Agent Trust Infrastructure (ATI): A Federated Identity, Trust, Attestation, and Accountability Architecture for Cross-Organization Autonomous AI Agents
AI agents increasingly collaborate across organizational boundaries, invoking tools and interacting with agents operated by independent entities. Although existing agent frameworks support planning and coordination, they lack a common mechanism for verifying cross-organizational agent identity, execution integrity, and authorization. We propose the Agent Trust Infrastructure (ATI), a framework-independent security architecture built on Public Key Infrastructure (PKI), Zero Trust Architecture (ZTA), and remote attestation. ATI establishes federated trust through peer organizational certificate authorities (AgentCAs), eliminating the need for a shared root authority. Before executing requests, ATI validates agent certificates, verifies approved model, tool, and system-prompt configurations, and enforces authorization through three policy checks. We implement ATI as an external security enforcement layer that integrates with existing agent frameworks without modifying their underlying models. We evaluate ATI in a simulated federation of eight organizations using six agent frameworks, each deploying a five-agent team. Security experiments involve two organizations and assess 14 attack scenarios and six benign tasks across five agent models, using five attack variations and three repetitions per variation. A panel of judge models evaluates the experimental outcomes. With all enforcement mechanisms enabled, attack containment increases from 43% to 84% (40% to 80% for the 11 attack scenarios evaluated consistently across all configurations), while enforcement introduces approximately 174 microseconds of overhead per request. ATI is released as an open-source, framework-independent package that uses IETF WIMSE workload token formats for identity and request verification and Entity Attestation Tokens for model, tool, and system-prompt integrity verification.
Small open-weight language models now run on a laptop, but they fail at multi-step work in ways a large model does not: they stop with the files missing, repeat one action until the budget ends, and report success without checking. At the one-to-nine-billion-parameter tier the harness supplies almost all of the procedural control that a large model supplies for itself.We present Mingbird, an open-source harness for that tier on a consumer graphics card: 12k lines of Python, a 628-test suite, and eighteen mechanisms for context economy, completion, time budget and the host boundary, plus a crawl guard outside those four groups. Its design point is completion control: the harness runs executable checks on the working directory and refuses a claim of completion when they fail.Five evaluations accompany it. A configuration-surface audit finds four defects that the 625-test suite of v3.1.1 missed, each because a test checked a property adjacent to the one that broke. A ten-task probe suite gives each mechanism a reason to fire and reports where the change reaches the artifact, including one mechanism that cuts the signal it protects. Disabling five of six completion-time checks costs 0.152 of the artifact score on eighteen held-out tasks, three replications per cell (95% CI [+0.064, +0.246], Holm p = 0.014). On a 288-cell matrix the harness reaches 0.886 against 0.631, 0.479 and 0.405 for three open scaffolds, and on τ 2 -bench it ranks first or tied first in all three domains against the arms evaluated.Tasks, scorer, source and every per-cell result are public.
As financial institutions increasingly interact with blockchain ecosystems through decentralized finance, stablecoins, tokenized assets, and distributed ledger infrastructure, they come to depend on systems governed wholly or partly by decentralized autonomous organizations (DAOs). This paper examines how DAO governance differs from traditional corporate governance and what this implies for risk management in regulated financial institutions. Using a qualitative comparative framework grounded in OECD governance principles, Basel Committee guidance, and FINMA circulars, the study maps DAO deliberation, voting, and accountability mechanisms against established governance and risk-management principles and illustrates associated risks through selected real-world cases. The analysis shows that DAOs preserve core governance functions but redistribute them across token holders, delegates, community forums, smart contracts, and legal wrapper entities. This creates distinct risks, including voting-power concentration, weak proposal controls, legal uncertainty, and unclear accountability, which established frameworks do not fully capture, while also offering two transferable strengths: a transparent, traceable decision record and a direct link between approval and execution. The paper contributes a structured assessment framework linking DAO governance characteristics to existing governance, outsourcing, technology-risk, and operational-resilience processes, enabling institutions to address DAO-governed dependencies through targeted extensions of current practices rather than a separate risk framework.
