Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (CS.CL) 2026-06-16

KVEraser: Learning to Steer KV Cache for Efficient Localized Context Erasing

Post-hoc context erasing over the KV cache is challenging because a local edit has a global consequence: once a span has been processed, its influence propagates into the cached states of all subsequent tokens. This issue arises naturally in long-context LLM applications, where stale retrieved facts, incorrect tool observations, retracted user preferences, or harmful prompt injections may be identified only after prefill. Exact erasing must then recompute all tokens after the deleted span, making its computational cost depend on suffix length rather than erased-span length. We introduce KVEraser, a learned KV-cache editing method for efficient localized context erasing. Given a processed context and a span to remove, KVEraser replaces only the KV states of the erased interval with learned steering states while reusing the remaining cache unchanged. To learn a transferable erasing mechanism, we build a two-stage training pipeline: generic span-neighbor pre-training teaches the eraser to suppress the influence of the erased span, while task-specific fine-tuning adapts this capability to downstream scenarios. Experiments show that KVEraser nearly matches full recomputation in post-erasure performance on in-domain tasks across 1K–32K context lengths, while its latency increases by only 24% compared with a 17.6x increase for full recomputation. KVEraser also generalizes to unseen long-document QA tasks with harmful factual distractors, achieving the best performance among approximate baselines with a 3–4x speedup over full recomputation.

02.
arXiv (CS.CL) 2026-06-19

Quality Over Clicks: Iterative Reinforcement Learning for Early-Stage E-Commerce Query Suggestion

Existing dialogue systems rely on query suggestion to enhance user engagement. Recent approaches mainly optimize generative models using click-through rate (CTR) models to align with user preferences. However, these methods are less effective in early-stage deployment scenarios, where click feedback is sparse and insufficient for training a reliable CTR model. To bridge this gap, we propose QualEQS, a quality-first iterative reinforcement learning framework for e-commerce query suggestion. We formalize actionable suggestion quality along three dimensions that directly affect downstream usability: answerability, factuality, and information gain. To continuously improve from online traffic without click supervision, we further propose group-level disagreement among candidate suggestions to identify ambiguous query contexts and mine hard training cases for iterative refinement. We also introduce EQS-Benchmark, a dataset of 16,949 real-world e-commerce queries for offline training and evaluation. Experiments show that our quality-based offline metrics correlate strongly with online performance, providing a practical evaluation recipe for sparse-feedback deployment. In both offline and online settings, QualEQS consistently outperforms strong baselines, yielding a 6.81% improvement in online ChatPV in a real-world enterprise-level conversational shopping assistant system.

03.
arXiv (quant-ph) 2026-06-16

Hyperinvariant Spin Network States – An AdS/CFT Model from First Principles

arXiv:2510.06602v2 Announce Type: replace Abstract: We study the existence and limitations of hyperinvariant tensor networks incorporating a local SU(2) symmetry. As discrete implementations of the anti de-Sitter/conformal field theory (AdS/CFT) correspondence, such networks have created bridges between the fields of quantum information theory and quantum gravity. Adding SU(2) symmetry to the tensor network allows a direct connection to spin network states, a basis of the kinematic Hilbert space of loop quantum gravity (LQG). We consider a particular situation where the states can be interpreted as kinematic quantum states for three-dimensional quantum gravity. We show that important aspects of the AdS/CFT correspondence are realized in certain quantum states of the gravitational field in LQG, thus justifying, from first principles, a class of models introduced by [F. Pastawski et al., JHEP 06, 149 (2015)]. We provide examples of hyperinvariant tensor networks, but also prove constraints on their existence in the form of no-go theorems that exclude absolutely maximally entangled states as well as general holographic codes from local SU(2)-invariance. We calculate surface areas as expectation values of the LQG area operator and discuss further possible constraints as a consequence of a decay of correlations on the boundary.

