arXiv:2510.21712v2 Announce Type: replace
Abstract: Retrieval-Augmented Generation (RAG) systems have emerged as a pivotal methodology for enhancing Large Language Models (LLMs) through the dynamic integration of external knowledge. To further improve RAG's flexibility, Agentic RAG introduces autonomous agents into the workflow. However, Agentic RAG faces several challenges: (1) the success of each step depends on both high-quality planning and accurate search, (2) the lack of supervision for intermediate reasoning steps, and (3) the exponentially large candidate space for planning and searching. To address these challenges, we propose DecoupleSearch, a novel framework that decouples planning and search processes using dual value models, enabling independent optimization of plan reasoning and search grounding. Our approach constructs a reasoning tree, where each node represents planning and search steps. We leverage Monte Carlo Tree Search to assess the quality of each step. During inference, Hierarchical Beam Search iteratively refines planning and search candidates with dual value models. Extensive experiments across policy models of varying parameter sizes demonstrate the effectiveness of our method.
Science Journals
arXiv:2511.17392v3 Announce Type: replace
Abstract: Deformable image registration (DIR) remains a fundamental yet challenging problem in medical image analysis, largely due to the prohibitively high-dimensional deformation space of dense displacement fields and the scarcity of voxel-level supervision. Existing reinforcement learning frameworks often project this space into coarse, low-dimensional representations, limiting their ability to capture spatially variant deformations. We propose MorphSeek, a fine-grained representation-level policy optimization paradigm that reformulates DIR as a spatially continuous optimization process in the latent feature space. MorphSeek introduces a stochastic Gaussian policy head atop the encoder to model a distribution over latent features, facilitating efficient exploration and coarse-to-fine refinement. The framework integrates unsupervised warm-up with weakly supervised fine-tuning through Group Relative Policy Optimization, where multi-trajectory sampling stabilizes training and improves label efficiency. Across three 3D registration benchmarks (OASIS brain MRI, LiTS liver CT, and Abdomen MR-CT), MorphSeek achieves consistent Dice improvements over competitive baselines while maintaining high label efficiency with minimal parameter cost and low step-level latency overhead. Beyond optimizer specifics, MorphSeek advances a representation-level policy learning paradigm that achieves spatially coherent and data-efficient deformation optimization, offering a principled, backbone-agnostic, and optimizer-agnostic solution for scalable visual alignment in high-dimensional settings.
arXiv:2512.03280v2 Announce Type: replace
Abstract: The conceptual design of Blended Wing Body (BWB) aircraft is often constrained by the high computational cost of resolving complex aerodynamics over a high-dimensional design space. While deep learning offers a pathway to rapid aerodynamic prediction and inverse design, its adoption in aerospace engineering is limited by a lack of large-scale, field-resolved training data. This work addresses this gap by introducing BlendedNet++, a comprehensive aerodynamic dataset comprising 12,492 unique BWB geometries, each evaluated using steady Reynolds-Averaged Navier--Stokes (RANS) simulations to provide integrated forces and dense surface fields (Cp, Cf). Leveraging this data, we establish a robust framework for two critical engineering tasks: (1) real-time prediction of surface aerodynamic fields using geometric deep learning models, and (2) generative inverse design. We benchmark five surrogate architectures, identifying Transolver as the most accurate for field predictions. Furthermore, we demonstrate a generative inverse design pipeline using conditional diffusion models combined with gradient-based refinement. This hybrid approach is shown to generate multiple feasible designs that satisfy specific lift-to-drag targets with high accuracy (R^2 > 0.99), as confirmed by computational fluid dynamics (CFD) simulation. These resources enable a shift from iterative analysis to direct generation in early-stage BWB design.