Trajectory prediction models for automated driving are usually trained on data with a fixed sampling frequency, while positions received via vehicle-to-everything (V2X) communication arrive at 1-10 Hz. We study whether a learned predictor that uses only the observation history keeps its advantage over constant velocity (CV) and constant acceleration (CA) models when the history frequency drops. On the Argoverse 2 motion forecasting dataset, the 5 s history of the focal agent was downsampled from 10 Hz to 5, 2, and 1 Hz, and the 6 s future was predicted without maps or neighboring agents. An LSTM-based recurrent neural network (RNN) trained for each frequency was compared with the same network trained at 10 Hz and applied to sparse histories after linear interpolation. Hypotheses and thresholds were fixed before training, the main RNNs were trained with ten seeds each, and results on 24,984 test scenarios have bootstrap confidence intervals over scenarios and seeds. The average displacement error of the frequency-specific RNN grew by 3.1% at 5 Hz, 10.1% at 2 Hz, and 20.3% at 1 Hz. Contrary to our hypothesis, its advantage over the better kinematic model did not vanish on straight motion and was 24-30% on both straight motion and turns at every frequency, while for pedestrians CV was more accurate. The 10 Hz model with interpolation was at most 1.3% less accurate than the frequency-specific model down to 2 Hz but 11.5% less accurate at 1 Hz. Each hypothesis was only partly supported.
Structure--Residual Inductive Biases for Language Model Pretraining
Hai Huang, Bartosz Żuk, Rassul Magauin, Jiaxin Huang, Xue Liu
AI 速读 基于摘要
作者提出结构-残差归纳偏置,通过边界瓶颈概念和连续概念两种方法,在语言模型预训练中改进下一词预测。
cs.CL · cs.AI
查看原文摘要
Next-token prediction is relied upon to learn both predictive behavior and internal representations. We ask whether representation structure itself should be an explicit inductive bias. Motivated by the two-part coding view of Kolmogorov complexity, we study a structure--residual principle in which a representation separates structure from realization detail. We instantiate this principle in two complementary ways. Boundary Bottleneck Concept (BBC) routes information from token segments through boundary carriers while preserving local autoregressive computation. Continuous Concept (CC) regularizes representation geometry and induces a spectral decomposition into low-frequency structure and high-frequency residual. We derive coding bounds for both views and show that multiplicity-corrected boundary attention becomes exact as the within-segment path energy vanishes. Empirically, BBC and CC exhibit strikingly different depth preferences: BBC works best in early layers, whereas CC is most effective in later layers. At 300M training tokens, combining BBC and CC improves NLL by 1.44\% on GPT-2 Medium. These gains persist with training runs up to 20B tokens. Representation diagnostics show that BBC increases the relative predictive importance of boundary states, while CC reduces spectral residual energy. Together, these results suggest that structure--residual biases can improve language-model pretraining beyond next-token prediction. Code: https://github.com/haihuang-ml/structure-residual
Ensemble prediction of nonlinear dynamics propagates a probability density whose linear transport equation retains the nonlinear drift, but grid-dependent circuit costs leave end-to-end feasibility in early fault-tolerant quantum computing unresolved. We couple polynomial reduced-order models fitted by sparse identification of nonlinear dynamics to Walsh-synthesized Koopman-von Neumann (KvN) transport and surface-code resource estimation. At fixed model dimension and drift degree, per-step transport rotation counts scale polylogarithmically with cells per reduced coordinate. Balancing model accuracy against circuit costs makes reduced-model polynomial degree a design variable for early fault-tolerant quantum simulation. For a cylinder wake approaching saturation, preparation, propagation and sampling to 1% relative root-mean-square field error against the radial model are estimated to take 1.62 h on about 560,000 physical qubits. Estimated three-coordinate wake runtimes span hours to weeks below one million qubits. IBM hardware experiments demonstrate nonlinear radial KvN transport for this wake, using reference circuits to mitigate noise.
