Artificial Intelligence
Mathematical Transfer in LLMs Follows Reasoning Approach More Than Topic
Sajad Goudarzi, Samaneh Zamanifard, Seyed Amin Seyed Haeri et al. (+2 more)
When selecting mathematical training data for LLMs, a natural organizing principle is topic: probability examples for probability targets. An alternative is
reasoning approach: worked solutions that share a solution method with the target, even when the mathematical domain differs. We ask which relation produces greater transfer after fine-tuning. We evaluate two counterbalanced $2\times2$ designs: probability and combinatorics crossed with invariant
reasoning and double counting (2,000 problems), and number theory and geometry crossed with complement and pigeonhole
reasoning (800 problems). In each design, every
cell serves as the held-out target in turn: same-approach (SA) sources share the target's method but change the topic, while same-topic (ST) sources share the topic but change the method. Every source appears once in each role, so additive source-quality effects cancel from the equally weighted aggregate contrast. Across five base models and three training seeds per design, SA outperforms ST in all 40 seed-pooled model–target comparisons. Model-level advantages range from 8.2 to 16.2 percentage points in the primary design (mean: 10.8) and from 12.0 to 16.0 in the second design (mean: 14.3); all ten model-level 95% confidence intervals exclude zero. In both designs, ST sources are more similar to targets under
embedding and lexical measures, so the SA advantage runs opposite to the measured ordering of statement-level resemblance. These findings identify
reasoning approach as a more effective matching criterion than topic for mathematical transfer across the evaluated topic–approach combinations.
Read the full abstract ▾
All authors Sajad Goudarzi, Samaneh Zamanifard, Seyed Amin Seyed Haeri, Moloud Nasiri, Hamed Rahimian
Robotics
Retrospective Open-Vocabulary Memory for Long-Term Object Search
Jiaming Wang, Zhiwei Xue, Chen Jizhuo et al. (+2 more)
Long-term object search requires learning where objects usually appear from repeated but uneven observations of a changing environment. We formulate retrospective open-vocabulary
memory as probabilistic inference from censored observations, where the key idea is to reason with evidence per opportunity: a detection or non-detection should influence belief only in proportion to the
robot's opportunity to observe the corresponding location. We introduce ECROM, which uses this principle to estimate long-term prevalence for concepts specified only at query time and converts the resulting belief directly into an active-search prior. To evaluate this problem, we introduce a controlled long-term
benchmark in ten HM3D homes that independently varies object placement and observation opportunity across repeated traversals. ECROM improves support-level AP on held-out queries by 4.5 points and search SPL by 4.2 points over the strongest competing
memory in each metric. The
benchmark,
dataset, and code will be open-sourced. Project page: https://jiaming.im/ecrom/
Read the full abstract ▾
All authors Jiaming Wang, Zhiwei Xue, Chen Jizhuo, Peng Shiqi, Harold Soh
Machine Learning
Non-Commutative Lie-Algebraic State-Space Models: Finite-Division Variety Laws, Semidirect Cascades, and Gradient Saddle Escape
Zeyu Jia
Structured state-space models (SSMs) achieve linear-time inference and logarithmic-time parallel training via associative scans, yet homogeneous diagonal transitions generate solvable
semigroups unable to track non-solvable group words or 3D rigid-body
manifolds. We develop a theoretical and empirical framework for Non-Commutative Lie-Algebraic State-Space Models (Lie-SSMs), parameterizing recurrent updates on Lie
algebra tangent spaces via the
matrix exponential. First, via an o-minimal multi-rate cylindrical flag collar, we establish an unconditional single-layer finite-division criterion for homogeneous contracting $SO(3)$ and $SU(2)$ heads, proving odd-
prime Sylow abelianity, odd-$p$-core centrality $[u^2,x]=e$, exponent-30 commutator laws ($[u^{30},v^{30}]=e$ in $SO(3)^m$ and $[u^{30},v^{30}]\in Z(G),[u^{30},v^{30}]^2=e$ in $SU(2)^m,SO(4)^m$, excluding $S_5$), and sharp threshold $d^*(S_5)=d^*(A_6)=5$ in $SO(5)$. Second, Depth-2 semidirect and Schreier $2$-cocycle cascades resolve $S_5$, $2I\cong SL_2(F_5)$, and $\mathrm{Heis}(F_p)$ in $O(\log T)$ depth, whereas unnormalized contracting coupling collapses exponentially without Directional RMSNorm. Third, we characterize the $O(\varepsilon^{c-1})$ Chen–Magnus–
Hall–Witt commutator saddle and $1/|[G,G]|$ coset-centroid dilution, proving $\varepsilon=0$
symmetry breaking and local Polyak–Łojasiewicz convergence ($\theta_{\max}<\pi$), solving $S_5$ ($99.90\%/99.40\%$ at $T=128$) and $D_{16}$ ($100.00\%$) via dense prefix loss and factored readouts. Across $Q_8,A_5,S_5,D_{16}$, Dyck-2, $SE_2(3)$ Left-Jacobian odometry, GPU scans ($26.03\times$ speedup), and three 3D IMU cohorts (UCI HAR, HAPT-PT 30-subject CV, and REALDISP), Depth-2 Lie-$SO(3)$ cascades reach $95.69\%\pm 0.14\%$ on HAR, $87.16\%\pm 0.64\%$ ($88.32\%\pm 0.93\%$ 30-subject episodes) on HAPT-PT, and $84.23\%\pm 1.00\%$ under displacement on REALDISP.
