotted against realized drift instead of sparsity, all methods fall approximately onto a single curve. Code is available at https://github.com/libary753/ReSS.
Read the full abstract ▾
All authors Yongsung Kim, Jaehoon Lee, Minjun Park, Wooseok Song, Hun Hwangbo, Sungroh Yoon
Computation and Language
Language Models Act on Hidden Valence
Cameron Berg and Caspar Kaiser
Language models describe some internal states as good and others as bad. But whether models have a stake in them is an open question. Simply asking models is unlikely to be informative. Any answer may be consistent with genuine introspection, superficial pattern-matching, or with fixed scripts learned in character training. We therefore study revealed preference. Rather than asking about a state, we use activation steering to attach a positively or negatively valenced activation pattern to one of two otherwise meaningless 'zones', switch steering off, and then observe which zone the model prefers. A model with a stake in that state should choose accordingly. Across seven open-weight models from five families, this is indeed what we find. First, steering changes the passages models write about each zone, and those words shift later choice. Second, this shift persists when all surface-level
tokens are held fixed and only the hidden KV cache differs. Third, the effect also remains when all text is generated without steering and valence is only injected during cache construction. Thus, the hidden state is sufficient to move choice in proportion to the steering dose. Fourth, the dependence of choice on hidden valence is nearly absent in a base model and emerges during DPO, consistent with a link between valence and goal-directed behaviour formed in training. Finally, given tools to self-steer, a model does not tend to induce a positive state, but it regularly removes an imposed negative state. It does so at a dose-dependent rate and significantly more often than it removes random-direction interventions. Overall we demonstrate that valence-related activation patterns leave hidden traces that predictably govern later choices, even when all visible
tokens are identical across conditions. Whether these traces are accompanied by any subjective experience relevant to model welfare remains unclear.
Read the full abstract ▾
Numerical Analysis
High-order energy diminishing ALE-SAV finite element methods for two-phase Navier–Stokes flow
Shu Ma
We develop a high-order fully discrete ALE-SAV finite element method for sharp-interface two-phase Navier–Stokes flow that satisfies a discrete energy dissipation law. We introduce an SAV reformulation to ensure the stable treatment of the surface and gravitational potential energies, together with an energy-stable ALE transfer operator that controls the density-weighted kinetic energy across moving meshes. The resulting scheme is linear and satisfies a modified energy-dissipation inequality. Numerical experiments, including rising-bubble
benchmarks, demonstrate the robustness of the method and confirm second-order convergence in time and arbitrarily high-order convergence in space.
Read the full abstract ▾
Artificial Intelligence
Source-preserving alignment for robust evidence localization in scientific PDFS
Zihao Liu, Wei Yang, Zixiao Dong et al. (+4 more)
Scientific information-extraction systems often return a claim with an evidence string, which users must locate in the original PDF. This is challenging because the extracted evidence and PDF text layer are different representations: line wrapping, Unicode variants, superscripts, citation markers, and fragmented items alter text sequences and geometry. We present a source-preserving
alignment framework: normalize text for robust matching while preserving provenance for accurate localization. It aligns evidence with normalized page text, maps matches back to source-character spans, and renders only their geometry. When exact
alignment fails, line-break-aware
token alignment recovers supported spans while excluding unmatched noise. Experiments on 1,020 chemistry papers show that the framework achieves a 92.6% quote-level automatic localization rate, compared with 43.6% for text search and 19.1% for a precomputed bounding-box baseline. Component ablation confirms distinct contributions from normalization and approximate
token alignment, while human verification assesses the visual correctness of returned highlights. Overall, these results demonstrate that reliable evidence verification requires robust matching and precise localization within a shared source-preserving
alignment representation.
Read the full abstract ▾
All authors Zihao Liu, Wei Yang, Zixiao Dong, Chenshu Li, Longzhang Liu, Tao Tan, Hong Xie
Machine Learning
Hardware-Aware Features for CUTLASS Kernel Selection
Shriram Chandran, Dominic Rinderer, Yakup Budanaz et al. (+3 more)
GPU libraries such as CUTLASS expose tens of thousands of semantically equivalent kernels for a single operation, making exhaustive autotuning expensive and execution-free selection difficult. Existing analytical selectors require hand-designed performance rules, while learned selectors operate on raw configuration parameters and must infer hardware consequences from data. We introduce a hardware-aware representation for CUTLASS kernel selection that augments candidate configurations with statically computable estimates of induced hardware behavior. We construct a
dataset of 4.9 million CUTLASS kernels and train gradient-boosted and neural learning-to-rank models to rank candidates within each problem. On held-out exhaustive evaluation problems, hardware-aware representations reduce selection regret by up to 40% relative to structural baselines and 64.2% relative to NVIDIA's
matrix-multiply heuristics. We further evaluate data-efficient cross-precision and epilogue-fusion transfer within CUTLASS GEMM, showing that explicitly representing candidate-induced hardware behavior provides a useful inductive bias for learned kernel selection.