04.
arXiv (CS.AI) 2026-06-11

Are LLMs Bad at Moral Reasoning?

arXiv:2606.11635v1 Announce Type: cross Abstract: For highly capable AI systems to operate safely in dynamic, open-ended environments, they must be able to identify, understand, and respond to moral reasons for action, and constrain their behaviour accordingly. A growing body of research aims to evaluate this capacity – moral competence – in today's most capable AI systems, recently reaching broadly pessimistic conclusions. One of the most ambitious such papers collects gold-standard human-authored rubrics for evaluating moral reasoning in 1,000 cases, and benchmarks frontier AI models against those rubrics, with underwhelming results. In this paper, we argue that the MoReBench dataset can be redeployed to give a much more optimistic picture of LLMs' moral reasoning (an essential part of moral competence). We show that if, instead of scoring LLMs' responses to these cases against these rubrics, we instead give the LLMs the same task given to humans – to generate scoring rubrics for the moral analysis of particular cases – the rubrics they generate are both better calibrated to the human rubrics than their open-ended responses, and, where they differ, plausibly reflect nothing more than the vast dimensionality of most moral problems, as well as highlighting some human departures from the "rubric for creating rubrics". Taking these points into consideration, the MoReBench dataset suggests that LLMs are significantly more capable at moral reasoning than was previously believed.

05.
arXiv (CS.CV) 2026-06-19

SA-VIS: Sparse frame Annotations for training Video Instance Segmentation

Recent online video instance segmentation (VIS) methods have achieved impressive results, thus becoming the preferred approach to segment instances in videos. Despite the resurgence of impressive single image models, the online (or semi-online) VIS approaches outperform single-image models (e.g., based on SAM) by using long sequences of densely annotated frames during training. However,such a training setup of VIS is expensive in the sense of compute as well as dense annotations required. In order to solve these major flaws, we argue that the effective modeling of the instances and their evolution in videos do not require densely annotated frames. To that end, we propose a simple and effective module, called Past-frames Feature Propagation (PFP) which aggregates low-dimensional features from the image encoder of multiple frames. This simple low-compute module provides tremendous learning capability in using sparse video frame labels for end-to-end training. Combined with a light-weight frame-specific Instance Queries, our Sparse frame Annotation VIS (SA-VIS) significantly improves performance over its baseline. Most interestingly, our simple design that avoids complexities effectively bridges the gap in accuracy between training on sparsely and densely annotated video sequences. This translates to a mere 0.4% drop in performance of SA-VIS when using annotations for only 1/5 of the images in the dataset. Empirically, SA-VIS shows strong improvements over the baseline on YouTube-VIS 2019/2021/2022 and Occluded VIS (OVIS) and an over 1% improvement in AP on the state-of-the-art in a limited annotations scenario.

06.
arXiv (CS.CL) 2026-06-15

Non-Parametric Machine Text Detection via Multi-View Gaussian Processes

Adversarial conditions such as paraphrasing and targeted style transfer sharply degrade the accuracy of machine text detectors. A document, however, carries multiple complementary signals (e.g., stylistic features, likelihood and rank-order features, and structural features), and an attack that suppresses one may leave others intact. While a parametric classifier can learn to combine these features given sufficient supervision, classifiers are prone to making confidently incorrect predictions when the distribution shifts (e.g., novel attacks or unseen language models). To address this, we propose a multi-view, non-parametric detection framework that extracts complementary feature views from the same document and aggregates per-view evidence through a Gaussian process ensemble. By aggregating evidence across views, an adversary must simultaneously defeat multiple independent axes of detection, substantially raising the cost of evasion. The Gaussian process formulation additionally provides calibrated probabilities and principled abstention on out-of-distribution inputs, supporting reliable deployment in high-stakes settings. We evaluate on three benchmarks spanning diverse generators and attacks: the DetectRL and RAID benchmarks, and the PAN2025 shared task and demonstrate that our multi-view detector maintains strong performance under the considered attacks, outperforming existing approaches against held out attacks.

07.
arXiv (CS.CL) 2026-06-11

Measuring language complexity from hierarchical reuse of recurring patterns

We introduce the ladderpath index as a measure of language complexity grounded in algorithmic information theory. It counts the minimum steps needed to reconstruct a sequence through hierarchical reuse of repeated substructures, capturing an exactly computable but constrained form of algorithmic compressibility related to, but distinct from, Kolmogorov complexity. We apply the ladderpath approach to 21 parallel corpora from the Parallel Universal Dependencies dataset. The ladderpath index is approximately invariant across the languages, and varies much less than the corpus length. This is more pronounced when all corpora are mapped to a unified binary representation, providing evidence for the equi-complexity hypothesis from a representation-independent perspective. We also observe trade-offs between character inventory size and corpus length, and between vocabulary-level and corpus-level reconstruction complexity, supporting the trade-off hypothesis that total complexity is conserved and redistributed across linguistic levels. The reusable substructures identified by the ladderpath approach, without any linguistic input, overlap with words and morphological components attested in the natural vocabulary. The hierarchical reuse captured by the ladderpath approach parallels the chunking mechanisms proposed in cognitive science, where the human cognitive system compresses linguistic input into nested, reusable units under shared memory and processing constraints. This connection between cognitive chunking and the ladderpath approach provides a new interpretation for the equi-complexity and trade-off hypotheses, grounding both in the shared cognitive architecture that underlies language processing across human languages.

