Academic Intelligence · Curated Daily

Explore the Frontier of Global Academia

AcademicHub aggregates real-time literature from top journals and preprint platforms. Build your personal research radar and let large language models compile cross-disciplinary analysis briefings automatically.

01.
arXiv (quant-ph) 2026-06-19

Many-Body Protection of Topological Edge Memory in Strong Interacting Quenches

arXiv:2606.19437v1 Announce Type: cross Abstract: Quantum quenches drive edge states far from equilibrium, yet whether the memory of a topological initial state survives in a non-integrable, interacting system has remained largely unexplored. We study this question in the bond-alternating XXZ chain – an interacting Su–Schrieffer–Heeger model hosting symmetry-protected topological edge modes with markedly enhanced boundary magnetization – and analyze quenches across all combinations of single-particle and many-body initial and final Hamiltonians. The results organize by a single distinction as we rigorously establish in this work: whether the post-quench Hamiltonian is free or genuinely interacting. For a free post-quench Hamiltonian, the dynamics is solved exactly by a correlation-matrix approach; the boundary-mode return amplitude decays as $t^{-3/2}$, and initial interactions enter only through a dressed one-body density matrix. For a genuinely interacting post-quench Hamiltonian, finite-time stability bounds prove that away from local resonances the first-dimer magnetization remains stable on time windows growing as arbitrarily large powers of the inverse inter-dimer coupling. Matrix product state simulations across all four protocols show that interactions in the final Hamiltonian markedly extend finite-time boundary memory – with local suppression near the isotropic $SU(2)$ point – revealing a many-body protection mechanism in a non-integrable system where scrambling would otherwise wash out initial-state memory fast.

02.
arXiv (CS.LG) 2026-06-18

Toward Simultaneously Optimal Regret in U-Calibration

arXiv:2606.18527v1 Announce Type: cross Abstract: U-calibration studies online forecasting algorithms whose predictions can be consumed by any unknown downstream agent, guaranteeing sublinear regret simultaneously for all proper loss functions. Existing U-calibration algorithms achieve worst-case optimal $O(\sqrt{T})$ regret for every bounded proper loss, but they fail to adapt to easier losses: as we show, even for smooth losses such as squared loss, they incur $\Omega(\sqrt{T})$ regret instead of the optimal $O(\log T)$ regret. In this work, we show that this limitation is not inherent. Specifically, we design a single forecast algorithm that simultaneously achieves $\tilde O(\sqrt{T})$ regret for every bounded proper loss and $O(\log T)$ regret for every bounded smooth proper loss. More generally, our algorithm also attains logarithmic regret for losses that are smooth relative to the log-barrier, which include several non-Lipschitz examples. Our approach is based on a novel variant of Follow-the-Perturbed-Leader (FTPL) in which perturbations are applied directly in the prediction space using self-concordant noise. The resulting analysis also departs substantially from prior FTPL analyses due to the complex nature of this noise and may be of independent interest.

03.
arXiv (CS.CV) 2026-06-15

HPSv3++: Scaling Reward Models Across the Full Spectrum of Diffusion Model Capabilities

Reward models guide text-to-image (T2I) systems toward outputs aligned with human preferences. However, typical reward models such as HPSv3 are trained on pre-annotated data from earlier T2I models, without accounting for quality discriminative shifts arising from evolving model capabilities and reinforcement learning (RL) iterations, limiting their broader applicability. In this work, we propose HPSv3++, a reward model framework that elevates the HPSv3 model for varying T2I model capabilities and their RL iteration changes across the full capability-iteration spectrum. Specifically, we first introduce HPDv3++, a 212K dual-dimension preference dataset annotated for text fidelity and aesthetic quality using a recent high-capability (Qwen-Image) model with human supervision. We then propose a two-stage training framework. Stage 1 employs data-aware orthogonal gradient projection to incorporate diverse aesthetic perception from HPDv3++ while preserving the original effective human preference knowledge in HPSv3. Stage 2 further leverages unlabeled data from T2I models spanning different capability levels and RL iterations, and introduces a joint capability-iterations conditioned signal for the reward model together with a standard deviation-driven unsupervised guidance mechanism, strengthening reward model across the capability-iteration spectrum. HPSv3++ achieves state-of-the-art preference prediction, outperforming HPSv3 9.8% on HPDv3, 5.5% on GenAI-Bench, while achieving 79.1%/88.1% on our proposed HPDv3++. When used for T2I RL training, it consistently improves GenEval scores across diverse T2I models, demonstrating its wide-range capabilities. The code is available at https://github.com/PlantPotatoOnMoon/HPSv3-PlusPlus.