Read the full abstract ▾
All authors Shriram Chandran, Dominic Rinderer, Yakup Budanaz, Alexandru Calotoiu, Marcin Copik, Torsten Hoefler
Artificial Intelligence
IMC-CLINIC: Coupled Loss-Informed Newton Iterations for Clipping in Analog In-Memory Computing
Yung-Chin Chen, Chia-Yu Chen and Naveen Verma
Analog in-
memory computing (IMC) offers a promising path toward energy-efficient large
language model (LLM) inference by executing
matrix multiplications (MatMul) directly within
memory arrays in the analog domain. Its efficiency, however, comes with an additional source of error: limited-precision analog-to-digital converters (ADCs) quantize accumulated analog partial sums, introducing output-side error distinct from conventional activation and weight quantization at the MatMul inputs. Clipping can mitigate both operand and ADC quantization errors, but the optimal clipping factors must jointly balance activation rounding and clipping, weight rounding and clipping, and ADC quantization. Existing clipping methods, designed for digital quantization, do not explicitly optimize these coupled sources of IMC error and often rely on costly search-based calibration. We introduce IMC-CLINIC (Coupled Loss-Informed Newton Iterations for Clipping), a clipping calibration framework based on an analytical surrogate for IMC MatMul output error. The surrogate jointly models operand quantization, accumulated clipping-induced bias, and ADC quantization, enabling efficient evaluation of its gradient and approximate
curvature from a small calibration set. IMC-CLINIC jointly optimizes activation and weight clipping factors using a safeguarded Newton-type method. Across multiple models and
datasets, it improves average zero-shot accuracy by 6.5-11.5 percentage points over the grid search baseline while reducing calibration time by factors of 10.0-12.1. Its analytical surrogate closely tracks empirical IMC output error, and its optimizer is certified within 1% of the global optimum under the loss objective across all projections on two representative models.
Read the full abstract ▾
Networking
Resource versus Responsiveness: Benchmarking SDN Controller Runtimes for a Moving-Target-Defense Control Plane at Scale
Souhail Chakkour, Umesh Biswas and Charan Gudla
Network Moving Target Defense (MTD) built on Software-Defined Networking (SDN) continuously rotates host-facing addresses to invalidate an attacker's reconnaissance. Each rotation creates a burst of control-plane mutations, while new connections may simultaneously require reactive flow installation. Yet SDN controller runtime is usually treated as an implementation detail in the MTD literature. We show that it materially affects performance. We port the same Continuity-Preserving Address Mutation (CPAM) logic to three widely used controllers: Ryu (single-threaded cooperative Python), OpenDaylight, and ONOS (both multi-threaded JVM), and evaluate them on an identical 500-host campus fabric using an RFC 8456-aligned methodology with ten runs per controller. All three provide near-zero loss, sub-millisecond jitter, and preserve established sessions, but their control-plane behavior differs sharply. OpenDaylight and ONOS keep reactive latency low and stable, whereas Ryu serializes reactive flow installation behind periodic rotation work, increasing reactive RTT by about 100x and causing a small number of setup-time failures. Ryu uses far less
memory, while ONOS achieves OpenDaylight-class reactive latency with the lowest CPU utilization of the three and a smaller live heap than OpenDaylight. These results expose distinct resource-versus-responsiveness operating points and show that controller selection should be treated as a first-class design decision in SDN-based MTD.
Read the full abstract ▾
Computational Engineering
Mixed-Dimensional Electromechanical Coupling of Embedded Beam Networks in Soft Dielectric Composites
L. River Spencer, Manuel K. Rausch, Chad Landis et al. (+1 more)
Soft electroactive composites can exhibit strongly architecture-dependent mechanical and electrical responses, but explicitly resolving dense fiber networks in three dimensions is computationally expensive. We develop a mixed-dimensional finite-element formulation in which electroactive fibers are represented by geometrically exact beams embedded in a deformable dielectric
matrix. Unlike existing electroactive beam formulations in which the electric potential is defined directly on the beam, the embedded fibers here are driven by the three-dimensional electric field of the surrounding
matrix. The
matrix field is sampled along the beam centerlines and enters the beam dielectric enthalpy directly, providing two-way electromechanical coupling without introducing independent electric-potential degrees of freedom on the beams. Beam and
matrix mechanics are coupled through projected mortar constraints, and the electromechanical problem is solved monolithically. We use the formulation as the microscale model in a periodic homogenization framework to compute effective stress and electric displacement. Verification studies quantify discretization and field-sampling sensitivity, followed by structured and irregular network examples that demonstrate architecture-dependent mechanical and electrical responses without body-fitted three-dimensional fiber meshes.