08.
arXiv (CS.CL) 2026-06-19

Think Again or Think Longer? Selective Verification for Budget-Aware Reasoning

Test-time reasoning is increasingly used as a serving-time control knob, but extra reasoning is not uniformly valuable: it can repair failed attempts, waste compute on already-correct answers, or introduce harmful answer changes. We study this as a deployment allocation problem rather than a new-verifier problem. We introduce \sevra, Selective Verification for Reasoning Allocation, a serving-layer controller that decides whether to preserve a frozen solver's initial answer or invoke active verification. Using a frozen Qwen3-4B solver, we log intervention outcomes and train recoverability-aware gates from serving-visible attempt state. On \mathfive, selective verification reaches 76.3\% accuracy, compared with 75.5\% for always verifying, while reducing post-generation tokens by 26.8\% and harmful flips from 2.2\% to 1.0\%. However, an 8,192-token initial solve reaches 76.0\% accuracy with 28\% fewer total model tokens, showing that selective recovery is useful but not the best tested cost frontier. In frozen transfer to \gsm, the selective policy verifies only 3.0\% of examples, improves accuracy from 93.4\% to 94.5\%, and reduces verification tokens by 91.2\% relative to always verifying; again, a longer initial solve matches its accuracy with fewer realized tokens. On CommonsenseQA, always-on verification hurts, while Self-Consistency@5 improves accuracy at about five times the realized token cost. The resulting deployment rule is: tune the initial budget first, then use selective recovery when explicit checks, bounded retries, auditability, or regression-risk control matter.

09.
arXiv (CS.CV) 2026-06-11

Auditing Demographic Bias in Facial Landmark Detection for Fair Human-Robot Interaction

Fairness in human-robot interaction critically depends on the reliability of the perceptual models that enable robots to interpret human behavior. While demographic biases have been widely studied in high-level facial analysis tasks, their presence in facial landmark detection remains unexplored. In this paper, we conduct a systematic audit of demographic bias in this task, analyzing the age, gender, and race biases. To this end, we introduce a controlled statistical methodology to disentangle demographic effects from confounding visual factors. Our analysis demonstrates that visual confounders, particularly head pose and face resolution, heavily outweigh the impact of demographic attributes. Notably, after accounting for these confounders, performance disparities across gender and race vanish. However, we identify a statistically significant age-related bias, with higher localization errors for older individuals. This shows that fairness issues can emerge even in low-level vision components and can propagate through the HRI pipeline. We argue that auditing and correcting such biases is a necessary step toward trustworthy and equitable robot perception systems.

10.
arXiv (CS.LG) 2026-06-16

Best Arm Identification with Minimal Regret

arXiv:2409.18909v2 Announce Type: replace Abstract: Motivated by real-world applications that necessitate responsible experimentation, we introduce the problem of best arm identification (BAI) with minimal regret. This variant of the multi-armed bandit problem elegantly amalgamates two of its most ubiquitous objectives: regret minimization and BAI. More precisely, the agent's goal is to identify the best arm with a prescribed confidence level $\delta$, while minimizing the cumulative regret up to the stopping time. Focusing on single-parameter exponential families of distributions, we leverage information-theoretic techniques to establish an instance-dependent lower bound on the expected cumulative regret. Moreover, we present an impossibility result that underscores the tension between cumulative regret and sample complexity in fixed-confidence BAI. Complementarily, we design and analyze the Double KL-UCB algorithm, which achieves asymptotic optimality as the confidence level tends to zero. Notably, this algorithm employs two distinct confidence bounds to guide arm selection in a randomized manner. Our findings elucidate a fresh perspective on the inherent connections between regret minimization and BAI.