04.
arXiv (CS.CL) 2026-06-12

Marginal Alignment Does Not Guarantee Joint-Distribution Fidelity: An Official-Reference Audit of Nemotron-Personas-Korea with Cross-Locale Replication

Authors:

Synthetic persona datasets cite alignment with official demographics as a basis for trust, yet downstream users consume them as joint structures across age, sex, region, occupation, education, name, and institutional status. Marginal alignment does not imply that these joints are preserved. We propose the Independence-Assumption Footprint (IAF), an audit primitive that operates on the attribute combinations a dataset card itself documents as treated independently. For each such combination, IAF compares the synthetic joint against an external official or institutional reference, using direct joint tables where available and rule-implied checks otherwise. Applied to NVIDIA Nemotron-Personas-Korea (one million Korean synthetic personas), IAF finds that NPK aligns with KOSIS marginals while three joints fail. The major-by-occupation distribution against the KEIS graduate universe carries a large conditional mismatch. The age profile of military service is institutionally inconsistent. Female representation in male-dominated occupations is substantially over-flattened toward parity, with the strict screening verdict mapping-dependent and age-robust under direct standardisation. A transferability demonstration across six further NPK locales finds locale-dependent rather than universal diagnostics, with reference-taxonomy cardinality confounding cross-locale flag counts. For synthetic personas used as silicon samples, marginal claims must therefore be paired with disclosure-anchored joint audits before reuse. The released audit artefacts (reference manifests, occupational crosswalks, derived metrics, reproducibility scripts) instantiate this protocol on the NPK family and are released for retargeting at other synthetic persona resources.

05.
arXiv (CS.LG) 2026-06-15

Minimum Distance Summaries for Robust Neural Posterior Estimation

arXiv:2602.09161v2 Announce Type: replace-cross Abstract: Simulation-based inference (SBI) enables amortized Bayesian inference by first training a neural posterior estimator (NPE) on prior-simulator pairs, typically through low-dimensional summary statistics, which can then be cheaply reused for fast inference by querying it on new test observations. Because NPE is estimated under the training data distribution, it is susceptible to misspecification when observations deviate from the training distribution. Many robust SBI approaches address this by modifying NPE training or introducing error models, coupling robustness to the inference network and compromising amortization and modularity. We introduce minimum-distance summaries, a plug-in robust NPE method that adapts queried test-time summaries independently of the pretrained NPE. Leveraging the maximum mean discrepancy (MMD) as a distance between observed data and a summary-conditional predictive distribution, the adapted summary inherits strong robustness properties from the MMD. We demonstrate that the algorithm can be implemented efficiently with random Fourier feature approximations, yielding a lightweight, model-free test-time adaptation procedure. We provide theoretical guarantees for the robustness of our algorithm and empirically evaluate it on a range of synthetic and real-world tasks, demonstrating substantial robustness gains with minimal additional overhead.

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

RegMix-D: Dynamic Data Mixing via Proxy Training Trajectories

Data mixture selection is critical for Large Language Model pretraining. Existing methods such as RegMix select a single static mixture by fitting a regression model on small-scale proxy runs. We propose RegMix-D, a simple extension of RegMix to dynamic mixing. Our key observation is that proxy runs produce not only endpoint losses, but also full loss trajectories, which can be used to further improve data mixture. By training regression model on these trajectories, we can predict optimal mixtures at multiple training stages. RegMix-D supports two deployment modes: an offline variant that generates a complete mixture schedule before target training, and an online variant that adapts the mixture during training using observed loss. Experiments on 25B tokens of the Pile dataset with a 1B parameter target model show that RegMix-D consistently improves over RegMix and DoReMi across 13 downstream tasks while remaining proxy-efficient: it surpasses RegMix even with only 128 proxy models (25% of RegMix's proxy compute budget).