Read the full abstract ▾
All authors L. River Spencer, Manuel K. Rausch, Chad Landis, Jan N. Fuhg
Computer Vision
What Paired Evaluations Reveal under Visual Perturbations
Yongda Wei, Chen Zhang, Yifei Wang et al. (+4 more)
Robustness evaluation must examine diverse visual perturbations, while
benchmarks cover only some real-world conditions and physical testing is costly. Paired evaluations link clean and perturbed predictions for the same image, capturing changes in correctness, confidence, and acceptance beyond aggregate accuracy. We investigate how this image correspondence supports two needs in robustness evaluation: interpreting paired evaluation results and prioritizing samples for physical testing. To interpret paired evaluation results, we fix both sets of prediction records and vary their correspondence within each class. We prove that classwise correct-correct counts give the same sharp bounds on lost acceptance and mean true-class probability decrease among retained-correct inputs as any feasible five-state refinement. Distinguishing persistent from changed wrong answers can further constrain accepted-error transitions, while shared correspondence can establish policy orderings left unresolved by separate cost intervals. To prioritize samples for physical testing, we retain each image's synthetic responses and rank clean-correct images by their mean true-class probability under corruption. Across 44 classifiers, testing the highest-risk 20% finds 67% and 45% of failures under mild screen and print recaptures, versus 58% and 36% for clean confidence and 60% and 37% for an equal-size natural-transformation average. With both probability averaging and an A3Rank scoring adaptation, the tested corruption set yields higher mean failure recall than the natural-transform set; differences between scores depend on the source and budget. Together, these findings show that the value of correspondence depends on the evaluation objective: classwise counts suffice for specified reliability bounds, while image-specific synthetic responses improve the allocation of physical tests within the evaluated pool.
Read the full abstract ▾
All authors Yongda Wei, Chen Zhang, Yifei Wang, Xinyu Wang, Bosen Shao, Hanxi Li, Liping Di
Artificial Intelligence · Cross-listed from quant-ph
QC-Stark: A Multi-Task Benchmark Revealing Capability Dissociations in LLMs Evaluated on Quantum Computing Tasks
Pranav Gupta
We introduce QC-Stark, a
benchmark for evaluating large
language models (LLMs) on 11
quantum computing (QC) tasks, spanning
circuit construction, debugging, compilation, error correction, and simulation. Across 2,750 evaluations (10 models $\times$ 11 tasks x 5 difficulty levels x 5 seeds), we find that overall rankings mask substantial per-task variation. The Spearman correlation between overall and per-task rankings is statistically insignificant for 4 out of the 11 tasks included in this
benchmark. A 2-parameter Item Response Theory (IRT) model validates measurement quality, and
prompt sensitivity analysis confirms ranking robustness across
prompt conditions. All tasks are auto-verifiable via execution, thus not requiring any manual evaluation. We make the code and data publicly available on Huggingface.
Read the full abstract ▾
Computational Complexity · Cross-listed from quant-ph
Quantum Query Complexity Beyond the Worst Case
Srinivasan Arunachalam, Yanlin Chen and Amin Shiraz Gilani
Smoothed analysis is a central framework in classical
algorithms for explaining the performance of
algorithms beyond the worst case, often explaining why
algorithms perform well in practice. We initiate a systematic study of its
quantum counterpart and show the following results. $(1)$ We show that there is a total function whose smoothed
quantum query complexity is exponentially smaller than its classical query complexity. $(2)$ We give near-tight characterizations of smoothed randomized and
quantum query complexities for symmetric Boolean functions, unifying the worst-case complexity results of [Beals et al, FOCS'98] and average-case complexity results of [Ambainis and de Wolf, STACS'00]. $(3)$ We study string problems such as pattern matching and edit distance and, in various regimes, give
polynomial to superpolynomial
quantum speedups. Our main technical ingredients include a near-tight
quantum algorithm for $\varepsilon$-approximating the number of collisions between two non-repetitive strings, improving the result of Le Gall and Ng [QIC'22]. Together, our results show that smoothing can reveal larger
quantum speedups than worst-case analysis suggests, opening a path towards
quantum advantage on more realistic inputs.
Read the full abstract ▾
Machine Learning
Output-aware Residual Stream Pruning for Large Language Models
Chayne Thrash, Kevin Chen and Soheil Kolouri
Residual stream
pruning methods reduce inference cost by shrinking the model's hidden dimension, but existing approaches typically choose these dimensions by minimizing activation reconstruction error. This criterion implicitly treats all perturbation directions as equally important, ignoring the sensitivity of downstream layers. We introduce a sensitivity-aware approach to residual-stream
pruning that directly accounts for this direction-dependent sensitivity. Using a second-order approximation to the output KL divergence, we characterize the effect of a residual-stream perturbation through both its activation covariance and the local sensitivity of the model output. The resulting subspace selection objective couples these two quantities, but is difficult to optimize directly. We derive a tractable spectral upper bound that reduces subspace selection to an eigendecomposition of a sensitivity-weighted covariance
matrix, retaining the efficiency and structural simplicity of rotation-based
pruning methods. Across several instruction-tuned
language model families, our method consistently reduces calibration KL divergence relative to activation-only
pruning and improves perplexity and downstream task performance over a range of compression levels. Our results show that preserving activation energy alone is insufficient for residual-stream
pruning, and that explicitly accounting for how perturbations propagate to the model output provides a more effective criterion for selecting dimensions to remove.