Read the full abstract ▾
Artificial Intelligence
ContractRL: Shielded Group-Relative Policy Optimization for Auditable Tool-Call Repair
Miaobo Hu, Shuhao Hu, Xiaobo Guo et al. (+4 more)
Structured tool calls often fail after only a small number of fields violate a schema or an execution contract. Regenerating the complete object enlarges the action surface and makes repeated repair difficult to audit. We introduce ContractRL, a contract-constrained sequential repair protocol that models verifier-guided JSON repair as a bounded decision process. At each step the policy observes the candidate, typed verifier feedback, JSON Pointer, immutable repair history, and remaining budget; a contract-derived action mask filters malformed or prohibited RFC-6902 operations before a deterministic validator performs the transition. We specify a contract-constrained group-relative objective for patch, retry, and abstention decisions while keeping canonical targets and semantic labels outside the online state until trace freeze. Under identical verifier information, ContractRL attains 0.9362 semantic success with 34.4 generated
tokens, compared with 0.9076 and 44.9
tokens for Patch-SFT and 0.9148 and 137.2
tokens for full regeneration over 192 cases per seed and five seeds. Policy optimization improves semantic success from 0.9186 for supervised ContractRL to 0.9375. A separate three-seed paired evaluation against Patch-SFT yields a semantic difference of +0.0396 (95% CI $[+0.0137,+0.0662], p=0.0039$). Feedback, action-mask, budget, and schema-shift analyses connect these gains to localized correction, while
adversarial and multi-turn evaluations characterize the remaining failure modes.
Read the full abstract ▾
All authors Miaobo Hu, Shuhao Hu, Xiaobo Guo, Xin Wang, Bokun Wang, Yina Sa, Daren Zha
Cryptography and Security
Actions with Receipts: Jointly Binding Claims, Evidence, and Execution for Replayable Tool-Agent Auditing
Miaobo Hu, Shuhao Hu, Xiaobo Guo et al. (+4 more)
Tool-using
agents can expose citations and execution logs while leaving a critical association unaudited: whether the claim shown to a user is the claim emitted by the committed execution and supported by the cited source. A valid citation and a valid trace can therefore remain individually well formed while being transplanted across claims, actions, runs, or source versions. We introduce a claim-anchored execution contract that jointly binds the emitted claim, its exact source span, the ordered execution prefix that produced it, and the source version and access state observed by that execution. Each receipt contains an emission anchor that deterministically locates the claim inside a committed answer or claim-bearing action, together with source identifiers, offsets, hashes, quotes, and a domain-separated execution commitment. A deterministic integrity verifier reconstructs these bindings before semantic or task labels are joined. We separate this integrity plane from a pluggable support plane, so structural validity is not used as a proxy for entailment. The contract exposes seven independently testable properties: claim-emission binding, source binding, ordered-execution binding, oracle separation, persisted-object replay, execution-rerun consistency, and version/access binding. Across 1,280 cross-object attacks, the joint contract detects 1,275 substitutions (0.9961). Removing a targeted property reduces its attack-detection rate to 0.0156-0.0625. On an independently adjudicated 384-pair split, the conflict-aware support guard reaches F1 0.8865 and false acceptance 0.0729; on unseen failure families, these rates are 0.8679 and 0.0938.
Read the full abstract ▾
All authors Miaobo Hu, Shuhao Hu, Xiaobo Guo, Xin Wang, Bokun Wang, Yina Sa, Daren Zha
Machine Learning
Bellman-Certified Rounding for Sparse Policy Deployment in MDPs
Zhaojun Peng
Continuous policy optimization may spread an update across many states, even when deployment permits only a few complete state-level changes. We study how much discounted return can be retained when continuous row mixtures are rounded to sparse binary policies in finite MDPs. Policy-dependent visitation couples the row edits, while long horizons make global
curvature bounds conservative. From $2d+2$ Bellman solves, we derive reusable envelopes that support uniform and candidate-specific guarantees before rounding. A rank-two rational representation of each exchange further permits weighted
curvature integration along the realized trajectory. We prove that linear dimension dependence is unavoidable when the budget scales, and that exact global
curvature thresholding is hard. Candidate-specific bounds raise pre-rounding certification coverage from $48.2\%$ to $74.1\%$ on the structured suite. At $\gamma=0.95$, local integration lowers the median bound-to-loss ratio from $402.3$ to $2.08$ on coupled instances.
Read the full abstract ▾
Computation and Language
CAST: Cost-Aware Speculative Trees from One-Pass Block Drafters
Jungseob Lee and Sugyeong Eo
Speculative decoding accelerates large
language model inference by drafting future
tokens cheaply and verifying them with the target model in parallel. Block drafters score a whole block of future
tokens in one forward pass, yet standard decoding verifies only the top-scoring chain and discards the other candidates. Because these candidates are already scored, verifying more of them adds target computation but no extra drafting. We introduce CAST (Cost-Aware Speculative Trees), which packs these candidates into a tree and verifies it in a single target pass, leaving the target model, drafter weights, and decoding rule untouched. To decide how wide the tree should be, CAST adds candidates while the expected gain from the next one outweighs the verification time it adds. The width therefore adapts to each deployment from a latency measurement, without sweeping over widths. We evaluate CAST across five domains on three GPU generations and two model families. At its predicted width, CAST is faster than the standard chain in all eight settings, by up to 43%. We also find that the best width depends strongly on the deployment. Where verification cost jumps at a kernel boundary, a 128-
token tree is only 2% faster than the standard chain, whereas the tree at the predicted width is 20% faster. Furthermore, we prove that CAST leaves the target output distribution unchanged under both greedy and sampled decoding. Code is available at https://github.com/js-lee-AI/CAST.