07.
arXiv (quant-ph) 2026-06-16

Black Hole–Entropy Container or Creator

arXiv:2603.18374v3 Announce Type: replace-cross Abstract: Do black holes possess entropy or do they create it? The dominant assumption is that they possess entropy, and a they evaporate that entropy is emitted and decreases. In this paper I use a model of a linear amplifier, in which I argue that the amplifier has not entropy and yet it emits entropy in the process of it operation. This model is closely related to behaviour of black holes, resulting in answer the question of that title that black holes do not have entropy, but nevertheless them create and emit entropy with the total entropy emitted being the same as the usual expression proportional to the square of the mass of the black hole.

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

Language-Instructed Vision Embeddings for Controllable and Generalizable Perception

Vision foundation models are typically trained as static feature extractors, placing the burden of task adaptation onto large downstream models. We propose an alternative paradigm: instead of solely feeding visual features into language models, we use language itself to dynamically guide the vision encoder. Our method, Language-Instructed Vision Embeddings (LIVE), leverages language as high-level guidance to produce task-centric embeddings at inference time, removing the need for task-specific retraining. This enables the encoder to focus on contextually relevant aspects of the input, yielding more controllable and generalizable representations. Empirically, LIVE reduces visual hallucinations (+34 points on MMVP), surpasses vision-language models with orders of magnitude more parameters on visual question answering, and generalizes to unseen instructions and tasks – offering a direct path toward adaptive, instruction-driven visual intelligence.

09.
arXiv (CS.AI) 2026-06-16

Green SARC: Predictive Cost and Carbon Governance for Agentic AI Systems

arXiv:2606.15954v1 Announce Type: cross Abstract: Agentic AI systems act through tools and sub-agents, yet the controls meant to bound their financial and environmental cost still sit on dashboards evaluated beside or after execution. Green SARC applies the SARC governance-by-architecture framework – four enforcement sites in the agent loop – to FinOps and GreenOps, contributing the theory of what to enforce and how to predict it. We report four policy-independent results. (i) The unconstrained "State Snowball" is $\Theta(n^2)$ in loop depth; on 3,000 real multi-step plans (SWE-rebench) it holds on 100%, with median curvature $\hat{c}_2=216$ exceeding the linear-accretion prediction $p/2=134$ – real plans accrete faster than the model. (ii) On real residuals the Normal-$\sigma$ gate under-covers (92% at nominal 95%); split-conformal calibration holds (95.2%). (iii) A soft Lagrangian penalty tuned to the budget in expectation breaches it on 91.5% of seeds; the architectural gate breaches 0%. (iv) Under binding budgets the gate's over-budget incidence is 0% on synthetic and real (BurstGPT) arrivals. End-to-end token/USD/carbon savings (47–55%) are real but policy-dependent in magnitude – set by a scope-cap knob, not by gate rejections. The library is open-source, dependency-free, and ships a regeneration script for every cited number.

10.
arXiv (quant-ph) 2026-06-19

Exploiting More Than Symmetry in Variational Quantum Machine Learning

arXiv:2606.20316v1 Announce Type: new Abstract: The success of variational quantum learning models crucially depends on choosing parametrizations that reflect the structure of the problem at hand. Symmetries provide one of the clearest such structures: whenever transformations of the input leave the desired outcome unchanged, this invariance should be built into the model rather than discovered during training. However, imposing a symmetry does not by itself determine a useful ansatz. Even within the symmetry-preserving space, one must decide where the trainable degrees of freedom should be placed. In this work, we study this remaining design freedom in equivariant variational quantum circuits. Building on symmetry-based parameter sharing, we disentangle two architectural choices: how much symmetry should be enforced, and which symmetry-respecting interactions should be trainable. Using Tic-Tac-Toe as a fully enumerable and structurally transparent test case, we find that suitable subgroups preserve most of the generalization benefit. By contrast, the dominant gains arise from gates acting directly on decisive task motifs. Thus, symmetry defines the admissible design space, while effective ansatze require an additional task-informed choice of trainable interactions.