WoRD: Diffusion-Based Posterior Inference for Online Goal Recognition
Almog Anschel, Sarah Keren
AI 速读 基于摘要
WoRD 用学习到的轨迹先验与扩散去噪链,从部分观测中推断智能体目标,并在模糊前缀下保持校准的多模态后验。
cs.AI · cs.LG · cs.RO
查看原文摘要
Goal recognition is the task of inferring an agent's intended goal from its observed behavior. Most existing work focuses on high-level symbolic recognition, abstracting away the continuous motion through which agents reach their goal. Methods that do reason over continuous trajectories rely on an explicit dynamics model that is hard to acquire and rarely transfers across settings. Neither approach explicitly learns how agents move toward each candidate goal, limiting accuracy on ambiguous trajectory prefixes where such knowledge is most needed. We present WoRD (Waypoint Recognition with Diffusion), which uses a learned trajectory prior to infer the agent's goal from partial observations. At each step, WoRD runs one warm-started denoising chain per candidate goal, steers each chain toward the observed trajectory through a closed-form log-likelihood gradient, and reweights the resulting samples by Bayes' rule into a calibrated posterior over goals. We evaluate WoRD on simulated robot settings, pedestrian trajectories, and vehicle trajectories using four metrics targeting distinct failure modes. Across all three settings, the same inference procedure preserves calibrated multimodal beliefs through ambiguous prefixes where filter- and classifier-based observers collapse to a single mode or grow overconfident.
People experiencing high-acuity mental health concerns are turning to AI chatbots for emotional support, but little is known about what drives this use, their experiences, and their preferences. In a sequential mixed-methods study, we inductively analyzed Reddit posts and used the findings to develop a survey. We surveyed 202 adults with recent mental health concerns who had used AI for emotional support. Most participants were already receiving clinical care and reported turning to AI, most commonly because they did not want to burden others, rather than being driven by barriers to care. Benefits were broadly shared, but participants reporting past-year suicidal ideation described more harm-facilitating AI behavior and reliance than those without. Overall, participants ranked honesty above warmth, and preference for crisis escalation declined as clinical acuity rose. These findings motivate AI support that addresses perceived costs of disclosure, evaluates benefits alongside harms, and widens feasible paths to human connection.
Automated segmentation is routinely used as the first step in quantitative image-based clinical workflows, where predicted masks are subsequently reviewed and corrected by human operators. Automated correction of the residual structural errors is currently addressed one anatomy at a time, so the number of trained correctors scales with the number of segmentation targets. We present GRACE, a retrieval-augmented refiner that edits an input mask under context drawn from a shared knowledge base of verified prior cases. The refiner is trained once on a multi-organ corpus spanning compact organs, cavities, branching vessels, and tumour lesions. In the held-out dataset-transfer evaluation, three held-out datasets query a fixed knowledge base built from training data only. A separate leave-one-anatomy-out experiment tests parameter-free index extension: verified cases of an unseen solid organ enter the knowledge base, and the refiner stays frozen. A light shape-aware projection head on a frozen multimodal encoder maps embedding similarity to the shape agreement of the masks, so the knowledge base returns cases that the refiner can use as a shape prior. We evaluate GRACE on 18 public datasets spanning CT, MRI and ultrasound, using nine refinement configurations and seven upstream segmentors that span weak, mid, and strong operating regimes. We report each change with the DSC of its input mask. On the weak tier, the input DSC is 0.386 and GRACE adds up to +0.35 DSC. On a strong segmentor, the input DSC is 0.932 and the mean change is small. Transfer to an imaging modality that is absent from training and from the index is partial and depends on the anatomy, and remains an open research direction. Our results suggest retrieval-conditioned refinement is a practical alternative to anatomy-specific correction models. Code: https://github.com/robustml-eurecom/grace
We prove that isomorphism of integral $2\times3\times3$ tensors under $\mathrm{GL}_2(\mathbb{Z})\times\mathrm{GL}_3(\mathbb{Z})\times\mathrm{GL}_3(\mathbb{Z})$ is decidable in bounded-error quantum polynomial time. For a tensor $T=(A,B)$ with nonzero determinant form $\det(xA-yB)$, we use the parametrization of Wood by balanced modules to reduce the isomorphism problem to principality tests for compatible lattices, and use the algorithms of Bley, Hofmann, and Johnston to solve this in BQP. When the determinant form vanishes identically, we get either an integral $2\times2\times2$ core, or a polynomial kernel vector of degree one or two. The linear-kernel case yields another compatible lattice problem, while the quadratic-kernel case reduces to a fixed system of polynomial congruences and unimodular lifting. A quantifier-elimination argument decides these congruences in deterministic polynomial time given the prime factorization of the modulus, which quantum integer factorization supplies. Finally, we show that zero padding preserves integral $2\times2\times2$ tensor isomorphism, yielding a polynomial-time reduction from real-quadratic principality to the identically singular $2\times3\times3$ problem.