Read the full abstract ▾
Computation and Language
Refusal Localizes, the Damage Relocates: Safety Layers Under Few-Sample Fine-Tuning
Jungseob Lee, Dongyub Jude Lee, Sugyeong Eo et al. (+3 more)
Fine-tuning adapts aligned large
language models (LLMs) to downstream tasks, but a few dozen harmful examples can remove their refusal of harmful requests. Prior work localizes safety-related behavior to specific layers, directions, and
tokens, suggesting targets for protection. We test whether successful localization and recovery support defenses that survive changes in the attack. Across six checkpoints from four model families, harmful and benign
prompts remain linearly separable after attack, and patching full clean hidden states into the compromised model restores refusal at a reproducible transition depth. Building on a prior layer-freezing defense, we freeze every layer up to this depth and repeat the attack. At a hundred harmful examples, refusal remains near zero on all six checkpoints, with recovery transitions above the frozen boundary. In a second study, removing the update's top two singular directions restores refusal after short
attention-only fine-tunes on four checkpoints. On Llama-3.1-8B, ordinary training changes weaken this repair and an attacker who spreads the update defeats it. A spectral detector calibrated on benign Llama fine-tunes misses most repair failures on that checkpoint. Localized freezing can nevertheless help preserve refusal when a few harmful examples enter training data unintentionally. These results show that an attacker can bypass a region identified by recovery and defeat a repair that works across multiple checkpoints, motivating five checks for defenses against adaptive fine-tuning. Code is available at https://github.com/js-lee-AI/refusal-relocates.
Read the full abstract ▾
All authors Jungseob Lee, Dongyub Jude Lee, Sugyeong Eo, Seongtae Hong, Seungyoon Lee, Heuiseok Lim
Computer Vision
EgoRefine: Ego-Referenced Predictive Alignment and Trajectory-Conditioned Reliability-Aware Fusion for Asynchronous Collaborative Perception
Lingzhao Kong, Yongsheng Zang, Yu Kang et al. (+4 more)
Collaborative perception enables connected
agents to share complementary observations for 3D object detection, extending sensing range and mitigating occlusion. Under asynchronous communication, however, cooperative features arrive with temporal delay. Existing prediction-based methods compensate for these features mainly from the transmitting
agent's own history, leaving residual misalignment with the ego
agent's current observation; subsequent fusion also often overlooks spatial variations in
alignment quality. We propose EgoRefine, an ego-referenced predictive
alignment and reliability-aware fusion framework for asynchronous collaborative perception. Its Ego-referenced Predictive
Alignment module uses the current ego feature to guide cooperative trajectory-field prediction and refines the sampling offsets along an ego-referenced trajectory direction. Its Trajectory-conditioned Reliability-aware Fusion module treats the trajectory discrepancy between the ego and cooperative streams and the directional refinement magnitude as
alignment cues, using them to condition the relation between aligned features and adaptively reweight the two streams before convolutional fusion. Experiments on V2V4Real and DAIR-V2X-Seq show that EgoRefine outperforms TraF-Align by 1.6 and 2.9 points on average in AP@0.5 and AP@0.7, respectively. The source code will be made publicly available at https://github.com/godk0509/EgoRefine.
Read the full abstract ▾
All authors Lingzhao Kong, Yongsheng Zang, Yu Kang, Kailun Yang, Jie Fu, Yukun Zuo, Zhiyong Li
Computer Vision · Cross-listed from eess.IV
LensBridge: Frequency-Guided Compound Degradation Adaptation for Lens Aberration Correction and Veiling Glare Removal
Xiaolong Qian, Zhonghua Yi, Qi Jiang et al. (+4 more)
Simplified optical systems often exhibit residual lens aberrations and Veiling Glare (VG), resulting in spatially varying blur and contrast reduction. Large-scale Lens Libraries (LensLib) enable reusable aberration correction models by covering diverse Point Spread Functions (PSFs), but their aberration-only training distribution does not include target-specific veiling glare. Extending such foundations to compound degradation is challenging because realistic target-system compound pairs are difficult to obtain. To address this challenge, we propose LensBridge, a two-stage framework that first establishes a reusable aberration correction foundation and then adapts it to compound optical degradation using only a few unpaired target observations. In Stage I, we build a PSF-aware one-step
diffusion foundation by constructing discrete degradation priors from LensLib PSFs and learning to retrieve them directly from aberrated images, enabling PSF-aware correction without requiring explicit PSF at inference. In Stage II, we adapt this foundation to compound degradation through frequency-domain guidance. At the data level, Frequency-guided Degradation Completion (FDC) transfers target low-frequency characteristics to LensLib aberrated images while preserving aberration structures to synthesize compound training pairs; at the model level, Frequency-guided Pseudo Decomposition (FPD) forms aberration- and VG-dominant pseudo observations to condition separate adaptation branches. Extensive experiments across multiple optical systems demonstrate that LensBridge effectively extends reusable aberration correction foundations to joint aberration correction and veiling glare removal without target-system paired supervision. All code will be available at https://github.com/XiaolongQian/LensBridge.
Read the full abstract ▾
All authors Xiaolong Qian, Zhonghua Yi, Qi Jiang, Kailun Yang, Shuhang Xie, Shaohua Gao, Kaiwei Wang
Robotics
DriftOPD: Sequence-Level Reverse-KL Distillation for One-Step VLA Policies
Youngjun Jun, Kyumin Choi, Youngmin Kim et al. (+4 more)
Vision-
Language-Action (VLA) models increasingly rely on action experts that generate short action chunks under receding-horizon control. While chunk-level training is convenient across
robot embodiments, it optimizes local action likelihood without explicitly accounting for long-horizon task success. Sequence-level reinforcement learning can address this limitation, but typically requires policy rollouts and closed-loop interaction, which are costly for real-
robot manipulation. We introduce DriftOPD, a teacher-free, rollout-free framework for sequence-level on-policy
distillation of continuous VLA action experts. We show that the sequence-level reverse Kullback-Leibler (KL) divergence decomposes into a chunk-level reverse-KL term and a future-potential term that captures the long-horizon effect of the current action. DriftOPD optimizes these two terms using a one-step drifting objective and a Q-function critic learned from offline demonstrations, respectively, enabling sequence-level optimization with only offline data and one-step action generation. Across multiple VLA architectures in simulation and real-world manipulation, DriftOPD generally outperforms existing one-step
distillation baselines while achieving task success performance comparable to multi-step teacher policies. These results demonstrate that long-horizon behavior can be effectively distilled into one-step VLA action experts without online interaction or a separate teacher.