arXiv:2601.06633v2 Announce Type: replace
Abstract: Open-ended tasks, such as coding problems that are common in computer science education, provide detailed insights into student knowledge. However, training large language models (LLMs) to simulate and predict possible student errors in their responses to these problems can be challenging: they often suffer from mode collapse and fail to fully capture the diversity in syntax, style, and solution approach in student responses. In this work, we present KASER (Knowledge-Aligned Student Error Simulator), a novel approach that aligns errors with student knowledge. We propose a training method based on reinforcement learning using a hybrid reward that reflects three aspects of student code prediction: i) code similarity to the ground-truth, ii) error matching, and iii) code prediction diversity. On two real-world datasets, we perform two levels of evaluation and show that: At the per-student-problem pair level, our method outperforms baselines on code and error prediction; at the per-problem level, our method outperforms baselines on error coverage and simulated code diversity.
arXiv:2601.16287v3 Announce Type: replace
Abstract: Active learning for photonic crystals explores the integration of analytic approximate Bayesian last layer neural networks (LL-BNNs) with uncertainty-driven sample selection to accelerate photonic band gap prediction. We employ an analytic LL-BNN formulation, corresponding to the infinite Monte Carlo sample limit, to obtain uncertainty estimates that are strongly correlated with the true predictive error on unlabeled candidate structures. These uncertainty scores drive an active learning strategy that prioritizes the most informative simulations during training. Applied to the task of predicting band gap sizes in two-dimensional, two-tone photonic crystals, our approach achieves up to a 2.7x reduction on average in required training data compared to a random sampling baseline while maintaining predictive accuracy. The efficiency gains arise from concentrating computational resources on high uncertainty regions of the design space rather than sampling uniformly. Given the substantial cost of full band structure simulations, especially in three dimensions, this data efficiency enables rapid and scalable surrogate modeling. Our results suggest that analytic LL-BNN based active learning can substantially accelerate topological optimization and inverse design workflows for photonic crystals, and more broadly, offers a general framework for data efficient regression across scientific machine learning domains.
arXiv:2601.17240v2 Announce Type: replace
Abstract: In-person small-group conversations play a crucial role in everyday life; however, facilitating effective group interaction can be challenging, as the real-time nature demands full attention, offers no opportunity for revision, and requires interpreting non-verbal cues. Using Mixed Reality to provide proactive information support shows promise in helping individuals engage in and contribute to group conversations. We present a preliminary participatory design and qualitative study (N = 10) using focus groups and two technology probes to explore the opportunities of designing proactive information support in in-person small-group conversations. We reveal key design opportunities concerning how to maximize the benefits of proactive information support and how to effectively design such supporting information. Our study is crucial for paving the way toward designing future proactive AI agents to enable the paradigm of augmented in-person small-group conversation experience.
arXiv:2601.19667v3 Announce Type: replace
Abstract: We present SynCABEL (Synthetic Contextualized Augmentation for Biomedical Entity Linking), a framework that addresses a central bottleneck in supervised biomedical entity linking (BEL): the scarcity of expert-annotated training data. SynCABEL leverages large language models to generate context-rich synthetic training examples for all candidate concepts in a target knowledge base, providing broad supervision without manual annotation. We demonstrate that SynCABEL, when combined with decoder-only models and guided inference, establishes new state-of-the-art results across three widely used multilingual benchmarks: MedMentions for English, QUAERO for French, and SPACCC for Spanish. Evaluating data efficiency, we show that SynCABEL reaches the performance of full human supervision using up to 60% less annotated data, substantially reducing reliance on labor-intensive and costly expert labeling. Finally, acknowledging that standard evaluation based on exact code matching often underestimates clinically valid predictions due to ontology redundancy, we introduce an LLM-as-a-judge protocol. This analysis reveals that SynCABEL significantly improves the rate of clinically valid predictions. Our synthetic datasets, models, and code are released to support reproducibility and future research.