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

SegmentAnyTreeV2: Scaling Transformer-Based Tree Instance Segmentation Across Sensors, Platforms, and Forests

We present SegmentAnyTreeV2, a sensor- and platform-agnostic framework for semantic and instance segmentation of forest point clouds. The model combines a serialization-based Point Transformer v3 backbone with a lightweight semantic head and a tree-focused cross-attention mask decoder. Semantic predictions restrict instance decoding to tree-class voxels, while instance-aware query initialization, one-to-many seed supervision, and asymmetric mask scoring improve separation in dense and structurally complex stands. We further introduce FOR-instance v3, an expanded benchmark comprising 427 scenes and 26,496 annotated trees across diverse biomes, forest structures, and LiDAR platforms. On the FOR-instanceV2 test split, SegmentAnyTreeV2 achieves 90.5% precision, 80.2% recall, 85.0% F1, 90.7% coverage, and 87.6% semantic mIoU, outperforming previous learning-based methods in both instance detection and mask completeness. Zero-shot evaluation on independent sites further demonstrates strong cross-domain generalization.

12.
arXiv (CS.CL) 2026-06-17

Scaling Enterprise Agent Routing: Degradation, Diagnosis, and Recovery

Production LLM assistants route user requests to growing libraries of specialized tools, but how does routing accuracy degrade as the catalog scales? We study single-step routing on a 110-agent, 584-tool catalog from a deployed enterprise productivity assistant, evaluating three frontier models from 10 to 110 agents. Routing F1 on under-specified requests drops 16–23 percentage points across models. An oracle analysis decomposes the degradation into a retrieval gap (the model cannot surface the right tool) and a confusion gap (even with perfect retrieval, the oracle ceiling drops 10pp). Embedding-based shortlisting recovers +10–11pp F1 at full scale across all three models and two providers. A production annotation study (1,435 human-labeled utterances, three annotators) confirms the recovery on real traffic at +10–17pp despite 10–15pp lower absolute performance.

14.
arXiv (CS.AI) 2026-06-15

The Journal of Prompt-Engineered (Moral) Philosophy Or: Why AI-Assisted Ethics Research Requires Process Transparency

Authors:

arXiv:2511.08639v4 Announce Type: replace-cross Abstract: Existing AI disclosure mandates in scholarship require that AI assistance be reported but leave transparency philosophically unspecified: they fix the duty without explaining what the duty serves. We argue that ethical inquiry is essentially contested at two independent levels – about what it is, and about what it demands of the inquirer – defeating output-only evaluation and welfare-economic dismissal of the transparency question, and, by extension, reproducibility framings imported from the empirical sciences. The transparency duty is grounded instead in agent-integrity: the legibility, before a community of inquiry, of the identity-constituting commitments that the author's mode of philosophising expresses. Because the standards for evaluating such work are not communally settled, the achievable goal for transparency is not evaluation against agreed criteria but tracking – accumulating the evidentiary record that lets each tradition assess the work on its own terms and makes future normative judgments possible. We develop a documentation-adequacy framework that operationalises Meaningful Human Control through five transparency elements – declaration, navigation, documentation account, process documentation, and development records – demonstrated by the paper itself, whose full documentation record is archived at a persistent identifier. The framework is a first iteration subject to revision, not a settled standard.

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

Position: Align AI to Our Aspirations, Not Our Flaws

arXiv:2606.13755v1 Announce Type: cross Abstract: We argue that aligning AI to aggregated human preferences is the wrong target. With current technology, one can train AIs to share the values of a Silicon Valley techno-optimist, a degrowth environmentalist, a national-conservative culture warrior, a single-party state cadre, or a devout religious traditionalist. We should not. Human values produce societies that thrive or fail on the merits of those values - from failed states and extreme inequality to declining happiness, political polarization, and government dysfunction in the world's wealthiest democracies. The pluralistic-alignment program correctly diagnoses that there is no single "humanity" to align with, but is dangerous if taken as the main directive. We argue that AI should be trained to a non-negotiable floor of objective alignment goals - competence, bounded by the constraints of factual accuracy, honesty, and lawfulness and that pluralism belongs at the surface (language, register, conventions, missing-context defaults) and across the wide band of legitimate value tradeoffs that respect the floor, but not at the level of values that violate it. We highlight the empirical reality of unfiltered pluralistic values, propose four commitments as a constructive alternative, and engage six credible objections: commercial pressure and practical feasibility, democratic legitimacy, regulatory compliance, over-reliance on institutionalist explanations, the charge that the floor itself is culturally laden, and the limits of Coherent Extrapolated Volition.