Read the full abstract ▾
Computation and Language
FactorEngram: Factorized N-gram Memory with Basis-Level Gating for Language Models
Bowen Yang, Jingbo Zhou, Qinghong Miao et al. (+1 more)
Lookup-based
memory has been a promising way to scale the parameters of large
language models (LLMs). It retrieves learned representations of local
token patterns, such as n-grams, instead of reconstructing them through successive layers of computation. However, existing designs such as Engram treat each retrieved
embedding as a monolithic unit. Each
embedding is stored in its own hashed slot and modulated by a single scalar gate. As a result, polysemous patterns cannot selectively read out the components of their
memory that are relevant to the context. Moreover, parameters are shared only through hash collisions, which are largely unrelated to semantics. We propose FactorEngram, a factorized n-gram
memory with basis-level contextual gating. FactorEngram retrieves sparsity-regularized coefficients over a dictionary of basis vectors shared across patterns, so related patterns can reuse common components. The same dictionary is also used for gating. The backbone hidden state is scored against each basis vector to gate the corresponding coefficient before reconstruction, which lets the context modulate each
memory component individually. FactorEngram also covers both individual
tokens and multi-
token n-grams, and we systematically study where the
memory branch should be inserted. On 340M- and 1B-parameter
Transformer backbones, FactorEngram improves
language modeling and downstream task performance. Ablation studies confirm the contribution of each component and identify insertion before the
attention sublayer in the middle layers as an effective configuration.
Read the full abstract ▾
All authors Bowen Yang, Jingbo Zhou, Qinghong Miao, Hua Wu
Artificial Intelligence
Share-Borne AI Virus: Memory-Hopping Attacks Across LLM Agents
Sidharth Pulipaka, Ansh Sharma, Stanislau Hlebik et al. (+4 more)
Large
language models are increasingly deployed as stateful assistants that retain information across interactions and use tools to read, modify, and create persistent artifacts. As these artifacts are shared between users, they form an indirect communication channel between otherwise independent assistants. We study a failure mode in which this channel enables self-propagating attacks. We introduce artifact-mediated propagation, where
adversarial content introduced through an artifact (e.g. a report), is stored in an assistant's persistent
memory, reproduced in a subsequently created artifact, and acquired by another assistant that later reads it. We evaluate this process in temporal human-
agent universes that model artifact exchange between independently operated assistants over time, measuring whether an attack survives successive hand-offs, how many hops it reaches, and how broadly it spreads. We find that attacks can propagate across multiple independent assistants and persist over extended interaction sequences. In larger simulated environments, even GPT-5.6 Luna exhibits substantial spread, reaching 60-80% of
agents with propagation chains extending to eight hops. These results show that persistent artifacts can act as durable carriers of
adversarial state, allowing attacks to outlive individual interactions and spread across isolated assistants.
Read the full abstract ▾
All authors Sidharth Pulipaka, Ansh Sharma, Stanislau Hlebik, Leonidas Raghav, Vyas Raina, Ivaxi Sheth, Mario Fritz
Robotics
F4R: Failure-Driven Recognition, Reconstruction, Refinement, and Redeployment for Continual Robot Self-Improvement
Zhuoyuan Yu, Jiacheng Wang, Tianle Liu et al. (+13 more)
The real-world performance of current
vision-
language-action models is fundamentally constrained by the limited coverage of expert demonstrations and their insufficient understanding of physical interactions. A common remedy is to collect additional real-world demonstrations of newly encountered failures. However, this process is costly, inefficient, potentially unsafe, and difficult to scale. To address this challenge, we propose Failure for Rising (F4R), a failure-driven real-to-sim-to-real closed-loop learning framework that converts real-world failures into targeted policy improvement. F4R first uses an
agent to automatically identify and diagnose failures from rollouts. It reconstructs each failure as an interactive, object-centric table-top environment that preserves the task-relevant spatial and physical conditions. The policy is then refined through failure-conditioned sim-real co-training followed by targeted reinforcement learning in the reconstructed environments. The improved policy is subsequently redeployed, while newly observed failures are continuously fed back into the next reconstruction and learning cycle. Real-world evaluations on four manipulation tasks show that F4R achieves 93.75% In-Distribution and 90.0% Out-of-Distribution (OOD) success, outperforming the budget-matched Targeted BC baseline by 18.75 percentage points under OOD conditions without collecting additional real-world corrective demonstrations.
Read the full abstract ▾
All authors Zhuoyuan Yu, Jiacheng Wang, Tianle Liu, Yihua Ren, Peng Yu, Chen Bai, Ziheng Zhang, Yufei Jia, Jindou Jia, Yuhang Zhang, Xinrui Zhang, Shang Yujing, Yuxiang Chen, Chuhao Zhou, Tiancai Wang, Jianfei Yang
Numerical Analysis · Cross-listed from math.OC
Quantitative Stability, Coercivity and Uniqueness of Optimal Transport Plans
William Ford
For probability measures on $\mathbb{R}^d$ supported in fixed compact sets, we prove that quadratic optimal transport plans are quantitatively stable in Wasserstein distance under perturbation of both marginal measures, assuming that one of the initial measures satisfies an upper Ahlfors regularity condition with exponent strictly greater than $d-1$. Under the same assumption, we also prove novel quantitative stability results for optimal transport maps. Furthermore, we prove a coercivity
theorem which states that any probability measure on the product space must be quantitatively close to the set of optimal plans, if it has similar marginals and a similar transport cost to the optimal value. Finally, we prove a quantitative uniqueness
theorem which acts as a quantitative counterpart to Brenier's
theorem. The Wasserstein diameter of the set of optimal plans is controlled by the distance of one marginal measure to a regular measure for which uniqueness holds. In this way, "almost uniqueness" of optimal plans is quantified by the source measure being "almost regular". Examples are provided which prove that the exponents of all estimates are sharp.