arXiv:2601.21170v3 Announce Type: replace
Abstract: Emergent phenomena -- onset of epileptic seizures, sudden customer churn, or pandemic outbreaks -- often arise from hidden causal interactions in complex systems. We propose a machine learning method for their early detection that addresses a core challenge: unveiling and harnessing a system's latent causal structure despite the data-generating process being unknown and partially observed. The method learns an optimal feature representation from a one-parameter family of estimators -- powers of the empirical covariance or precision matrix -- offering a principled way to tune in to the underlying structure driving the emergence of critical events. A supervised learning module then classifies the learned representation. We prove structural consistency of the family and demonstrate the empirical soundness of our approach on seizure detection and churn prediction, attaining competitive results in both. Beyond prediction, and toward explainability, we ascertain that the optimal covariance power exhibits evidence of good identifiability while capturing structural signatures, thus reconciling predictive performance with interpretable statistical structure.
arXiv:2601.21458v2 Announce Type: replace
Abstract: Modern deepfakes have evolved into localized and intermittent manipulations that require fine-grained temporal localization to mitigate severe digital security risks. The prohibitive cost of frame-level annotation makes weakly supervised methods a practical necessity, which rely only on video-level labels. To this end, we propose Reconstruction-based Temporal Deepfake Localization (RT-DeepLoc), a weakly supervised temporal forgery localization framework that identifies forgeries via reconstruction errors. Our framework uses a Masked Autoencoder (MAE) trained exclusively on authentic data to learn its intrinsic spatiotemporal patterns; this allows the model to produce significant reconstruction discrepancies for forged segments, effectively providing the missing fine-grained cues for accurate localization without demanding dense human annotations. To robustly leverage these indicators, we introduce a novel Asymmetric Intra-video Contrastive Loss (AICL). By focusing on the compactness of authentic features guided by these reconstruction cues, AICL establishes a stable decision boundary that enhances local discrimination while preserving generalization to unseen forgeries by advanced generative models. Extensive experiments on large-scale datasets, including LAV-DF, demonstrate that RT-DeepLoc achieves state-of-the-art performance in weakly-supervised temporal forgery localization.
arXiv:2601.21531v2 Announce Type: replace
Abstract: Visual token compression is widely used to accelerate large vision-language models (LVLMs) by pruning or merging visual tokens, yet its adversarial robustness remains unexplored. We show that existing encoder-based attacks cannot fully disclose the robustness vulnerabilities of compressed LVLMs, due to an optimization-inference mismatch: perturbations are optimized on the full-token representation, while inference is performed through a token-compression bottleneck. To address this gap, we propose the Compression-AliGnEd attack (CAGE), which aligns perturbation optimization with compression inference without assuming access to the deployed compression mechanism or its token budget. CAGE combines (i) expected feature disruption, which concentrates distortion on tokens likely to survive across plausible budgets, and (ii) rank distortion alignment, which actively aligns token distortions with rank scores to promote the retention of highly distorted evidence. Across diverse representative plug-and-play compression mechanisms and datasets, our results show that CAGE consistently achieves lower robust accuracy than the baseline. This work highlights that robustness assessments ignoring compression can be overly optimistic, calling for compression-aware security evaluation and defenses for efficient LVLMs.
arXiv:2601.22530v2 Announce Type: replace
Abstract: Large Language Models (LLMs) struggle with multi-step reasoning over structured tables. The primary reason is the lack of explicit supervision for intermediate reasoning states. Existing learned reward models or executor-based verifiers are either unscalable or rely on answer-checking environments unavailable for many tabular tasks. This leaves no signal that is scalable and grounded in the query. To address this, we introduce TABROUGE, a training-free and deterministic state reward. By adapting the Longest Common Subsequence (LCS) metric from text summarization to evaluate tabular states, TABROUGE assesses the lexical coverage and structural integrity of intermediate tables against the query without requiring learned models or external executors. Built upon this metric, we propose RE-TAB, a plug-and-play, training-free framework. RE-TAB reframes table reasoning as deterministic control over intermediate states, utilizing TABROUGE for stepwise feedback and trajectory-level test-time scaling (TTS) signals. Across six backbones and three benchmarks, RE-TAB improves accuracy by an average of 26.7 pp over no-reward baselines. It also reduces TTS samples by up to 33%. Preliminary GRPO experiments further indicate TABROUGE's viability as a scalable post-training reward, increasing gains by 8.34 pp. We further analyze failure modes of TABROUGE, including paraphrase under-rewarding and echo-column hacking, and identify when structure-aware lexical rewards remain reliable.