We investigate how language models process negation in natural language inference (NLI). We propose the \textit{Polarity Composition Hypothesis}, a high-level causal model of this computation. Behavioral evaluations show that language-model predictions largely conform to the proposed hypothesis. Using Distributed Alignment Search, we identify causally effective subspaces aligned with the variables in the polarity-composition hypothesis. These representations transfer across negation forms and partially to a subset of MNLI. We release the dataset and code.
This paper proposes a measurement standardization framework that compresses expert-AI interactions into structured, comparable fields for prospective risk detection in deployed AI systems, without access to model internals. This concept paper defines the framework's scope, semantically and statistically, and specifies a protocol for its empirical testing. The population-level claims it is designed to support therefore belong to a staged research program rather than to results claimed here. Measurement standardization underpins three claims. The first is a reliability claim: under bounded conditions, large language models can produce reliable, standardized assessments of the evidential and policy alignment of expert-AI interactions. The second is a governance claim: alignment scores give experts an immediate signal during deployment and give institutions a basis for monitoring alignment patterns across mission types, models, and domains. The third is an outcome validation claim: once measurement standardization is established, aggregate alignment scores could be used to study associations with downstream outcomes in regulated professional settings. This introduces the possibility of an "AI epidemiology", a form of risk detection based on correlated variables instead of mechanistic analysis, inspired by epidemiological reasoning. A minimal application of the protocol to a published expert-AI corpus shows that the judge reproduces its policy and evidential alignment scores across two runs under the specified conditions. Judge reliability at scale remains to be validated in future work. The paper sets out a defined grammar of eight interaction fields, together with a statistical protocol based on paired bootstrap inference, DeLong's test for paired AUCs as a sensitivity check, a pre-specified one-sided non-inferiority margin of 0.05, and Holm-Bonferroni correction.
Decentralized Autonomous Organizations (DAOs) run protocol governance by letting token holders vote on proposals. The dominant rule, voting power proportional to wallet balance, concentrates control among a small number of large holders, fueling the token-control governance attacks that have already compromised real protocols. To counter this concentration, the community has turned to anti-plutocratic voting mechanisms such as Quadratic Voting (QV), which assign sublinear voting power per token with the goal of dampening the influence of large holders. We prove that no voting rule that derives power solely from wallet balance can succeed on a permissionless blockchain. Through a costed model of on-chain voting that captures realistic blockchain frictions -- including per-wallet splitting and voting costs, fixed setup costs, and minimum-balance requirements -- we show that whenever a wallet of any size yields nonzero voting power, a Sybil attacker who splits tokens across many wallets achieves total voting power that grows at least linearly in their token holdings. For concave rules actually proposed to dampen governance power -- those that are positive, increasing, and finite -- we show that the optimal strategy yields power that is asymptotically linear in token holdings, regardless of the cost scheme. Instantiating the model on real DAOs reveals attack costs orders of magnitude below the value at stake. Replaying the ten most recent finalized proposals of five major DAOs (ENS, Compound, Uniswap, Arbitrum, and ZKsync) under linear, quadratic, logarithmic, and power-($β= 0.25$) voting, we measure Sybil amplification factors between $1,172\times$ and $4,039\times$ under Quadratic Voting, and exceeding $229,000\times$ under steeper power rules.