11.
arXiv (CS.AI) 2026-06-18

AI Sovereignty as National Learning Capacity: A Human-Centered Learning Mechanics Viewpoint on France, the United States, and China

arXiv:2606.00729v2 Announce Type: replace Abstract: Artificial intelligence in France is often discussed through separate dimensions such as investment, compute, regulation, employment, sovereignty, and education. This viewpoint paper proposes a unified interpretation: France can be analyzed as a national AI learning system. Building on Human-Centered Learning Mechanics (HCLM), we use HCLM not as a validated econometric model, but as a conceptual and diagnostic lens for interpreting national AI development as a balance between information injection, absorptive capacity, and institutional dissipation. Information injection includes compute, data, talent, research, capital, industrial deployment, and policy experimentation. Institutional dissipation refers to avoidable frictions such as administrative overload, coordination failures, energy constraints, regulatory uncertainty, talent mobility pressures, and weak industrial absorption. Regulation is not treated as mere friction: adaptive governance, trusted data spaces, and safety-oriented standards may increase long-term learning capacity by improving legitimacy, interoperability, and social trust. The central claim is not that a country follows neural-network equations, but that AI sovereignty depends on how effectively it converts distributed information into absorbed, coordinated, and socially legitimate capability. The paper connects HCLM with neural scaling laws, endogenous growth theory, creative destruction, absorptive capacity, and coordination mechanisms. It offers a formal heuristic, policy indicators, illustrative scenarios, and implications for France. The numerical results are diagnostic scenarios, not econometric estimates or official rankings. The proposed viewpoint reframes AI policy as the governance of an open, strategic, non-equilibrium learning system that should be tested with historical and cross-country data.

12.
arXiv (CS.LG) 2026-06-16

Continual Backdoor Training in IoT/CPS

arXiv:2606.14987v1 Announce Type: cross Abstract: Internet of Things (IoT) and Cyber-physical systems (CPS) increasingly rely on continual learning (CL) to adapt to evolving environments, device heterogeneity, and concept drift, thereby improving overall utility. While continual adaptation is essential for long-lived IoT deployments where data patterns evolve, it also introduces new security vulnerabilities. In particular, backdoor attacks can exploit incremental updates, replay buffers, and representation reuse to implant persistent malicious behaviors that remain dormant during normal operation but activate upon specific triggers. In this paper, we present a backdoor attack in continual learning used in IoT/CPS systems. To this end, we formalize an IoT/CPS-specific threat model, analyze why continual learning amplifies backdoor persistence in IoT pipelines, and evaluate our technique under varying conditions. Our analysis highlights critical open challenges in securing lifelong learning in IoT/CPS and industrial IoT (IIoT) environments, as well as the need for heightened security controls.

13.
arXiv (math.PR) 2026-06-16

Pricing Excess-of-Loss Reinsurance and CAT Bonds under Climate Uncertainty: A Cox Process Framework with Temperature-Dependent Stochastic Intensity

arXiv:2606.14830v1 Announce Type: cross Abstract: This paper develops a climate-aware pricing framework for excess-of-loss (XL) reinsurance contracts and catastrophe (CAT) bonds under non-stationary catastrophe risk. Catastrophe arrivals are modeled as a Cox process whose stochastic intensity depends exponentially on a temperature-related climate index. To represent climate dynamics, the index is modeled as a mean-reverting Ornstein–Uhlenbeck process around a time-dependent warming trend. Within this setting, aggregate losses follow a compound Cox structure with lognormal severities. Pricing is performed under a reduced-form risk-adjusted measure, which provides a tractable valuation approach for XL reinsurance layers and binary zero-coupon CAT bond payoffs in an incomplete market setting. Because catastrophe losses are not dynamically replicable, the framework emphasizes scenario-based valuation rather than model-independent no-arbitrage bounds. A Monte Carlo valuation scheme is implemented to quantify the economic implications of climate-dependent catastrophe intensity. The numerical results show that climate dependence materially changes the loss-generation mechanism and affects the valuation of catastrophe-linked contracts. In the baseline calibration, the climate-aware model increases the excess-of-loss reinsurance premium and lowers the CAT bond price relative to the stationary benchmark. Furthermore, our analysis of the 99.5\% Tail Value-at-Risk (TVaR) indicates that stationary benchmarks may underestimate economic capital requirements by approximately 13.7\% compared to the climate-aware framework, highlighting the potential regulatory relevance of the proposed model. This finding highlights that benchmark design is critical for interpreting climate-pricing effects.