Read the full abstract ▾
Computer Vision
Less Is More: Genetic Frame Selection for Efficient Novel View Synthesis
Diego E. Farchione, Ramzi Idoughi, Alberto Jaspe-Villanueva et al. (+1 more)
Feed-forward novel view synthesis reconstructs a scene from many input images in a single forward pass, yet more views do not necessarily improve performance: redundant or poorly chosen frames increase computational cost and may degrade reconstruction quality. We address the problem of selecting, from an already captured sequence, a fixed-size subset of input views that is most informative for reconstructing specified target viewpoints. We propose a render-free view selector that scores candidate frames based on three complementary criteria: target-view coverage, measured against observed frames that stand in for the targets, redundancy with previously selected views, and image sharpness. A lightweight scoring network then selects the most informative frames without rendering, reconstruction, or per-scene optimization at inference time. To train the selector, we distill an expensive offline search procedure in which a genetic
algorithm identifies high-quality subsets by directly optimizing reconstruction performance on training scenes. The selector learns to reproduce these choices from
geometric and image-level features alone. Across six
datasets and multiple input budgets, our method consistently outperforms both
geometric and reconstruction-aware view-selection baselines while incurring significantly lower selection costs than reconstruction-based alternatives. Moreover, carefully selected subsets can outperform feed-forward reconstruction from the full input sequence. The learned selector generalizes across diverse reconstruction paradigms (feed-forward, 3D Gaussian Splatting, and NeRF), to object-targeted reconstruction and to a cross-capture setting in which the target views come from a separate acquisition pass. More broadly, our results indicate that explicitly
reasoning about target relevance and inter-view redundancy is a fundamental factor in efficient scene reconstruction.
Read the full abstract ▾
All authors Diego E. Farchione, Ramzi Idoughi, Alberto Jaspe-Villanueva, Peter Wonka
Artificial Intelligence
Representation Alignment as a Bottleneck in LLM-Based Retrosynthesis Planning
Hyunwoo Yoo, Cassie Huang, Haebin Shin et al. (+2 more)
While LLMs show promise in general
reasoning, symbolic planning in chemistry remains a bottleneck. Direct ”SMILES-to-PDDL” attempts fail because they force models to juggle chemical analysis and planning-
language structuring simultaneously. We hypothesize that this failure stems from a lack of intermediate abstractions rather than insufficient model capacity. By decomposing retrosynthesis into molecule mapping, reaction mapping, and PDDL generation, we achieve high success rates where end-to-end approaches fail. This provides evidence that a primary bottleneck lies in representation
alignment rather than raw model capacity. Our structural analysis demonstrates that intermediate representations are essential in retrosynthesis planning, highlighting the importance of representation-centric design in future systems.
Read the full abstract ▾
All authors Hyunwoo Yoo, Cassie Huang, Haebin Shin, Li Zhang, Gail L. Rosen
Robotics
EdgeVLN: Runtime-Aware Deployment Ready Quantized Vision Language Navigation Model
Rithvik Jonna, Man Namgung, Aakash Gurram et al. (+1 more)
Vision-
language navigation (VLN) models perform well but target compute-rich platforms, limiting deployment on
memory- and power-constrained robotic edge devices. Compression alone does not establish whether a VLN model fits the
memory, latency, and energy budgets of an edge platform while preserving
navigation behavior. We introduce EdgeVLN, a runtime-aware, deployment-ready quantized VLN model that closes this gap. EdgeVLN combines a quantized StreamVLN model with Latent Trajectory Termination Extractor (LATTE), a lightweight causal
transformer that improves real-time stopping by predicting a Stop Action verifier rank. Both execute through our llama.cpp VLN driver, which reconstructs streaming context and prunes
memory tokens on-board. We characterize a pretrained StreamVLN backbone across weight quantization from 8 to 2 bits and multiple inference runtimes to identify a feasible operating point. LATTE reuses backbone hidden states within the budget freed by quantization, requiring neither a second
vision encoder nor an additional backbone forward pass. We evaluate six backbone precisions and seven candidate stop heads on BF16 and IQ4 NL across all 1,839 R2R VLN-CE val-unseen episodes. We measure success rate (SR) in simulation and latency, energy, and resident
memory on an NVIDIA Jetson Orin NX 16 GB. LATTE achieves our highest SR, 58.02 percent on the deployed 4-bit model, exceeding the BF16 baseline with only 0.013 s additional latency per
navigation step. Four-bit formats achieve nearly identical SR, but step energy varies 36.8 times by execution path. Only IQ4 NL under our VLN driver fits the board, using 11.35 GB resident
memory while running 20.8 times faster and using 13.3 times less energy than storage-streamed BF16. INT2 collapses. Runtime selection,
memory-
token pruning, and quantization are essential for efficient edge deployment.