We present the first complete implementation of $β$-reduction in the untyped $λ$-calculus using exclusively spiking neural dynamics. Every step of the reduction---redex detection, subterm segmentation, variable-binding detection, De~Bruijn index arithmetic, token copying, and inter-step role maintenance---is performed by leaky integrate-and-fire (LIF) neurons with explicit membrane dynamics, exponential synaptic currents, axonal delays, and pair-based spike-timing-dependent plasticity (STDP). Each neural component maps onto a specific, experimentally documented mechanism: Jeffress-type delay-line coincidence for redex detection; a line-attractor balance counter for subterm segmentation; $θ$--$γ$ cross-frequency coupling for De~Bruijn variable binding; a phase-ring traveling wave for depth-index arithmetic; theta-cycle STDP for token copying; and recurrent persistent firing for role maintenance between reduction steps. All synaptic weights are derived analytically from measured EPSP amplitudes; no hand tuning is performed. The closed-loop system achieves 100\,\% agreement with a symbolic oracle across ten validation classes, including Church numeral arithmetic (\textsc{plus}, \textsc{mult}, \textsc{pred}), Boolean logic with divergent branches, Y-combinator steps, terms with up to six bound-variable occurrences, De~Bruijn indices up to~12, noise robustness, and determinism. To our knowledge this is the first Turing-complete symbolic computation performed entirely within spike dynamics using biologically grounded circuit primitives. We discuss implications for neuromorphic proof assistants and the neural basis of symbolic reasoning.
Long-context decoding, whether over long reasoning traces, long videos, or robot episodes, is increasingly limited by the memory of the key-value (KV) cache. Page-retrieval sparse attention makes each decoding step cheaper by attending only to the top-scoring KV pages. However, it still stores every page exactly, and adds a small digest per page for scoring, so the cache ends up slightly \emph{larger} than the full cache. We propose OVAL, which stores each page as a pair of output-aware low-rank bases that act as both the retrieval index and the storage. A key basis scores the page and rebuilds its keys when the page is selected, and a PCA basis rebuilds its values. The exact pages can then be dropped, and nothing extra is added to the cache. The key idea is that the key basis should be output-aware. A low-rank error in the keys perturbs the attention output at first order, but perturbs the page mass that retrieval ranks only at second order in expectation. OVAL therefore builds the key basis from a mix of key geometry (key PCA) and the attention-output error, with a per-head mixing weight set in closed form. At the same attended budget, OVAL is best or tied-best in eight of nine long-reasoning settings. It matches FullKV on MATH500 with 66\% of its memory, is 1.25 points above the best baseline on AIME24 with 56\%, and is on par with FullKV on LongBench~v2. On \mbox{Video-MME} and LVBench, it is within 0.32 points of exact-page retrieval at the same read, while storing 53\% of the full cache instead of 103\%. Its savings also stack with quantization and token pruning: with 4-bit bases, it is within 0.25 points of FullKV at 41\% of its memory.
Say Less, Hallucinate Less? Diagnosing and Restoring Object Coverage in Preference-Optimized Vision-Language Models
Hansang Lee
AI 速读 基于摘要
作者发现偏好优化在降低物体幻觉的同时会减少物体类别覆盖,并提出无需训练的基础模型引导恢复方法。
cs.CV · cs.AI
查看原文摘要
Preference optimization (PO) is the standard way to reduce object hallucination in large vision-language models. We find that four public PO checkpoints of LLaVA-1.5-7B also lower object coverage. They name fewer object categories, and their lower hallucination rates bring no significant gain in set-level object F1 on CHAIR. We therefore ask how much of the PO gain the base model can reproduce by stopping earlier. A positive end-of-sequence (EOS) logit bias lets the base model reach the recall of each PO model, and this early stopping alone reproduces 58 to 86 percent of the CHAIRi gain of three checkpoints on two disjoint COCO sets. Writing longer does not bring the lost categories back. Under greedy decoding, a negative EOS bias recovers only 2 to 9 percent of the lost CHAIR recall of these checkpoints, and 94 to 96 percent of the added object mentions repeat categories already named. On CHAIR, the base caption of the same image still names about half of the missed object categories. We therefore propose base-guided object-category restoration, which appends high-confidence base-only categories to the PO caption without training. With thresholds fixed on a development set, it brings three checkpoints within one point of the base recall on a held-out COCO set with a lower distinct-object hallucination rate than the base model. Against the recall-matched base stopping curve, its gain holds for RLAIF-V near the base recall and for RLAIF-V and CHAIR-DPO at an intermediate recall. We will release the code of the recall-controlled evaluation.