arXiv:2601.22678v3 Announce Type: replace
Abstract: Full-graph and mini-batch Graph Neural Network (GNN) training approaches have distinct system design demands, making it crucial to choose the appropriate approach to develop. A core challenge in comparing these two GNN training approaches lies in characterizing their model performance (i.e., convergence and generalization) and computational efficiency. While a batch size has been an effective lens in analyzing such behaviors in deep neural networks (DNNs), GNNs extend this lens by introducing a fan-out size, as full-graph training can be viewed as mini-batch training with the largest possible batch size and fan-out size. However, the impact of the batch and fan-out size for GNNs remains insufficiently explored. To this end, this paper systematically compares full-graph vs. mini-batch training of GNNs through empirical and theoretical analyses from the view points of the batch size and fan-out size. Our key contributions include: 1) We provide a novel generalization analysis using the Wasserstein distance to study the impact of the graph structure, especially the fan-out size. 2) We uncover the non-isotropic effects of the batch size and the fan-out size in GNN convergence and generalization, providing practical guidance for tuning these hyperparameters under resource constraints. Finally, full-graph training does not always yield better model performance or computational efficiency than well-tuned smaller mini-batch settings. The implementation can be found in the github link: https://github.com/LIUMENGFAN-gif/GNN_fullgraph_minibatch_training.
arXiv:2601.23154v2 Announce Type: replace
Abstract: Pain management in intensive care usually involves complex trade-offs, since both inadequate and excessive treatment can compromise patient safety. Prior work on reinforcement learning for sedation and analgesia has explored how to optimize these interventions, but has not considered patient survival or partial observability. To investigate the risks of these design choices, we developed an offline deep reinforcement learning framework that suggests hourly medication doses based on recurrent state representations. Using retrospective data from 47,144 ICU stays in the MIMIC-IV database, we trained and evaluated behavior-regularized actor-critic models that prescribe continuous doses of opioids, propofol, benzodiazepines, and dexmedetomidine according to two goals: reduce pain or jointly reduce pain and 30-day post-discharge mortality. Although the two resulting policies were associated with lower pain, clinician agreement with the pain-only policy was positively correlated with mortality ($\rho$=0.119, p<0.0001), while agreement with the joint policy was negatively correlated ($\rho$=-0.316, p<0.0001). We found that such divergence arose from a different response to high levels of comorbidity. This suggests that valuing post-discharge outcomes could be critical for learning safer treatment policies, even if a short-term goal remains the primary objective.
arXiv:2602.01705v3 Announce Type: replace
Abstract: Reinforcement learning has become a central paradigm for improving LLM reasoning, but most existing methods optimize policies over discrete token sequences. This creates a mismatch between the optimization space and the structure of reasoning: many important decisions are semantic, global, and trajectory-level rather than local token choices. Continuous latent-space RL offers a promising alternative by allowing policies to explore higher-level reasoning representations. However, simply moving to latent space is not sufficient. The resulting policy must model a complex, multi-modal distribution over valid reasoning trajectories. We therefore propose Latent Diffusion Reasoning with Reinforcement Learning (LaDi-RL), where a diffusion model generates latent reasoning trajectories through iterative denoising. This formulation enables structured exploration and expressive distribution modeling, but also introduces a fundamental credit-assignment challenge: the policy acts in latent space, while rewards are observed only after the latent is decoded into text. A naive rollout strategy therefore entangles latent reasoning quality with text decoding quality, making it unclear whether an incorrect answer results from a poor latent trajectory or from an imperfect textual realization. To address this, we introduce hierarchical latent-text rollouts. We sample multiple text completions for each latent trajectory and aggregate their rewards to obtain a decoder-marginalized estimate of latent utility. This provides a cleaner and lower-variance reward signal for optimizing the diffusion policy. Empirically, LaDi-RL outperforms token-level RL by 9.4% on code generation and 5.7% on math reasoning in pass@1, and even surpasses the base model's pass@k performance.