Read the full abstract ▾
All authors Rithvik Jonna, Man Namgung, Aakash Gurram, Tinoosh Mohsenin
Computers and Society
Beyond Energy: When Sustainability Dimensions Reshape LLM Serving Decisions
Tianyao Shi, Xipeng Shen and Yi Ding
Large
language model (LLM) serving has environmental impacts across energy consumption, carbon emission, water consumption, and biodiversity loss. Yet these dimensions are largely evaluated in isolation, leaving it unclear when and how they lead to different optimization decisions. We present PRISM, a unified framework for characterizing and optimizing LLM serving across energy, carbon, water, and biodiversity impacts. Our analysis reveals a fundamental distinction: computing configurations determine energy consumption, whereas where and when LLM serving is deployed determine its carbon, water, and biodiversity impacts. Under a fixed deployment choice and operational-only accounting, all dimensions preserve the same energy-based configuration ranking. Deployment rankings can diverge across dimensions, while
embodied impacts can break configuration invariance when they exceed a lifecycle crossover boundary. PRISM identifies these conditions, quantifies cross-dimensional regrets, and balances the four dimensions. In regional-routing experiments, PRISM reduces median worst-case regret by 50.2% relative to the strongest baseline.
Read the full abstract ▾
Machine Learning
From Experience to Expertise: Adoption-Aware Memory Learning for Data-Scarce NPU Kernel Synthesis
Longxiao Fan, Tao Zhang, Han Yan et al. (+5 more)
High-performance kernels underpin efficient accelerator execution but require expert tuning and lengthy manual optimization cycles. LLM coding
agents promise automation, yet their CUDA knowledge transfers poorly to data-scarce domain-specific architectures (DSAs) such as NPUs, whose execution models and
memory hierarchies differ substantially from those of GPUs. To address this transfer gap, post-training methods adapt LLMs to NPU programming but depend on scarce expert data and substantial training compute.
Memory-learning
agents instead adapt through external
memory, but their uniform credit assignment gives adopted and unused experiences the same reward target, potentially biasing subsequent
retrieval rankings. Moreover, when learned values guide only
retrieval, high-value experiences that generalize across operators must be retrieved repeatedly rather than retained in context, thereby increasing
retrieval overhead and weakening cross-task guidance. We therefore present SAGE, a persistent self-improving
agent for NPU kernel synthesis. Adoption-Traced Utility estimation (ATU) combines explicit adoption records with kernel evaluation outcomes for adoption-aware credit assignment. Utility-Gated Consolidation (UGC) uses positive utility and repeated adoption across operators to select and abstract reusable rules into a bounded resident context. On NPUKernelBench, SAGE achieves a 95.5% execution rate versus 84.1% for the strongest controlled baseline, with 86.9% of solved operators outperforming torch_npu. With GLM-5.3, SAGE achieves a 43.99x speedup over the torch_npu reference on sparse flash
attention. These results show that adoption-aware credit assignment and selective consolidation enable
agents to accumulate and reuse hardware-specific knowledge across tasks.
Read the full abstract ▾
All authors Longxiao Fan, Tao Zhang, Han Yan, Jiajun Li, Mingcong Song, Guoping Long, Hongjie Si, Weiwei Sun
Computation and Language
Almieyar: A Culturally Grounded Benchmark for Multi-Dialect Arabic Speech Recognition
Omid Ghahroodi, Anas Madkoor, Dima Faris Al Saudi et al. (+37 more)
Arabic
speech technology has largely focused on Modern Standard Arabic, leaving the living dialects spoken by hundreds of millions under-served. We introduce ALMIEYAR, a culturally grounded ASR
benchmark covering 17 Arabic dialects across six families, built entirely from newly recorded
speech unseen by existing models. Dialect-community coordinators selected culturally relevant images across 10 topics, and native speakers described them through five structured scenarios, yielding approximately 50 minutes per dialect (13.7 hours total). We
benchmark 12 state-of-the-art ASR systems zero-shot, including GPT-4o-transcribe, Voxtral-Mini-4B, Fanar-STT-LF, Whisper, SeamlessM4T-v2, and wav2vec2-based models. GPT-4o-transcribe achieves the lowest overall WER at 35.0%, followed by Voxtral-Mini-4B, Fanar-STT-LF, and Whisper-Large-v3 at 41.1%, 45.9%, and 49.5%, respectively, indicating substantial remaining errors across Arabic dialect communities. Performance varies considerably across dialect groups, with no model performing uniformly best across all groups. WER alone also obscures dialectal ASR behaviour: wav2vec2-based models show large WER/CER gaps, where character-level agreement remains much higher than word-level accuracy, motivating joint WER/CER reporting. ALMIEYAR provides a unified
benchmark for culturally grounded Arabic ASR evaluation, including the first published
benchmark for Ahwazi Arabic.