16.
Science (Express) 2026-06-18

Dynamic asymmetric strain imprinted into substrates by an oxide thin film | Science

Authors: Unknown Author

In film-substrate systems, the substrate role is often considered to be limited to providing static mechanical constraints. Dynamic film-substrate interactions when a structural change in the film modifies the substrate are generally disregarded. Using combined X-ray and electron microscopies, we observed that the electrically induced filament in a VO 2 film created strong asymmetric strain in the underlying Al 2 O 3 substate. This asymmetric substrate strain fed back into the film and defined the filament expansion direction, revealing the importance of film-substrate dynamic interactions in determining film functionality. Furthermore, the strain imprint propagated at least tens of microns deep into the substrate, exceeding the film thickness more than 200 times, potentially enabling substrate functionalization as an active mechanical coupling media in 3D-integrated microelectronics architectures.

17.
arXiv (CS.LG) 2026-06-19

Pseudo-Formalization for Automatic Proof Verification

arXiv:2605.20531v2 Announce Type: replace-cross Abstract: Reliable verification of proofs remains a bottleneck for training and evaluating AI systems on hard mathematical reasoning. Fully formal proofs, in languages like Lean, are easy to verify because they are unambiguous and modular. Most proofs, particularly those written by AI systems, have neither property, and translating them into formal languages remains challenging in many frontier math settings. We propose Pseudo-Formalization (PF), a proof format that captures the modularity and precision of formal proofs while retaining the flexibility of natural language. A Pseudo-Formal proof is decomposed into self-contained modules, each stating its premises, conclusion, and proof in natural language. To verify the correctness of a regular natural language proof, an LLM translates it to Pseudo-Formal and then verifies each module independently, an algorithm we call Block Verification (BV). We evaluate PF+BV on two benchmarks spanning olympiad and research-level mathematics, where it pareto-dominates LLM-as-judge baselines on error-finding precision and recall. To support future work, we release our research-level proof verification benchmark ArxivMathGradingBench.

18.
PLOS Computational Biology 2026-06-15

Fung-AI: An AI/ML-driven pipeline for antifungal peptide discovery

by Daniel S. Berman, Libby M. Lewis, Tom D. Curtis, Olivia N. Tiburzi, Daniel F. Q. Smith, Arturo Casadevall, Laura J. Dunphy Emerging fungal pathogens represent a concerning threat to both global health and food security. In this study, we aimed to address our rising vulnerability to fungal pathogens through the development of the Fung-AI pipeline: an AI/ML-driven approach for antifungal discovery. A generative adversarial network (GAN) was trained to generate novel candidate antifungal peptide sequences. Next, in silico antifungal and hemolytic classifiers were built to further prioritize AI-generated peptides for experimental validation. From a pool of ~10,000 candidates, thirteen peptides were selected for testing over two-stages of experimentation. Five peptides were found to display mild antifungal activity against the wheat pathogen, Fusarium graminearum, with minimal inhibitory concentrations (MICs) ranging from 250 µg/mL to 500 µg/mL. Four of the five peptides also showed activity against the human pathogen, Candida albicans (MIC: 500 µg/mL). Two of our AI-generated antifungal peptides additionally demonstrated low cytotoxicity in HepG2 human liver carcinoma cells (LC50 > 704.2 µg/mL) indicating that they may be useful as scaffolds for future optimization for therapeutic applications. None of our peptides were found to considerably inhibit the emerging pathogen C. auris, suggesting the need for pathogen-specific down-selection of candidate peptides. Overall, we present a proof-of-principle, generative-AI-based approach for the rapid design of de novo antifungal peptides.

19.
arXiv (CS.AI) 2026-06-12

Iterating Toward Better Search: A Two-Agent Simulation Framework for Evaluating Agentic Search Architectures in E-Commerce

arXiv:2606.12924v1 Announce Type: new Abstract: We present a modular two-agent simulation framework for evaluating conversational shopping assistant architectures. An independent buyer agent, configured with personas, missions, and patience levels, is paired with an interchangeable responder that integrates with a real e-commerce search API. Holding the buyer constant across experiments enables controlled comparison of responder designs on identical scenarios. Using 2011 conversations across 14 persona buckets, we establish four empirical findings. First, rolling-window memory outperforms intent-extraction memory on all quality metrics while being 35% faster per query. Second, illustrating rapid evidence-driven iteration, a systematic failure analysis of a responder version enables targeted fixes that reduce failure and near-failure rates by 62% across the full dataset. Third, swapping the responder LLM backbone from Gemini~2.5 to Llama~3.3~70B costs 0.16–0.45 points despite identical architecture. Finally, we document systematic philosophical disagreement between frontier LLM judges: Gemini rewards process correctness while Claude demands concrete outcomes, despite using the same evaluation prompt.