14.
bioRxiv (Bioinfo) 2026-06-11

A quantitative coordinate system for developmental dynamics

Quantitative comparison of morphogenesis across individuals remains a fundamental challenge, as developing embryos vary in shape, orientation and developmental tempo. Moreover, real-time three-dimensional imaging generates large, heterogeneous four-dimensional datasets that are difficult to directly align. As a result, developmental variability is typically described qualitatively rather than measured. Here we introduce STERN, a quantitative framework that learns continuous spatiotemporal representations of morphogenesis directly from in vivo 4D imaging data. By embedding embryos into a shared spatiotemporal space, STERN defines a quantitative developmental coordinate system that enables direct comparison of developmental trajectories across individuals without requiring explicit registration or staging. Applied to mouse embryogenesis, STERN reveals that embryos follow conserved developmental trajectories while progressing at distinct temporal rates, providing a quantitative measure of developmental heterochrony. Extending this framework to zebrafish neural crest light-sheet timelapse imaging, we further show that developmental order is preserved across distinct imaging views even with altered anatomical coverage, supporting the generality of the learned representation across vertebrate imaging contexts. Finally, in developing mouse hearts, where morphogenesis proceeds through subtle and continuously evolving structural changes, STERN resolves fine-scale developmental dynamics at minute-scale temporal resolution that are difficult to localize reproducibly using human experts or general-purpose multimodal AI. Together, these results establish a shared quantitative coordinate system for morphogenesis, in which developmental trajectories become directly comparable across individuals and developmental variability becomes a measurable property.

15.
medRxiv (Medicine) 2026-06-17

Macrophage-targeted glucocorticoid prodrug resolves acute inflammation while preserving HPA axis function: mechanistic, preclinical, and Phase II/III clinical evidence

Glucocorticoids (GCs) remain the fastest-acting anti-inflammatory agents but are constrained by systemic exposure that suppresses the hypothalamic pituitary adrenal (HPA) axis, silences adaptive immunity, and drives chronic toxicities. Chronic inflammatory diseases are sustained by long-lived CD206+ macrophages containing immune-resistant pathogenic material not cleared physiologically. We developed 101-PGC-005 ('005), a macrophage-targeted type 1a dexamethasone prodrug engineered for low-affinity, recycling-compatible uptake via CD206, with intracellular release triggered by acidic endosomes. We evaluated '005 in mechanistic assays, pathogen-diverse preclinical models, three human pharmacokinetic (PK) studies, and an adaptive-design randomized Phase II/III trial in 309 hospitalized patients with moderate COVID-19. In two completed Phase I human studies, a first-in-human dose-escalation and repeated-dose study and a dedicated single/multiple-dose PK and safety study; '005 circulated as intact prodrug with rapid systemic clearance (Tmax ~0.5 h; terminal half-life ~1.9 h), with no measurable free dexamethasone after single dosing and only low, clinically non-significant free dexamethasone after repeated dosing, and intact prodrug recovered unchanged in urine. Morning cortisol and ACTH were preserved after 30 mg once daily for three consecutive days (1.5 times the intended therapeutic dose). A cerebrospinal fluid PK study is evaluating central-compartment penetration. In the Phase II/III trial, powered for non-inferiority, conducted across six sites in India under GCP with Ministry of Health approval and independent DSMB oversight; '005 (20 mg IV daily for 3 days) was superior to dexamethasone (6 mg IV daily for 3 -10 days) on the primary endpoint of time to > a 2-point improvement on the WHO ordinal scale (HR 2.31; 95% CI 1.83-2.93; p < 0.0001; median 3 vs. 4 days). '005 was also superior on viral clearance (HR 1.47; 95% CI 1.17-1.84; p = 0.0001), hospital discharge rate, SpO2; recovery, and fever resolution. Zero patients in the '005 arm received investigator-initiated corticosteroid supplementation despite protocol allowance. All 309 randomized patients completed the study (ITT = per-protocol). Safety profiles were equivalent (TEAEs 54.8% vs 54.5%; p = 0.958), with no Grade 3+ events, SAEs, deaths, or discontinuations in either arm. Mechanistically, '005 delivered dual benefit: acute debulking of inflammatory macrophages and selective depletion of chronically activated pathology-sustaining macrophages, while preserving CXCL10 antiviral signaling and physiologic HPA control. Critically, HPA preservation is not merely a safety feature, it is a core efficacy mechanism: by clearing the pathogenic macrophage burden that was overriding HPA regulation, '005 restores the conditions for endogenous cortisol to resume its pulsatile, demand-responsive anti-inflammatory role across all GR-expressing cells, lymphocytes, endothelial cells, neurons, and newly differentiated macrophages, that '005 itself cannot reach. These findings support regulatory-grade evidence for macrophage-targeted corticosteroid therapy and provide the foundation for further development across acute inflammatory indications (sepsis, viral pneumonia, cytokine-release syndromes) and chronic macrophage-driven diseases (atherosclerosis, metabolic steatohepatitis, neurodegeneration, tumor-associated macrophages).