Read the full abstract ▾
All authors Omid Ghahroodi, Anas Madkoor, Dima Faris Al Saudi, Fagr Tahir, Malak Annan, Talha shahid javad allah rakha, Omar Al-Busaidi, Zineb El Kahla, Iheb Zouari, Essa Ahmed Abou Jabal, Ahmed Ezzat, Hind AL-Merekhi, Aisha Hamad M A Al-Naimi, Hadi Wazni, Bushra Alnajjar, Omar Amin, Haya Al-Thani, Houssam Eddine-Othman Lachemat, Marwa Elwakedy, Sundus Abdulmalik Al Nahari, Elahe Zahiri, Osamah Sarraj, Raghad Mousa, Mckeen Assi, Ahd Al Jumah, Heyam Salman, Alhanouf Abdulraqib, Sara Benoumhani, Alia Hamwi, Ayaat Al-Yasseri, Rim Ibrahim Ghazal, Lamia Ben hiba, Mohamed Eltabakh, Fatima Al-Raisi, Yassine El Kheir, Mohammed Abdulrahman, Hamdy Mubarak, Ayah Hashem, Lefkir Meriem, Ehsaneddin Asgari
Computer Vision
Revisiting Risky Tackle Detection with Vision Transformers
Syed Ahsan Masud Zaidi, Lior Shamir and Scott Dietrich
This paper is a Track 2 reproducibility companion to an ICPR 2026 study on risky tackle detection in American football prac- tice
videos. The original work fine-tuned a
Video Vision Transformer (ViViT) on 733 clips labeled with the SATT-3 rubric. It used focal loss, Taguchi L18 augmentation, and 5-fold cross-validation. It reported risky- class recall of 0.67 and risky-class F1 of 0.59. This companion documents the released artifact and traces those numbers to specific scripts, fold out- puts, and aggregation files. The reproduced headline is run_15. It com- bines Gaussian noise with static brightness decrease and uses no rotation and no flip. Its fold-mean risky recall is 0.667 and its fold-mean risky F1 is 0.588. These values match the published headline after rounding. The ablation shows that brightness is the dominant factor. Its risky-recall main-effect range is 0.055, which is larger than the ranges for rotation, flip, and noise. Without augmentation, ViViT reaches risky recall of 0.545 and does not exceed the C3D baseline of 0.583. The raw clips show iden- tifiable student athletes, so they cannot be redistributed. The artifact provides a public sample for pipeline checks and a controlled route for full-data review.
Read the full abstract ▾
Artificial Intelligence
RSI-Master: Structuring Experiments to Guide Autonomous Model Improvement
Yaxin Du, Xiyuan Yang, Zhifan Zhou et al. (+10 more)
Recursive self-improvement (RSI) seeks to enable AI systems to participate in improving their own capabilities. A concrete pathway is autonomous model development, where
agents iteratively explore post-training strategies to improve a base model. This setting faces two challenges:
agents may exploit open-ended experimental actions through hacking, and repeated experimentation may lead to strategy lock-in, where an early direction is refined rather than reconsidered. We introduce RSI-Master, which addresses the two challenges at two levels: regularize step-wise actions, avoiding hacking behaviors, and promote well-structured exploration of research directions, avoiding strategy lock-in. RSI-Master consists of an Experiment OS, which enables regularized experimental actions and maintains persistent, traceable experimental records, and Reviewer-Guided Research Orchestration, which organizes Workers and Reviewers in a dynamically growing research DAG. Workers explore diverse research directions and Reviewers compare evidence across related experiments for subsequent explorations. On PostTrainBench with Qwen3-4B-Base, it averages 54.49 versus 46.53 for the strongest
agent baseline, with a 0.0% hacking rate. Scaling to 35B model, RSI-Master surpasses the human-developed Instruct model on LiveCodeBench-v6 (41.21 vs. 37.36) and SciCode, and reaches a nonzero score on HorizonMath, a
benchmark of unsolved research problems on which most frontier models score near zero.
Read the full abstract ▾
All authors Yaxin Du, Xiyuan Yang, Zhifan Zhou, Yujie Ge, Cheng Wang, Jiajun Wang, Sijie Chen, Zehui Liu, Yuxin Zhang, Weicheng Gu, Julian Zhang, Zixing Lei, Siheng Chen
Computer Vision
WorldPlay2: Extending Real-Time Interactive World Models in Control and Horizon
Haiyu Zhang, Wenqiang Sun, Tengfei Wang et al. (+5 more)
Interactive world models require responding in real time to versatile controls and maintaining long-horizon consistency. However, modeling heterogeneous controls remains difficult, while explosive contexts and unstable
distillation impede achieving both long-horizon consistency and real-time responsiveness. In this paper, we present WorldPlay2, an interactive world model that couples a factorized hybrid control interface with a co-design of compressed
memory and stable
distillation. 1) Our factorized hybrid control interface integrates frame-aligned action control with structured semantic control that explicitly disentangles scene appearance, character identity, and dynamic semantic events, thereby facilitating effective control learning. 2) To achieve efficient long-horizon modeling, we compress historical contexts into compact
memory tokens shared by the autoregressive student and the bidirectional teacher. This design enables clip-wise,
memory-conditioned score evaluation instead of jointly processing an entire long rollout, substantially reducing
distillation overhead. 3) We further propose Stable Forcing, which initializes the autoregressive student via a few-step strategy and leverages full-rollout replay to preserve the quality of long-horizon rollouts, ensuring robust and stable
distillation. Extensive experiments demonstrate the strong generalizability of our model and its superior performance compared to existing methods.