A working-memory model must remember stimuli and use its memory flexibly to perform different tasks. Here we introduce a model of working memory for images that combines dynamic activity with temporary associations between image features. We train the model on five tasks, including remembering a sample while other images are presented, categorizing distractor images while remembering the sample, and determining whether an image is new or has been presented before. The same trained model performs all tasks, guided by explicit cues about which response is required. We use this flexible memory model to ask how delay, distraction, and image history affect performance. Delays have little effect on sample-memory performance for the delays we tested, while seeing more images leads to more false alarms even when the target image appeared equally recently in all cases. On the sample-memory-plus-categorization task, the probability of correct responses on all questions decreases as more questions are asked, despite robust sample-memory performance. Finally, the model maintains stable performance across 40 brief trials of each task under the tested conditions. Together, these results demonstrate how a single working-memory system can be flexibly used to perform different tasks, and how its errors depend on both previous images and task demands.
Identifying the source of an epidemic, i.e., the patient zero, in a temporal contact network is important for effective outbreak containment and resource allocation. While several methods have been proposed for this task, they have rarely been evaluated under comparable conditions, and no systematic comparison exists for temporal contact networks. We address this gap with a comparative study of seven major methods, namely Steiner Tree (ST), Individual-Based Approximation (IBA), Dynamic Message Passing (DMP), Monte Carlo Mean-Field (MCMF), Soft Margin Estimator (SME), Backtracking Network (BN, a graph neural network), and Belief Propagation (BP). We unify all methods under a single problem formulation and evaluate them on eight diverse temporal networks. In the main experiment, we compare source detection performance for varying basic reproduction numbers. Three complementary experiments on a subset of networks study (i) learning curves, (ii) partial observability, and (iii) parameter misspecification. We find that BP and BN are the strongest source detectors overall. SME is strong under partial observation but mediocre otherwise. IBA, DMP, and MCMF share a common weakness rooted in their factorized-likelihood assumption, leading to overconfident predictions. We identify scaling these methods to larger networks as the most pressing direction for future work.
Empirical research is changing rapidly as AI coding agents are becoming increasingly proficient in every step of it, from writing code to manipulating data. We argue that while the potential of coding agents for science is vast, their use in hypothesis testing poses a substantial threat to classical inference, as these agents enable a supercharged form of selective inference. In particular, we argue that coding agents carrying out empirical research introduce a new type of post-selection inference that combines human-like, open-ended search over an implicit space of estimators with programmatic replayability of both the selection and the estimation process. Indeed, we find that on simulated data with zero effects, four frontier agents report significant p-values between 54% and 98% of the time when asked to produce a compelling finding. We propose a simple procedure for detecting and correcting these distortions. After an agent has completed its analysis, we reset and replay it on subsamples of the data to estimate the agent's relative probability of selecting that analysis when it looks statistically favorable. For agents with stable selection behavior, the resulting corrected test is asymptotically valid and consistent conditional on selection. Importantly, the correction is learned from the black-box agent's actual choices rather than a model of how it searches, so it remains valid even as AI technology continues to evolve rapidly. In simulations, the correction robustly lowers the high false rejection rates to near the nominal 5%. We illustrate the method on applications from two published papers.