20.
arXiv (CS.CL) 2026-06-19

LedgerAgent: Structured State for Policy-Adherent Tool-Calling Agents

Policy-adherent tool-calling agents in customer-service domains must maintain task states across turns while calling tools and obeying domain policies. Task states consist of relevant facts, identifiers, constraints, and conditions observed through user interaction and tool calls. In standard agents, task states are not represented separately. Observations, tool returns, and policy instructions are placed in the prompt, leaving agents to reconstruct the relevant states from the prompt each time they decide what to do next. This design makes state management implicit, creating two common failure modes. An agent may retrieve the right facts but later ground its decision in stale, missing, or incorrect information; and a syntactically valid tool call may still violate a domain policy that depends on the current task state. We introduce \textsc{LedgerAgent}, an inference-time method for tool-calling agents that maintains observed task states in a separate ledger and renders the states into the prompt. The ledger is also used to check state-dependent policy constraints before environment-changing tool calls are executed, blocking policy violations. Across four customer-service domains and a mixed panel of open- and closed-weight models, \textsc{LedgerAgent} improves average pass\textasciicircum{}k over a standard prompt-based tool-calling approach, with the largest gains under stricter multi-trial consistency metrics.

21.
arXiv (CS.LG) 2026-06-17

HeteRo-Select: Informativeness as the Participation Driver in Heterogeneous Federated Learning

arXiv:2508.06692v2 Announce Type: replace Abstract: Federated learning systems typically allocate gradient compression by link speed. This is sensible when bandwidth and data informativeness align. However, under non-IID data, these signals often decorrelate or invert. A bandwidth-driven allocator then risks compressing the most informative gradients hardest. We propose HeteRo-Select, a framework that replaces bandwidth with a per-client informativeness score as the primary driver of compression. The score jointly governs three decisions per round: client selection, compression ratio, and server aggregation weight, with bandwidth retained only as a hard ceiling. Score-proportional selection provably reduces the effective heterogeneity of the chosen subset; score-proportional compression provably lowers aggregate top-$k$ error at fixed traffic. Under the exact FedCG simulation protocol, HeteRo-Select delivers a $1.78\times$ speedup and an $18.2\%$ reduction in traffic on CIFAR-10. The same configuration, unchanged, scales from a $7{,}850$-parameter logistic regression to an $11.27$M-parameter ResNet-18, hitting the accuracy target on three of four benchmarks. When bandwidth and informativeness are deliberately anti-correlated, the method still achieves the target accuracy with less traffic than the normal-bandwidth run.

22.
arXiv (CS.AI) 2026-06-19

Measuring Biological Capabilities and Risks of AI Agents

arXiv:2606.19899v1 Announce Type: cross Abstract: This paper addresses a rapidly emerging policy challenge: how to generate and interpret credible evidence about the biological capabilities and risks of AI scientists, or agentic AI systems capable of autonomously or collaboratively performing multi-step scientific tasks. As these systems enter real research workflows, decision-makers increasingly face evaluation results whose meaning depends on underlying design choices that are often implicit or under-documented. We synthesize current evidence on AI-enabled biological risks and introduce biological agentic evaluations as a promising, but interpretation-sensitive, tool for assessing these systems. Our central contribution is a set of practical, experience-grounded considerations – drawing from our own evaluations – that show how choices around defining, designing, running, scoring, and documenting evaluations materially shape what results do and do not imply about risk. The analysis is intended to help policymakers interpret biological evaluation outputs with appropriate caution; guide public and private funders toward high-leverage investments in AI-biology evaluation research; and support biosecurity practitioners assessing emerging AI systems. A secondary audience includes researchers designing or conducting agentic evaluations within frontier AI labs, AI providers, scientific institutions, and third-party evaluation organizations.