Read the full abstract ▾
All authors Youngjun Jun, Kyumin Choi, Youngmin Kim, Seonghyun Jin, Sunwoo Park, Jangho Park, Jong Chul Ye
Computation and Language
DuplexSpeechBench-Document Grounding: Benchmarking Document Grounding and Hallucinations in Voice Agents
Puneet Mathur, Nedim Lipka, Zeyu Jin et al. (+1 more)
Voice
agents enable low-latency, natural interaction, yet their ability to faithfully ground responses in external documents remains underexplored. We introduce DuplexSpeechBench-Document Grounding (DSB-DG), a
benchmark for evaluating document grounding in voice
agents across five professional domains. DSB-DG targets three failure modes: Context Saturation, which measures grounding under increasing document length; Grounding Decay, which measures retention of document facts across multi-turn dialogue; and Proactive Grounding, which evaluates whether context re-injection mitigates conversational drift. The
benchmark contains 1,636 adversarially verified QA pairs from 50 documents covering five professional domains, and supports fully automatic evaluation of grounding accuracy,
hallucination, and response latency. Across systems spanning cascaded, proprietary full-duplex and real-time, and open-weight
speech2
speech architectures, we find substantial differences in effective grounding capacity. While cascaded pipeline (ASR-LLM-TTS) achieves the highest grounding accuracy, Gemini-Live and GPT-Realtime closely trail behind. Open-weight systems exhibit distinct failure modes, most notably an abrupt context-capacity collapse and multi-turn grounding decay. More broadly, grounding fidelity degrades with context and conversational load, and failures frequently manifest as unsupported generations rather than abstention. We show that contextual grounding as a key unresolved challenge for reliable full-duplex voice
agents.
Read the full abstract ▾
All authors Puneet Mathur, Nedim Lipka, Zeyu Jin, Dinesh Manocha
Computer Vision
Beyond Pixel Reconstruction: Retrieval-Guided Glyph-Aware Restoration for Low-Resource Manchu Historical Documents
Ting Huang, Dongdong Wang, Mingqiu Liang et al. (+1 more)
Historical Manchu documents preserve invaluable linguistic and cultural heritage, yet their digitization is hindered by severe degradations and the scarcity of paired training data. Existing document restoration methods primarily optimize pixel-level reconstruction, which can produce visually plausible results while failing to preserve the structural identity of Manchu glyphs. To address this limitation, we propose a
retrieval-guided glyph-aware restoration framework that goes beyond pixel reconstruction by explicitly incorporating glyph-level structural knowledge. Our method retrieves relevant glyph exemplars to provide structural guidance during restoration and integrates this information into the reconstruction process, improving the recovery of degraded character structures under low-resource conditions. Extensive experiments on Manchu historical documents demonstrate that the proposed approach improves both image restoration quality and glyph-level fidelity compared with existing restoration methods. These results highlight the importance of incorporating character-aware structural priors for reliable restoration of low-resource historical documents.
Read the full abstract ▾
All authors Ting Huang, Dongdong Wang, Mingqiu Liang, Siyang Lu
Artificial Intelligence
Predictive Credit: Measuring What Scientific Explanations Add to Experimental Forecasts
Jingjie Ning, Xueqi Li, Yibo Kong et al. (+1 more)
Research
agents explain planned experiments. We measure predictive credit with paired forecasts sharing an intervention, forecaster, and outcome while varying description, matched explanation, and donor context. Five checks track commitment, delivery, predictive gain,
alignment, and known-signal uptake. Across 336 prospective states in controlled learning, 12 Tox21 endpoints, and 24 OpenML tasks, v5's frozen credit decision was inconclusive. Tox21's preregistered ROC AUC interval-score harm test was unmet ($D-M=-.0026$, 95 percent interval [$-.0174$, .0104]); OpenML's joint formation, point-equivalence, and repeatability rule was unmet. Matched point-accuracy gains over description remained unconfirmed, and Tox21/OpenML seed-donor intervals spanned zero. Under requested DeepSeek V4 Pro, matched and donor cards reduced secondary Tox21 drift by 64.5 and 59.1 percent. A DeepSeek V4 Flash replay raised matched point MAE from .01823 to .02020 and missed matched-donor interval-score equivalence. OpenML full-card assignment widened nominal 80 percent intervals by 21 percent, with 49.3 percent coverage versus 51.4 percent for description and content in 66/144 cards. Direct-text Flash delivered all 144 notes without detectable matched point-accuracy gain. A researcher-authored mechanism positive control lowered point MAE by 2.60 percentage points versus description. The protocol measures predictive credit for research-
agent benchmarks and scientific
forecasting; natural-explanation credit remained unconfirmed at the tested donor resolutions.