arXiv:2602.02236v4 Announce Type: replace
Abstract: We study online fine-tuning of pretrained control policies for autonomous driving using Real-Time Recurrent Reinforcement Learning (RTRRL), a memory-efficient algorithm that updates policy parameters at every time step without backpropagation through time. We extend RTRRL to support LrcSSM, a recently proposed nonlinear diagonal state-space model, and combine offline behavioral cloning with online RTRRL fine-tuning to adapt policies to distribution shifts at deployment. We validate the approach in the CarRacing simulation and on a 1:10-scale RoboRacer platform equipped with an event camera, where a pretrained policy is fine-tuned online during real-world line-following. To our knowledge, this is the first demonstration of online RL fine-tuning with event-camera observations on standard (non-spiking) hardware in closed-loop control. LrcSSM-based policies improve fastest and most consistently across both settings.
arXiv:2509.21820v2 Announce Type: replace
Abstract: In this paper, we introduce a combination of novel and exciting tasks: the solution and generation of linguistic puzzles. We focus on puzzles used in Linguistic Olympiads for high school students. We first extend the existing benchmark for the task of solving linguistic puzzles. We explore the use of Large Language Models (LLMs), including recent state-of-the-art models such as OpenAI's o1, for solving linguistic puzzles, analyzing their performance across various linguistic topics. We demonstrate that LLMs outperform humans on most puzzles types, except for those centered on writing systems, and for the understudied languages. We use the insights from puzzle-solving experiments to direct the novel task of puzzle generation. We believe that automating puzzle generation, even for relatively simple puzzles, holds promise for expanding interest in linguistics and introducing the field to a broader audience. This finding highlights the importance of linguistic puzzle generation as a research task: such puzzles can not only promote linguistics but also support the dissemination of knowledge about rare and understudied languages.
arXiv:2602.08354v5 Announce Type: replace
Abstract: Recent advancements in large reasoning models (LRMs) have greatly improved their capabilities on complex reasoning tasks through Long Chains of Thought (CoTs). However, this approach often results in substantial redundancy, impairing computational efficiency and causing significant delays in real-time applications. Recent studies show that longer reasoning chains are frequently uncorrelated with correctness and can even be detrimental to accuracy. In a further in-depth analysis of this phenomenon, we surprisingly uncover and empirically verify that LRMs implicitly know the appropriate time to stop thinking, while this capability is obscured by current sampling paradigms. Motivated by this, we introduce SAGE (Self-Aware Guided Efficient Reasoning), a novel sampling paradigm that unleashes this efficient reasoning potential. Furthermore, integrating SAGE as mixed sampling into group-based reinforcement learning (SAGE-RL) enables SAGE-RL to effectively incorporate SAGE-discovered efficient reasoning patterns into standard pass@1 inference, markedly enhancing both the reasoning accuracy and efficiency of LRMs across multiple challenging mathematical benchmarks.
arXiv:2509.21092v2 Announce Type: replace
Abstract: This study investigates how the majority group influences individual judgment formation and expression in anonymous, spontaneous online conversations. Drawing on theories of social conformity and anti-conformity, we analyze everyday dilemmas discussed on social media. First, using digital traces to operationalize judgments, we measure the conversations' disagreement and apply Bayesian regression to capture shifts of judgments formation before and after the group's exposure. Then we analyze changes in judgment expression with a linguistic analysis of the motivations associated with each judgment. Results show anti-conformity behaviors: individuals preserve the majority's positive or negative orientation of judgments but diverge from its stance, with persuasive language increasing post-disclosure. Our findings highlight how online environments reshape social influence compared to offline contexts.