16.
arXiv (CS.AI) 2026-06-16

Intelligence Is Not the Bottleneck: Validating an LLM First-Pass Manuscript Score Against Peer-Review Outcomes

arXiv:2606.15887v1 Announce Type: cross Abstract: Large language model (LLM) systems are increasingly proposed to assist peer review, yet most evaluations judge the prose of machine-generated review text, not the validity of the numeric score a system assigns. We validate AIPR, which reads a submitted manuscript and emits five 0-100 quality dimensions and a weighted overall score, against the public decision outcomes of a major machine learning venue. AIPR grades by prompting alone, with no fine-tuning on reviews or decisions. Across 300 ICLR submissions with public decision tiers and reviewer ratings, graded under a frozen pipeline with hypotheses pre-registered before any score met any outcome, the overall score separates rejected from accepted submissions (AUROC 0.82, 95% CI 0.78-0.87), rises monotonically across tiers, and tracks the mean reviewer rating. The signal is strongest where we claim it: the lowest-scoring fifth is rejected far above the base rate, with oral papers absent. The validity comes mostly from the model: a one-paragraph prompt on the same model discriminates almost as well as the full pipeline (the small gap favours the pipeline but does not meet the pre-declared criterion, p = 0.09). What the engineering adds is reliability and a grounded review: AIPR's score barely moves across repeated runs (0.7 vs. 2.8 points within-paper SD) where the bare prompt swings, and the same pass returns a rubric-structured, evidence-grounded review rather than a bare number, with the human keeping the decision.

17.
Nature (Science) 2026-06-17

Rock weathering can counteract river CO<sub>2</sub> emissions induced by permafrost thaw

作者:

Climate-induced permafrost thaw unlocks large stores of organic carbon that are mineralized and emitted as carbon dioxide (CO2) from rivers to the atmosphere1. Concurrently, warming and permafrost thaw can increase mineral weathering rates, thus affecting the release and sequestration of inorganic carbon2–4. Yet how these biological and geological carbon cycles interact and jointly affect CO2 dynamics (emission compared with drawdown) in permafrost rivers remains unknown5. Here we combine CO2 emissions, organic and inorganic solute concentrations, dual carbon isotopes (δ13C–Δ14C) and geochemical modelling to infer how permafrost thaw may affect river biogeochemistry over decades to centuries across the Qinghai–Tibet Plateau. Leveraging a gradient of thermal permafrost degradation, we find that river CO2 emissions decline, whereas solute fluxes from rock weathering increase with decreasing permafrost cover. Across this region, net CO2 drawdown fluxes from rock weathering are about 35% of river CO2 emissions, varying from around 15% in catchments with continuous permafrost to more than 100% in catchments with discontinuous or isolated permafrost. Thus, carbon fluxes from chemical weathering may become increasingly important with ongoing permafrost thaw, potentially even outpacing river CO2 emissions. Our findings disentangle the interplay between biological and geological carbon fluxes that are important for the cryosphere and the global carbon cycle. Permafrost thaw on the Qinghai–Tibet Plateau increases rock-weathering rates while reducing river CO2 emissions, suggesting geological carbon fluxes may eventually outpace thaw-driven emissions.

18.
arXiv (CS.CL) 2026-06-12

ProPlay: Procedural World Models for Self-Evolving LLM Agents

Self-evolving agents are expected to improve through interaction without external supervision, but this remains difficult in partially observable environments where agents must explore actively, learn from limited feedback, and decide when to trust prior experience. Existing LLM-agent methods often rely on memory or planning modules, yet they rarely close the loop between them to continually refine an internal understanding of environment dynamics. We introduce ProPlay, a procedural world model that supports procedure-level preplay, where agents can rehearse future procedural paths using the learned world knowledge. Rather than representing experience as isolated rules or low-level action constraints, ProPlay abstracts successful trajectories into procedures and organizes them in a procedure graph that captures causal transitions among task stages. Each transition is associated with a reliability record embedding to estimate its task-specific contribution from past outcomes. Before each episode, ProPlay simulates future procedural trajectories over known graph structures as structured soft guidance; after execution, it refines the graph using environment feedback. Experiments on public benchmarks show that ProPlay consistently improves environment understanding and self-evolution capability over strong baselines. Our code has been released in https://github.com/antman9914/proplay.