Read the full abstract ▾
All authors Jingjie Ning, Xueqi Li, Yibo Kong, Dongting Li
Artificial Intelligence
Rules to Tools: Executable Checks for LLM Agents in Scientific Computing
Jingjie Ning, Guojiang Zhao, Chen Xu et al. (+4 more)
Scientific coding
agents receive equations, boundary conditions, and output requirements in writing, then must assess the programs they revise. Rules to Tools (R2T) supplies prepared executable checks of public scientific requirements. Matched SciCode repair groups share written checks, starting programs, model, and budgets; the tool group receives a callable implementation. Across two task-ID cohorts, complete repair is 26/30 with text and 29/30 with the prepared checks. Three task IDs favor tools, one favors text, and eleven tie. The eight-ID cohort scores 13/16 versus 15/16, with a task-cluster bootstrap 95% interval of [-12.5, 43.75] percentage points for the difference. The larger shared-definition SciCode cohort ties at 13/24 per group. Five development-exposed tasks with alternate starting programs score 3/10 versus 7/10. The tool group favors tasks 17, 77, and 11; initial checks flag task 17 and report no violation for tasks 77 and 11. Task 37 favors text and has no initial reported violation. A fresh source-through-Python arm also reaches 15/16, matching the dedicated command's aggregate. In a matched PDE comparison, detailed text scores 23/24 and checks score 24/24, with 31.2% lower reported model output for checks.
Agent-side output savings vary by cohort, while public CPU use rises in both task-ID cohorts. These results measure task-dependent repair outcomes and
agent-side costs with prepared checks.
Read the full abstract ▾
All authors Jingjie Ning, Guojiang Zhao, Chen Xu, Shanshan Zhong, Xiaochuan Li, Ji Zeng, Guolin Ke
Machine Learning
In-Distribution Imagination for Model-Based Offline Reinforcement Learning
Mintae Kim and Koushil Sreenath
Model-based offline reinforcement learning (MBORL) improves sample efficiency through model-generated trajectories. However, accumulative model error can drive imagined trajectories outside the offline data distribution, leading to unrealistic synthetic data and unstable policy optimization. Many existing methods primarily control rollouts using transition-level uncertainty. We propose in-distribution imagination (IDI), a rollout control framework that estimates trajectory support in a learned representation space and adaptively truncates rollouts that leave the offline trajectory
manifold. Combined with trajectory-regularized RL, an extension of
entropy-regularized RL, IDI consistently improves performance in limited-data settings. Experiments show that trajectory support predicts rollout failure substantially better than transition-level uncertainty, highlighting the importance of trajectory-level rollout control in MBORL.
Read the full abstract ▾
Machine Learning
Provable Test-Time Scaling for Beam Search in LLM Reasoning
Qijia He, Yu Huang, Yuan Cheng et al. (+2 more)
Beam-search-based test-time methods provide an effective way to improve large
language model (LLM) performance on long-horizon generation by
pruning invalid
reasoning paths early, leading to significantly improved
reasoning efficiency and more favorable test-time cost scaling. Despite strong empirical success, the theoretical understanding of beam search remains limited. In this paper, we study the test-time compute guarantee of the commonly used beam search framework that uses the model's internal log-likelihood for intermediate scoring, while relying on an external reward model only after a complete response is generated. We first establish a lower bound for vanilla beam search, showing that at least $\Omega(C^\star(x)^2)$ samples are required for the optimal response to survive, where $C^\star(x)$ is the
token-level coverage coefficient for
prompt $x$. This motivates our modified confidence-filtered beam search (CF-Beam), which reduces the sufficient coverage dependence from quadratic to nearly linear under prefix competitiveness, for fixed horizon, gap, and target accuracy. We then show that the regret of CF-Beam is upper-bounded by the probability of rare failure events and the reward estimation error scaled by a path-level coverage coefficient, where the rare-failure term vanishes as per-step sampling increases. Our results highlight a fundamental advantage of beam search over sequence-level inference methods such as Best-of-N and Best-of-Majority. While the guarantees of these approaches typically involve coverage coefficients that grow exponentially with the horizon $L$, CF-Beam controls the dominant search-induced term through a
token-level coverage coefficient that scales polynomially with $L$. Our numerical experiments further confirm that beam search is more robust on hard instances and under increasing
reasoning horizons.
Read the full abstract ▾
All authors Qijia He, Yu Huang, Yuan Cheng, Yuxin Chen, Yingbin Liang
Artificial Intelligence
Where Scientific Search Agents Fail: Decision-Checkpoint Auditing of Exposure and Inspection Attempts
Hongmin Li and Wanli Zhao
Final-answer accuracy does not reveal whether a scientific-search
agent failed to encounter a target paper, attempt to inspect it, or return an accepted answer after inspection. We introduce decision checkpoints that record observations and tool actions without
benchmark labels during inference, then join target identities and evaluator labels to assign outcome categories from recorded events. Across five conditions on 540 answerable AutoResearchBench Deep questions in a fixed, target-enriched environment, keyword search achieves 24.6% accuracy, compared with 17.8% for raw search. The keyword condition has fewer incorrect answers with neither target exposure nor inspection, but more incorrect answers after the target is exposed and left uninspected. Compared with keyword search, read-first has 27.4% more recorded evidence-search calls. Target inspection attempts occur on 199 questions under read-first and 191 under keyword search; both conditions achieve 24.6% accuracy. The checkpoint protocol makes these question-level differences explicit, distinguishing target exposure and inspection from aggregate accuracy and total tool use.