23.
arXiv (CS.AI) 2026-06-15

Causal Object-Centric Models for Planning with Monte Carlo Tree Search

arXiv:2606.14418v1 Announce Type: new Abstract: We introduce COMET (Causal Object-centric Model for Efficient Tree search), a model-based reinforcement learning algorithm that performs Monte Carlo Tree Search in a slot-structured latent space. COMET pairs a frozen unsupervised object-centric encoder with a transformer-based world model, in which actions are bound to objects through a novel action-slot fusion mechanism that is used in slot transition prediction. Policy and value heads use object-causal attention, modulating token interactions by learned per-slot relevance scores so that decision-making concentrates on task-relevant entities. COMET adds an explicit object-level inductive bias to MuZero-style latent planning. Across eight visually and dynamically diverse tasks from the Object-Centric Visual RL benchmark, ManiSkill, Robosuite, and VizDoom, COMET achieves a higher mean normalized score during the early stages of training compared to object-centric and monolithic baselines.

24.
arXiv (CS.CL) 2026-06-11

GrowLoop: Self-Evolving Conversation Evaluation Seeded by Human

With the rapid advancement of large language models, evaluating human-likeness in open-ended conversation has become increasingly important. However, human-likeness is a form of tacit knowledge that humans perceive intuitively, yet the underlying criteria resist explicit formulation. Human judgments vary widely, with strong agreement on some cases and legitimate disagreement on others. Meanwhile, the criteria behind human judgments remain implicit, leaving no clear basis for constructing cases. Further, what counts as human-likeness is not static, but evolving with model capability and human expectations. Despite progress in evaluation methods such as expert-authored benchmarks, Reward Models, and self-evolving benchmarks, none addresses all three challenges simultaneously. Therefore, we propose GrowLoop, a self-evolving conversation evaluation system that continuously adapts as models advance and scenarios shift. Starting from minimal human seed annotations, LLM agents iteratively extract and refine evaluation rubrics through Heuristic Learning. Human-AI agreement is required where annotators converge, while only plausibility is expected where they diverge. Moreover, the Rubric-Case co-evolution mechanism enables continuous evolution. When the evaluation target shifts, new human seeds expand the system's coverage accordingly. When applied to human-likeness evaluation in open-ended conversation, the AI judge guided by these rubrics not only substantially outperforms existing methods in alignment with human judgments, but also uncovers issues that annotators overlook. The resulting benchmark effectively discriminates models across capability tiers and reveals where they fall short, while generalizing to new scenarios and adapting as models advance. Our work shifts the benchmarking paradigm from manual updates or difficulty scaling to comprehensive, continuous self-evolution.

25.
arXiv (CS.AI) 2026-06-18

DeFAb: A Verifiable Benchmark for Defeasible Abduction in Foundation Models

arXiv:2606.18557v1 Announce Type: new Abstract: A rule-based logic solver resolves every instance in our benchmark in under 50 microseconds with 100% accuracy; the best frontier language model reaches 65% at best and drops to 23.5% under rendering-robust evaluation (worst case over four surface renderings). We introduce DeFAb (Defeasible Abduction Benchmark), a dataset and generation pipeline that converts four decades of publicly funded knowledge bases into formally grounded instances for defeasible abduction: constructing hypotheses that explain anomalies by overriding defaults while preserving unrelated expectations. Because every hypothesis must pass polynomial-time checks for valid derivation, conservativity, and minimality, DeFAb makes logical rigor the instrument for measuring creativity and theoretical reasoning, scoring the disciplined construction of theory revisions rather than fluent but theory-destroying prose. The pipeline pairs taxonomic hierarchies (OpenCyc, YAGO, Wikidata) with behavioral property graphs (ConceptNet, UMLS) to produce 372,648+ instances across 33.75M materialized rules from 18 sources, in three levels with polynomial-time verifiable gold standards. Four frontier models do not reliably internalize defeasible reasoning: rendering-robust Level 2 accuracy is 7.8-23.5%; chain-of-thought variance (~36 pp) exceeds any inter-model gap; and a matched contamination control isolates a +19.4 pp Level 3 gap. We further release DeFAb-Hard (a 235-instance Level 3 difficulty variant; best model 53.3% vs 100% symbolic) and CONJURE (a kernel-verified transformative-creativity variant of 560 Lean 4/Mathlib instances whose gold answers are definitions the proof kernel did not previously contain, judge-free verifier; a pilot finds zero novel concepts). The same verifier doubles as an exact reward for preference optimization (DPO, RLVR/GRPO). Released under MIT at https://huggingface.co/datasets/PatrickAllenCooper/DeFAb.