20.
arXiv (CS.CL) 2026-06-15

Did You Forget What I Asked? Prospective Memory Failures in Large Language Models

作者:

Large language models often fail to satisfy formatting instructions when they must simultaneously perform demanding tasks. We study this behaviour through a prospective memory inspired lens from cognitive psychology, using a controlled paradigm that combines verifiable formatting constraints with benchmark tasks of increasing complexity. Across three model families and over 8,000 prompts, compliance drops by 2-21% under concurrent task load. Vulnerability is highly type-dependent: terminal constraints (requiring action at the response boundary) degrade most, with drops up to 50%, while avoidance constraints remain comparatively robust. A salience-enhanced format (explicit instruction framing plus a trailing reminder) recovers much of the lost compliance, restoring performance to 90-100% in many settings. Interference is bidirectional: formatting constraints can also reduce task accuracy, with one model's GSM8K accuracy dropping from 93% to 27%. In additional stacking experiments, joint compliance declines sharply as constraints accumulate. All results use deterministic programmatic checkers without an LLM-as-judge component on publicly available datasets.

21.
arXiv (CS.AI) 2026-06-16

APEX: Adaptive Principle EXtraction A Three-Layer Self-Evolution Framework for Production AI Agents

arXiv:2606.15363v1 Announce Type: new Abstract: Self-improvement in AI agents has emerged as a key research frontier: systems that modify their own prompts, workflows, and decision rules based on accumulated operational experience. The state-of-the-art Self-Harness framework [1] achieves 14–21% improvement on Terminal-Bench-2.0 by mining failure clusters and patching the agent harness. However, Self-Harness optimises only one dimension – the prompt harness – leaving behavioural principles and workflow topology unchanged. We propose APEX (Adaptive Principle EXtraction), a three-layer co-evolution framework that simultaneously evolves: (L1) the harness via failure-mode patching, (L2) behavioural principles via success-trace distillation [2], and (L3) the agent workflow topology via structural fitness-based selection [6]. We implement APEX on Joe [13], a production-grade super AI Agent built on NVIDIA Nemotron and designed as an Edge AI Agent Factory for the NVIDIA Agent Challenge 2026, managing a 15-node compute fleet using 114 real task traces collected over 18 days. APEX achieves an APEX Health Score of 0.570 (+90% vs. baseline 0.300) in a single evolutionary run, distilling 6 novel reusable principles and selecting a research-first workflow topology scoring 0.900 (+20%). Our results demonstrate that multi-dimensional co-evolution substantially outperforms single-axis harness optimisation, at a cost of only 4 LLM calls (~270 s) on a local qwen2.5-coder:32b instance.

22.
arXiv (CS.CV) 2026-06-11

Ouroboros-Spatial: Closing the Data-Model Loop for Spatial Reasoning

Spatial reasoning remains a persistent challenge for multimodal large language models (MLLMs). Existing approaches largely rely on large-scale, statically curated datasets, where all training samples are treated uniformly regardless of the model's evolving capabilities. This static paradigm is inherently data-inefficient: training capacity is often spent on samples that are either trivial or overly difficult for the model at its current stage. To address this limitation, we propose Ouroboros-Spatial, a self-evolving training framework in which the model plays dual roles as a proposer and a solver. In each iteration, a frozen proposer generates spatial question-answer (QA) pairs from 3D scene metadata and raw video frames, together with executable code for deriving reliable ground truth. A learnable solver is then fine-tuned on the accepted samples, and its per-sample prediction confidence is used as a difficulty signal. This signal is fed back to the proposer in the next iteration, guiding it to generate questions better matched to the solver's current capabilities. Through this closed-loop design, the training distribution co-evolves with model ability, reducing redundant trivial examples while filtering out ambiguous or uninformative samples with limited learning value. Across six spatial reasoning benchmarks, Ouroboros-Spatial substantially improves Qwen3-VL-4B and Qwen3-VL-8B while using an order of magnitude fewer training examples than recent large-scale curated datasets. On VSI-Bench, it yields absolute gains of 9.9 and 6.8 points for the 4B and 8B models, respectively, enabling both to outperform a wide range of strong open-source and proprietary baselines.