Read the full abstract ▾
Cryptography and Security
Z-Sigil: A Public-Key Cryptosystem with Chained Selection over a Fiber Bundle of Module-Lattice Keys
Andrea Rondelli
Z-Sigil is a public-key cryptosystem in which the plaintext selects successive keys from a fixed module-
lattice family. Messages are length-prefixed, zero-padded and divided into 32-byte blocks. Each public vector is a shared public
matrix applied to a small secret vector, plus a small error. Key indices label torsion points of a flat Kähler
torus; the secret family forms a section of a key bundle over them. A public nonce initializes a hash state that selects each block's key and bit mask. The sender updates the state with the plaintext block; the receiver does so after recovering it. The stream and nonce determine a discrete walk along which decryption reads the secret section. We specify the
algorithms, prove correctness under an explicit noise condition and bound decoding failure for messages chosen after the public key. Under stated decisional Module-LWE assumptions, we establish IND-CPA confidentiality for the chain without modelling the state hash as a random oracle. The reduction covers
quantum adversaries under
quantum hardness assumptions, with classical keys, messages and ciphertexts; no concrete security level is established. With independent uniform selectors, a restricted direct-decryption model quantifies reduced fragment recovery under partial key exposure, without improving full-message recovery probability over an independent-block baseline. Known-plaintext and candidate-message attacks, and parallel candidate-table decryption, delimit this result. Neither a universal sequential lower bound nor authentication or chosen-ciphertext security is established. Replacing an earlier scalar-exposing proposal, we give an augmented-
lattice interpretation, integral-transport obstructions and a noise budget for research on curved, nontrivial bundles. A byte-level specification, pseudocode, test vectors and numerical checks support verification.
Read the full abstract ▾
Machine Learning
Understanding Off- vs On-Policy Distillation: A Tale of Distinct Training Objectives
Qiwei Di, Xuheng Li, Kaixuan Ji et al. (+3 more)
On-policy
distillation (OPD) learns from teacher feedback on student-generated responses and has shown promise in reducing forgetting relative to supervised fine-tuning (SFT). However, its benefits and fragility remain incompletely understood. We study sequential
distillation from multiple teachers, where the student minimizes its average divergence from the teachers. Forward Kullback–Leibler (KL) divergence yields a weighted arithmetic mixture, while reverse KL yields a normalized weighted
geometric aggregate. We develop
algorithms that learn these targets under off-policy and on-policy feedback, respectively, establishing logarithmic regret bounds in the tabular setting and extending the analysis to function approximation. By analyzing these aggregation targets, we identify mechanisms that help explain both the benefits and fragility of OPD. Relative to forward KL, reverse KL can better retain a confident expert's preferences under uninformative feedback, but is more sensitive to teachers that assign very low probabilities to correct responses. Its
token-level conditionals also reveal a dependence on continuation distributions that can favor incorrect prefixes over long horizons.
Read the full abstract ▾
All authors Qiwei Di, Xuheng Li, Kaixuan Ji, Chenggong Zhang, Heyang Zhao, Quanquan Gu
Neural and Evolutionary Computing
How much of fly walking is written in the wiring?
Isabel Guan, Yuntian Zhao, Dingyuan Zhang et al. (+2 more)
Connectome models of the fly nerve cord generate walking-like motor rhythms, but oscillation alone does not show that the specific wiring matters. Here we provide, to our knowledge, the first test of which features of motor output depend on the specific wiring. We simulated the leg motor systems of two independent Drosophila connectomes, with synapse counts as fixed weights and glutamatergic synapses treated as inhibitory, and compared each with six families of rewired networks that preserve progressively more of its structure, using pre-registered criteria. We find that rhythm is generic but antagonist coordination is not: many rewired networks were more rhythmic than the real ones, yet the real wiring coordinated antagonistic motor pools more strongly than every rewired network, most of all at the thorax–coxa joint. We trace this specificity to how premotor input is allocated between antagonistic pools. Both connectomes carry Sherrington's reciprocal innervation—
neurons that excite one pool inhibit its antagonist—and no rewired network does. Reassigning premotor inputs between the pools abolished coordination even when motor
neurons' typical input strength changed little (all pre-registered criteria met in one connectome; same direction in the other). Coordination, not rhythm, therefore reveals whether a connectome model's wiring matters.
Read the full abstract ▾
All authors Isabel Guan, Yuntian Zhao, Dingyuan Zhang, Shipeng Lyu, I-Ming Chen
Multiagent Systems
CollabFlow: Recursive Self-Improvement of Agent Collaboration
Xiao Huang, Mingda Zhang, Junming Zhang et al. (+5 more)
Recursive self-improvement (RSI) lets a system improve from its own outcomes; in LLM-based multi-
agent systems,
Agents refine one another within a task, and outcomes improve how they collaborate across tasks. However, existing multi-
agent collaboration leaves this loop open: collaboration is pre-defined at the operator level,
topology-only learning keeps verbatim exchange that propagates errors, and reward maximization on a system's own outcomes concentrates on a few teams. To address these challenges, we propose CollabFlow, an RSI system of Learned
Agent Collaboration: a trainable Collab-Director constructs teams of complete
Agents, a frozen executor runs them, and each round's outcomes retrain the director. Within each round, the edges of a collaboration graph carry protocols of Evidence-Conditioned Communication: a receiver adopts a differing answer only when the sender's evidence is stronger by a margin, so the director learns who communicates and how. Across rounds, we further propose Collaborative Trajectory Balance (CTB), a flow-based objective that credits each team once across its construction orders and targets a reward-proportional distribution over teams, so several good teams stay in play. We also bound how far this self-generated target moves between rounds, which shrinks as records accumulate. On twelve
datasets, CollabFlow outperforms all baselines and keeps improving across rounds. Code is available at https://anonymous.4open.science/r/CollabFlow-631E.