arXiv:2509.20881v2 Announce Type: replace
Abstract: Code retrieval aims to find relevant code snippets matching natural language queries within massive codebases, playing a vital role in software development. Recent advances leverage PLMs to bridge the semantic gap between natural language (NL) and programming languages (PL), significantly outperforming traditional information retrieval and early deep learning approaches. However, existing methods still face key challenges, including a fundamental semantic gap between human intent and machine execution logic, and limited robustness to diverse code styles. To address this, we propose PseudoBridge, a novel code retrieval framework that introduces pseudo-code as an intermediate, semi-structured modality to align NL semantics with PL logic. Specifically, PseudoBridge consists of two stages: First, we employ an LLM to synthesize pseudo-code, enabling explicit alignment between NL queries and pseudo-code. Second, we introduce a logic-invariant code style augmentation strategy, employing the LLM to generate stylistically diverse yet logically equivalent code implementations, and then align these varied code styles with pseudo-code to enhance robustness. We evaluate PseudoBridge across 10 PLMs and 6 mainstream programming languages. Extensive experiments demonstrate that PseudoBridge consistently outperforms baselines, achieving significant improvements in generalization, particularly in zero-shot scenarios like Solidity and XLCoST. Extended evaluations using open-source LLMs and advanced embeddings confirm that these gains stem from PseudoBridge's intrinsic design, independent of specific closed-source models. PseudoBridge achieves performance comparable to SOTA embedding methods, highlighting the effectiveness of explicit logical and semantic alignment via pseudo-code as a robust solution for code retrieval.
arXiv:2602.16990v2 Announce Type: replace
Abstract: Most recommendation benchmarks evaluate how well a model imitates user behavior. In financial advisory, however, observed actions can be noisy or short-sighted under market volatility and may conflict with a user's long-term goals. Treating what users chose as the sole ground truth, therefore, conflates behavioral imitation with decision quality. We introduce Conv-FinRe, a conversational and longitudinal benchmark for stock recommendation that evaluates LLMs beyond behavior matching. Given an onboarding interview, step-wise market context, and advisory dialogues, models must generate rankings over a fixed investment horizon. Crucially, Conv-FinRe provides multi-view references that distinguish descriptive behavior from normative utility grounded in investor-specific risk preferences, enabling diagnosis of whether an LLM follows rational analysis, mimics user noise, or is driven by market momentum. We build the benchmark from real market data and human decision trajectories, instantiate controlled advisory conversations, and evaluate a suite of state-of-the-art LLMs. Results reveal a persistent tension between rational decision quality and behavioral alignment: models that perform well on utility-based ranking often fail to match user choices, whereas behaviorally aligned models can overfit short-term noise. The dataset is publicly released on Hugging Face, and the codebase is available on GitHub.
arXiv:2603.01993v2 Announce Type: replace
Abstract: Recent advances in generative AI have significantly enhanced the realism of multimodal media manipulation, thereby posing substantial challenges to manipulation detection. Existing manipulation detection and grounding approaches predominantly focus on manipulation type classification under result-oriented supervision, which not only lacks interpretability but also tends to overfit superficial artifacts. In this paper, we argue that generalizable detection requires incorporating explicit forensic reasoning, rather than merely classifying a limited set of manipulation types, which fails to generalize to unseen manipulation patterns. To this end, we propose REFORM, a reasoning-driven framework that shifts learning from outcome fitting to process modeling. REFORM adopts a three-stage curriculum that first induces forensic rationales, then aligns reasoning with final judgments, and finally refines logical consistency via reinforcement learning. To support this paradigm, we introduce ROM, a large-scale dataset with rich reasoning annotations. Extensive experiments show that REFORM establishes new state-of-the-art performance with superior generalization, achieving 81.52% ACC on ROM, 76.65% ACC on DGM4, and 74.9 F1 on MMFakeBench.