23.
arXiv (CS.AI) 2026-06-12

SMSR: Certified Defence Against Runtime Memory Poisoning in Persistent LLM Agent Systems

作者:

arXiv:2606.12703v1 Announce Type: cross Abstract: Retrieval-augmented generation (RAG) agents increasingly run with persistent memory that accumulates across user sessions. This creates a new attack surface: an adversary interacting only through normal channels can inject crafted memories that, once retrieved, steer the agent's responses for future users, without touching model weights or code. We call this Multi-Session Memory Poisoning (MSMP) and show that no existing defence certifies against it; static-corpus defences (RobustRAG, ReliabilityRAG) assume a fixed knowledge base, and heuristic filters are bypassed by fluent enterprise-style text. We present Signed Memory with Smoothed Retrieval (SMSR), the first defence with a certified robustness bound for this setting. Component 1 adds HMAC-SHA256 provenance at write time, blocking unsigned injection. Component 2 applies randomised memory ablation with verdict-based majority voting at query time, bounding the influence of authenticated adversaries. We prove that no provenance-free retrieval-time filter can certify against adaptive injection, derive a hypergeometric certificate for Component 2, and formalise the Consistent Minority Effect, whereby a consistent adversarial answer wins string-based voting as a numerical minority while verdict-based voting removes it. Across 15 enterprise scenarios (3,150 repeated trials), Component 1 cuts attack success from 93-100% to 0% for all unsigned variants. For an authenticated adversary with a single injection, Component 2 holds success to 8.0% (95% CI [5.8, 10.9], n=450), below the certified worst case. In an end-to-end query-only attack where the agent itself writes the poison rather than it being pre-seeded, SMSR reduces success from 65.3% to 5.3% (n=150, non-overlapping CIs) on a live agent stack. Clean-query utility is 90% (Component 1) and 85% (combined).

24.
arXiv (CS.AI) 2026-06-16

Demystifying Variance in Circuit Discovery of LLMs

arXiv:2606.16920v1 Announce Type: cross Abstract: Circuit discovery is a key technique in mechanistic interpretability to pinpoint the model components that are crucial for performing a given task. Although the current state-of-the-art method (EAP-IG) performs well on the metric of (un)faithfulness, it suffers from substantial variability. This includes resampling variance, where the circuit changes when we probe with a new batch of data from the same distribution; rephrasing variance, where the discovered circuit shifts when the prompts are rephrased; and sample-wise variance, where a circuit with low population unfaithfulness exhibits large fluctuations in unfaithfulness across individual samples. This paper studies the roots of these variances. We demonstrate that CEAP, our new circuit discovery method that improves upon EAP-IG with a theoretical guarantee, can substantially lessen resampling variance. We further show that rephrasing variance arises because prompts with different templates tend to activate different circuits in the model. This leads us to argue that it may be challenging to find a comprehensive circuit that explains and controls the model's behavior on a task, which can be expressed in countless templates, suggesting that LLMs may be inherently hard to steer. We show that sparsity, which has been claimed to form more compact and interpretable task circuits, fails to solve this problem. Regarding sample-wise variance, we argue that it is largely benign: extremely poor unfaithfulness scores often stem from how unfaithfulness is defined, rather than from defects in the measured circuits. We show that the magnitude of unfaithfulness is affected by selective contribution scaling, a neural mechanism that accounts for the extremely poor scores sometimes observed.

25.
arXiv (CS.AI) 2026-06-12

Creating and Evaluating K-12 GenAI Assessment Graders Through Context Engineering

arXiv:2606.12422v1 Announce Type: cross Abstract: The integration of large language models (LLMs) into educational assessment represents a transformative shift in classroom grading practices. While automated scoring systems and machine learning techniques have existed for decades, generative AI (GenAI) now enables educators to implement standards-based grading (SBG) with unprecedented efficiency and scale. This paper examines the theoretical foundations and evaluates an LLM grader that uses commercially available foundation models with context and prompt engineering to score student work against a rubric. Drawing on an empirical interrater agreement study using Massachusetts Comprehensive Assessment System (MCAS) data, we observed the Quadratic Weighted Kappa (QWK) and Proportional Reduction in Mean-Squared Error (PRMSE) across mathematics, science, and ELA, using Claude Sonnet 4, Haiku 4.5, GPT-5, and GPT-5 Mini. The results demonstrate that LLM graders, especially when based on foundational models with more parameters, achieve substantial agreement with human raters in mathematics and science assessments, while the performances vary in ELA, suggesting generic foundation models can be effective at scoring in given contexts. Additional analysis of teacher and student feedback reveals strong acceptance of AI-generated narrative feedback but skepticism toward numerical scores, suggesting that LLMs function most effectively as formative tools rather than summative evaluators. Our findings indicate that thoughtfully designed hybrid models that combine AI efficiency with teacher judgment can reduce workload, enhance feedback quality, and support equitable assessment practices without displacing professional expertise.