Read the full abstract ▾
All authors Xiao Huang, Mingda Zhang, Junming Zhang, Qiang Huang, Hanwen Zhang, Yue Dai, Zijia Wang, Xiaoying Tang
Artificial Intelligence
EvoSteer: Online Self-Evolving Graph Orchestration via Reference-Anchored Credit Assignment
Mingda Zhang, Hanwen Zhang, Qiang Huang et al. (+5 more)
In recent years, LLM-based multi-
agent systems have been widely applied to orchestrate tool-using
agents into executable communication graphs. However, existing self-evolving orchestration still faces key challenges, including post-hoc
evolution that revises the team only after the trajectory ends, credit
diffusion that gives every action the same terminal advantage under confounded baselines, and skill admission that is uncalibrated and never retired. To address these challenges, we propose EvoSteer, a new paradigm of Online Self-Evolving Graph Orchestration – the orchestrator builds a running team and repairs its plausible but failing steps from execution features and a learned value estimate. To support this paradigm, we introduce Anchored Trajectory Balance (AnchorTB), a regression-style flow-matching loss that assigns each orchestration action a coefficient by balancing subtrajectories against a frozen reference. Built on the learned flow, we further propose Validated Skill Admission, in which a candidate skill is tried before promotion and promoted only if paired evidence passes a sequential test under a shared nominal testing budget. Moreover, AnchorTB combines measured task-level reference reward statistics with prefix-dependent corrections. Experimental results on twelve
datasets show that EvoSteer significantly outperforms baselines across question answering, mathematical
reasoning, code generation, and interactive decision making. Our code is available at https://github.com/beita6969/evosteer.
Read the full abstract ▾
All authors Mingda Zhang, Hanwen Zhang, Qiang Huang, Zijia Wang, Pengfei Guo, Yuchen Zhang, Jionghao Zhu, Xiaoying Tang
Computation and Language
Breaking Babel: A Self-Evolving Multi-Agent System for Long-Form Subtitle Translation
Haibo Jin, Xinjie Li, Najmeh Sadoughi et al. (+4 more)
Long-form subtitle translation requires
reasoning over discourse and cultural context spanning episodes or entire series, while maintaining consistent terminology and style. Existing single-LLM methods are largely sentence-level, and multi-
agent systems often use static workflows that do not adapt to scene complexity or production context. We propose SMART, a Self-evolving Multi-
Agent system for long-foRm subtitle Translation. During test-time training, SMART builds persistent series-level
memory and translates a subset of sentences through a dynamic router and Mixture-of-
Agents layer with tools for terminology verification, subtitle constraint validation, and contextual
retrieval. A judge-refiner loop scores candidates and uses textual critiques to update
agent prompts and routing policies without retraining the underlying LLMs. During test-time inference, the evolved configuration translates the remaining series. We also introduce Subtitle Arena, covering 14 genres, 2–198 episodes per series, production years 1959–2023, and 15 target locales, together with SubMQM, a subtitle-adapted MQM framework with seven dimensions and 19 error categories. SMART achieves the best overall MQM score in all 15 Subtitle Arena directions, reducing average penalty by 6.9% over the strongest competing
agent system. On a public
benchmark, MuSC, SMART obtains the best model result across all 4
language pairs. SMART also achieves the best result in human evaluation with an overall score of 4.50/5.
Read the full abstract ▾
All authors Haibo Jin, Xinjie Li, Najmeh Sadoughi, Yang Liu, Yibo Wang, Zhu Liu, Yuzong Liu
Artificial Intelligence · Cross-listed from stat.ML
Bandits with Multiple Optimal Arms: Minimax Regret and Non-Adaptivity
Kaixuan Ji, Qiwei Di, Qingyue Zhao et al. (+2 more)
We study multi-armed bandits (MAB) with multiple optimal arms, motivated by the fact that many practical decision making problems admit multiple correct answers. For $K$-armed bandits with $A$ optimal arms, we first provide a sharper analysis of previous sub-sampling
algorithms (De Heide et al., 2021; Zhu and Nowak, 2020), establishing a $\tilde{O}\Big(\frac{K-A}{\sqrt{KA}}\sqrt{T} \Big)$ minimax regret, where $T$ is the total number of interactions and $\tilde O(\cdot)$ drops all constant and logarithmic factors, improving the previous $\tilde{O}(\sqrt{KT/A})$ regret. We then provide a matching lower bound up to logarithmic factors, indicating that our established rate is nearly minimax-optimal. We further show that the knowledge of $A$ up to $\tilde{O}(1)$ factors is necessary to achieve near-optimal regret, as near-optimal
algorithms for one number of optimal arms must incur substantially larger regret than optimal regret for a smaller number. Overall, our results provide a comprehensive minimax characterization of $K$-armed bandits with $A$ over the entire range of $1 \leq A \leq K-1$.
Read the full abstract ▾
All authors Kaixuan Ji, Qiwei Di, Qingyue Zhao, Heyang Zhao, Quanquan Gu
Computation and Language · Cross-listed from eess.AS
Tacit-TTS: From Autoregressive Decoding to Masked Prediction for Efficient Transcript-Free Voice Cloning
Jian Chen, You Zhang and Mark Vinton
TTS systems with autoregressive semantic modeling have demonstrated strong zero-shot voice cloning performance and rich expressive variation, but their sequential decoding incurs substantial latency. Non-autoregressive alternatives offer much faster generation, yet often rely on more restrictive reference conditioning, such as requiring transcripts of the reference
speech during inference. We present Tacit-TTS, an efficient transcript-free zero-shot voice cloning system distilled from IndexTTS2. Our model replaces autoregressive text-to-semantic decoding with masked non-autoregressive generation, introduces training-free acoustic length estimation, and accelerates the flow-matching renderer through ReFlow
distillation. Across two English and two Mandarin
datasets, Tacit-TTS achieves competitive zero-shot quality while generating
speech over 10x faster than IndexTTS2 for utterances longer than 5 seconds. Its transcript-free conditioning further supports cross-lingual and non-lexical references. We validate this capability using references from eight other
languages, infant babble, and synthetic gibberish, where transcript-dependent systems often degrade or fail due to unreliable ASR transcripts.