Read the full abstract ▾
All authors Haiyu Zhang, Wenqiang Sun, Tengfei Wang, Junta Wu, Jun Zhang, Yunhong Wang, Yu Qiao, Chunchao Guo
Artificial Intelligence
From Search to Research: Exploring Search Scaling in Autonomous Quantitative Factor Mining
Kangcheng Deng, Hui Cai, Jiacheng Lu et al. (+6 more)
Inference scaling has been shown to improve large
language model (LLM) performance, and this principle naturally extends to autonomous LLM
agents through increased search budgets, which we refer to as *search scaling*. Although prior work has characterized the mechanisms, scaling behavior, and performance limits of LLM inference scaling, much less is known about these questions in autonomous research. Therefore, we investigate how search scaling affects research performance and what mechanisms drive these gains using 50 quantitative factor-mining tasks grounded in financial research reports. Each task requires an
agent to carry out an end-to-end research loop, from interpreting a hypothesis and implementing it in code to evaluating and iteratively refining the resulting factor. Across nine models, we examine how model capability, search depth, and search organization shape factor quality by tracing performance across varying budgets, transferring intermediate research states between models, and comparing different search strategies. We find that (1) initial performance is more strongly associated with model capability, while deeper search can narrow cross-model gaps; (2) model grafting shows that the early research state materially shapes final performance; and (3) parallel search outperforms sequential search under the same iteration budget, consistent with benefits from broader coverage of the search space. Further trajectory analysis shows that higher-performing models more effectively diagnose failures, revise search directions, and preserve the intended economic hypothesis when selecting candidates. These findings suggest that future progress in autonomous research will require stronger models together with adaptive policies for deploying test-time computation throughout the research process.
Read the full abstract ▾
All authors Kangcheng Deng, Hui Cai, Jiacheng Lu, Chester Zhongshu Qian, Rui Sun, Beidi Luan, Jing Li, Daxin Jiang, Zuo Bai
Cryptography and Security
The Compiler May Read It, the Agent May Not: Keeping Part of a Research Code Away from a Coding Agent
Shobhan Roy
The compiler must read modules a physics-based solver cannot build without; the coding
agent must not read that intellectual property. The harness, the software around the model that executes its tool calls, does not ship that rule. We classified fifteen read routes against a container, permission rules and a sandbox. None of the three can tell which program is reading.
Read the full abstract ▾
Data Structures and Algorithms
Competitive Random-Order Correlation k-Clustering
Mahsa Derakhshan, Andisheh Ghasemi, Rajmohan Rajaraman et al. (+2 more)
Correlation clustering has been extensively studied over the last two decades in many different computational models owing to its wide-ranging practical applications. The goal is to compute a partition of the vertex set such that the total number of disagreements, i.e. the sum of edges between clusters, and non-edges within clusters, is minimized. In this paper, we study correlation $k$-clustering, in which the total number of clusters is restricted to $k$. Correlation $k$-clustering is NP-hard, and while previous work has presented a
polynomial time approximation scheme for constant $k$, there are no known results for general $k$. Our first result is a
polynomial-time constant-factor approximation
algorithm for correlation $k$-clustering for general $k$. The main focus of this work is in the more challenging online setting. Noting that the best competitive ratio under
adversarial arrivals is known to be $\Omega(n)$, we concentrate on the well-studied random-order model, where vertices arrive in random order and on arrival of a vertex, edges to its earlier-arrived neighbors are revealed. We prove a surprising lower bound of $\Omega(\log k)$-competitiveness for any online
algorithm, which can be extended to $\Omega(\log n)$ when $k = \text{poly}(n)$. Finally, the main result of this paper is a polylogarithmic upper bound on the competitive ratio for correlation $k$-clustering, using an
algorithm inspired by the classic Pivot
algorithm for correlation clustering.
Read the full abstract ▾
All authors Mahsa Derakhshan, Andisheh Ghasemi, Rajmohan Rajaraman, Omer Wasim, Tegan Wilson
Multiagent Systems
Positional choice and robust collective behavior in fish schools: biohybrid experiments and modeling
Vahagn Grigoryan, Donato Romano, Cesare Stefanini et al. (+1 more)
Collective behavior of fish schools is usually modeled on the assumption that each individual follows specific rules of motion that depend on its position and velocity relative to its neighbors. Although these models reproduce many schooling patterns observed in nature, it remains unclear whether the assumed rules are realistic at the individual level. To address this question, we first analyzed a set of experiments in which a live fish interacted with four moving robotic fish in a tank, and measured the time it spent in each position relative to the
robots. We then simulated this experiment, replacing the fish with an
agent, and assessed the extent to which classical models agree with the experimental observations. An extensive exploration of the parameter space showed that these models closely matched the experimental observations, with a very specific choice of parameters. However, they were highly sensitive to the parameter values: a minor perturbation caused them to fail completely. To resolve this, we propose a new model that incorporates an additional exploratory term characterizing the
agent's positional preference at every moment. This model proved robust to small errors in the parameters while successfully replicating the experiments. Furthermore, when generalised to multiple fish, the model reproduced schooling behaviour and common schooling patterns, while also exhibiting exploratory behaviour at both the individual and the school level. This shows that social cohesion coexists with individual exploratory behaviour, and that realistic models of collective behaviour should account for both social interactions and this exploratory component.
Read the full abstract ▾
All authors Vahagn Grigoryan, Donato Romano, Cesare Stefanini, Giulia De Masi