arXiv:2602.03033v2 Announce Type: replace
Abstract: MetaML-style metaprogramming languages allow programmers to construct, manipulate and run code. In the presence of higher-order references for code, ensuring type safety is challenging, as free variables can escape their binders. In this paper, we present Contextual MetaML, \textit{the first metaprogramming language that supports storing and running open code under a strong type safety guarantee}. The type system utilises contextual modal types to track and reason about free variables in code explicitly.
A crucial concern in metaprogramming-based program optimisations is whether the optimised program preserves the meaning of the original program. Addressing this question requires a notion of program equivalence and techniques to reason about it. In this paper, we provide a semantic model that captures contextual equivalence for Contextual MetaML, establishing \textit{the first full abstraction result for an imperative MetaML-style language}. Our model is based on traces derived via operational game semantics, where the meaning of a program is modelled by its possible interactions with the environment. We also establish a novel closed instances of use theorem that accounts for both call-by-value and call-by-name closing substitutions.
arXiv:2509.19824v2 Announce Type: replace
Abstract: Tube-based Model Predictive Control (MPC) is a widely adopted robust control framework for constrained linear systems under additive disturbance. The paper is focused on reducing the numerical complexity associated with the tube parameterization, described as a sequence of elastically-scaled zonotopic sets. A new class of scaled-zonotope inclusion conditions is proposed, alleviating the need for a priori specification of certain set-containment constraints and achieving significant reductions in complexity. A comprehensive complexity analysis is provided for both the polyhedral and the zonotopic setting, illustrating the trade-off between an enlarged domain of attraction and the required computational effort. The proposed approach is validated through extensive numerical experiments.
arXiv:2509.19590v2 Announce Type: replace
Abstract: Evaluations of generative models are now ubiquitous, and their outcomes critically shape public and scientific expectations of AI's capabilities. Yet skepticism about their reliability continues to grow. How can we know that a reported accuracy genuinely reflects a model's underlying performance? Although benchmark results are often presented as direct measurements of capability, in practice they are inferences: treating a score as evidence of capability already presupposes a theory of what it means to be capable at a task.
We argue that AI evaluations should instead be framed as inference tasks grounded on an explicit theory of capability. While this perspective is standard in fields like psychometrics, it remains underdeveloped in AI evaluation, where core assumptions are often left implicit. As a proof-of-concept, we empirically show that reported performance can depend strongly on the evaluator's modeling assumptions, underscoring the need for transparent, theory-driven evaluation practices. We conclude by offering an Evaluation Card to help researchers document, justify, and scrutinize the modeling decisions underlying AI evaluations.
arXiv:2509.19102v2 Announce Type: replace
Abstract: General-purpose robotic skills from end-to-end demonstrations often leads to task-specific policies that fail to generalize beyond the training distribution. Therefore, we introduce FunCanon, a framework that converts long-horizon manipulation tasks into sequences of action chunks, each defined by an actor, verb, and object. These chunks focus policy learning on the actions themselves, rather than isolated tasks, enabling compositionality and reuse. To make policies pose-aware and category-general, we perform functional object canonicalization for functional alignment and automatic manipulation trajectory transfer, mapping objects into shared functional frames using affordance cues from large vision language models. An object centric and action centric diffusion policy FuncDiffuser trained on this aligned data naturally respects object affordances and poses, simplifying learning and improving generalization ability. Experiments on simulated and real-world benchmarks demonstrate category-level generalization, cross-task behavior reuse, and robust sim2real deployment, showing that functional canonicalization provides a strong inductive bias for scalable imitation learning in complex manipulation domains. Details of the demo and supplemental material are available on our project website https://sites.google.com/view/funcanon.