Read the full abstract ▾
Systems and Control
Forward-Invariant Policy Classes for Safe Reinforcement Learning in Multicopter Control
Chieh Tsai, Muhammad Junayed Hasan Zahed, Jinzhi Shen et al. (+5 more)
This paper proposes a reinforcement learning (RL) framework for safe gain scheduling based on forwardinvariance-induced action-space design. Under stated nominalmodel and inversion-domain assumptions, rather than enforcing safety through runtime shielding or penalty-based constraints, safety is embedded directly into the policy class. Specifically, we construct a finite library of feedback controllers sharing a common Lyapunov certificate that establishes forward invariance of a prescribed admissible set under arbitrary switching. Consequently, any policy whose actions are restricted to this library, including policies encountered during RL exploration, inherits the same certificate. Policy optimization can therefore focus on closed-loop performance without runtime safety filtering or action projection. The framework is instantiated for quadcopter hover regulation, where a DQN schedules among certified feedback controllers. Nonlinear MuJoCo simulations demonstrate state-dependent gain scheduling and empirically evaluate robustness to wind, model mismatch, sensor noise, and sensing delay. The results illustrate how safety certification can be separated from policy optimization by learning over a forward-invariant policy class.
Read the full abstract ▾
All authors Chieh Tsai, Muhammad Junayed Hasan Zahed, Jinzhi Shen, Yi Xie, Ruoshan Lan, Majed Obaid, Salim Hariri, Hossein Rastgoftar
Robotics
TERRA: Terrain-Aware Reconstruction, Retargeting and Control for Musculoskeletal Locomotion
Merkourios Simos, Chengkun Li, Bianca Ziliotto et al. (+1 more)
Recent advances in musculoskeletal modeling and reinforcement learning have enabled muscle-actuated
agents to reproduce increasingly complex human motions. Yet these capabilities remain largely confined to flat ground, in part because motion
datasets rarely include aligned terrain geometry and because retargeting terrain interactions to complex musculoskeletal bodies is challenging. We present TERRA, an end-to-end pipeline for terrain-aware retargeting and control of musculoskeletal locomotion. From kinematic trajectories alone, TERRA combines terrain priors, estimated contacts, and negative free-space evidence to recover task-relevant support geometry. TERRA further considers anatomical, tendon-continuity, and contact constraints during retargeting. Using the resulting motion-terrain pairs from five
datasets, we successfully train a single muscle-actuated control policy on 9.4 hours of diverse locomotion. Across reconstruction, retargeting, and held-out tracking
benchmarks, TERRA improves terrain accuracy, sharply reduces anatomical and interaction violations, and achieves the highest observed completion rate over supported terrain families. Overall, TERRA provides a practical route from scene-less motion data to muscle-actuated locomotion over diverse non-flat terrain. Project website: https://cnai.epfl.ch/terra/
Read the full abstract ▾
All authors Merkourios Simos, Chengkun Li, Bianca Ziliotto, Alexander Mathis
Artificial Intelligence
AgBench: Agentic AI Benchmarks for Personal AI Devices
Yizhou Han, Di Wu, Dhananjay Saikumar et al. (+1 more)
Agentic AI systems increasingly rely on cloud-hosted large
language models for planning, tool use, and iterative execution, raising concerns about API cost and data exposure. Advances in personal AI devices enable
agents to execute locally, but limited resources on device may affect task success and performance. Existing
benchmarks are inadequate for systematically characterizing these trade-offs across devices, workloads, and deployment architectures. We present AgBench, a
benchmark suite and open artifacts for reproducible evaluation of agentic AI on personal devices. Using AgBench, we evaluate local, hybrid, and cloud execution across agentic workloads, examining task success, latency, cloud API cost, and data exposure. Our results, drawn from over 162.07 million data points, show that personal AI devices can complete many
agent tasks locally, but local-only execution generally has lower task success and longer completion times than cloud-only execution, especially as concurrency increases. Local-only execution eliminates cloud model API costs and sensitive-information exposure to cloud
agents. Hybrid execution can improve task success, but its cloud cost and data exposure depend on how
agents divide work and share information. No single architecture performs best across task success, goodput, cloud cost, and data exposure; deployment choices should reflect the intended workload and device capabilities. AgBench is available at https://anonymous.4open.science/r/AgBench-2777.
Read the full abstract ▾
All authors Yizhou Han, Di Wu, Dhananjay Saikumar, Blesson Varghese
Operating Systems
StateFork: Branchable Infrastructure for Agent Exploration
Jiakai Xu, Tianle Zhou, Georgios Liargkovas et al. (+5 more)
AI
agents improve task success by exploring multiple trajectories, but for computer-use
agents each trajectory modifies external environment state. Branching from an intermediate point is correct only when restoration is observation-equivalent - future actions produce the same observations - and practical only when creating, restoring, and discarding branch states is physically efficient. We study this problem for terminal-using
agents, where tasks modify files, shell context, running processes, and local services. We introduce StateFork, a logical control plane that separates exploration policies from physical state materialization, exposing sessions, commands, snapshots, restores, and cleanup over multiple execution substrates. We also build Waypoint, a checkpoint/restore substrate for terminal execution sessions that combines filesystem layering, process checkpointing, and a persistent terminal-compatible command session. Together, StateFork and Waypoint improve terminal-
agent exploration by combining sample-efficient search with efficient restoration of the right execution state. On Terminal-Bench, branch-based exploration through StateFork and Waypoint improves task completion over pass@20 at the same visited-node budget, and using Waypoint achieves 26% higher task accuracy than other execution substrates while completing exploration up to 70% faster. These results show that observation-equivalent, physically efficient execution sessions are a key systems abstraction for exploratory AI
agents.
Read the full abstract ▾
All authors Jiakai Xu, Tianle Zhou, Georgios Liargkovas, Danielle Gillai, Ruizhe Fu, Patrick Shen, Eugene Wu, Kostis Kaffes