EmbeWebAgent: Embedding Web Agents into Custom Enterprise UIs
Chenyang Ma, Clyde Fare, Matthew Wilson, Dave Braines
AI 速读 基于摘要
提出一种把智能体直接嵌入企业自有前端界面的框架,借助轻量前端钩子与可复用后端工作流,实现比截图或原始 DOM 观察更稳健、动作表达更丰富的 Web 自动化。
cs.AI · cs.SE
查看原文摘要
Most web agents operate at the human interface level, observing screenshots or raw DOM trees without application-level access, which limits robustness and action expressiveness. In enterprise settings, however, explicit control of both the frontend and backend is available. We present EmbeWebAgent, a framework for embedding agents directly into existing UIs using lightweight frontend hooks (curated ARIA and URL-based observations, and a per-page function registry exposed via a WebSocket) and a reusable backend workflow that performs reasoning and takes actions. EmbeWebAgent is stack-agnostic (e.g., React or Angular), supports mixed-granularity actions ranging from GUI primitives to higher-level composites, and orchestrates navigation, manipulation, and domain-specific analytics via MCP tools. Our demo shows minimal retrofitting effort and multi-step behaviors grounded in a live UI setting. Demo Video: https://youtu.be/Cy06Ljee1JQ
In-context learning has recently been linked to implicit gradient descent in linear self-attention models, suggesting that context can induce a forward-pass update. Retrieval-augmented generation (RAG) also relies on context, but retrieved documents are usually treated as static evidence rather than signals for adaptation. We study RAG as an in-context optimization process. First, we show that one linear self-attention layer can implement one gradient-descent step on a unified linearized RAG objective covering both projection-based and dot-product retrieval interfaces. This gives an exact regime where retrieval-augmented prediction and in-context optimization coincide. We use this result not as a literal model of LLM computation, but as a guide for adapting the interaction between queries and retrieved evidence. We then test the boundary of this correspondence: it remains stable under controlled linear extensions, but becomes feature-distribution dependent under nonlinear architectures. Finally, we turn this view into a lightweight method for frozen RAG LLMs. The method keeps the retriever and backbone fixed, and predicts a context-conditioned update to a generator-side evidence-use interface. Across seven QA benchmarks, two retrievers, and two frozen LLM backbones, this forward-only update improves a shared-interface baseline, transfers to held-out tasks, and approaches test-time gradient adaptation at much lower per-query cost.
We consider estimating the one-step-ahead conditional distribution of a multivariate stochastic process. Many existing approaches rely on structural assumptions such as finite-range memory, sparsity, or additivity, which can be poorly suited to processes with long-range nonlinear interactions. However, without such structural assumptions, nonparametric estimation is known to be challenging due to the curse of dimensionality. To address this challenge, we introduce a new estimator based on the idea of learning representations of the \emph{predictive states} of a stochastic process. We theoretically analyze several aspects of this problem. As a bonus, we establish convergence rate guarantees for an instantiation of our method based on deep neural network estimators. Our results suggest that faster convergence rates are achievable when the predictive state space is lower-dimensional. That is, when the past history can be compressed into a lower-dimensional statistic that is sufficient for predicting the future. We implement our method in PyTorch and evaluate its performance on synthetic data and real \emph{C. elegans} data.
Retrieval is increasingly moving from one-shot matching toward interactive reasoning, where language agents iteratively inspect evidence, reformulate queries, and search again. Training such agents raises a credit-assignment challenge: executable actions such as queries or summaries can be directly evaluated by the retriever, while latent reasoning steps are not directly observable and only affect future executable actions. This asymmetry makes outcome-level reward assignment unreliable, as the same final reward may credit reasoning steps that did not actually shape retrieval success. We propose RICE-PO, a critic-free policy optimization framework that converts retrieval interactions into localized learning signals. RICE-PO selects high-uncertainty executable actions as anchors, evaluates local counterfactual branches using retrieval metrics, and propagates credit to latent reasoning steps only when reasoning-to-action influence is strong and future residual effects are stable. On BRIGHT and BEIR, RICE-PO consistently outperforms prompt-based agents and group-based RL baselines under the same retriever setting. These results show that the structure of agent-environment interaction itself can provide useful supervision for training reasoning-based retrieval agents.
Networks that predict angles, gaze directions, and rotations do not output these quantities directly: they emit raw coordinates that a fixed decoder, such as normalization, maps to the final prediction. Attribution methods are usually applied to the raw coordinates, so they can credit inputs for changes the decoder discards, such as rescaling a vector when only its direction matters. The resulting explanation can highlight inputs that have no effect on the prediction. We instead attribute through the decoder differential, which for normalization decoders reduces to a scaled tangent projection. We characterize invariance to output rescaling, show that every fixed smooth scalar target on a sphere has points of zero tangential sensitivity, and show that an endpoint-projected variant of Integrated Gradients can still attribute to scale-only inputs, whereas direct angular Integrated Gradients cannot. On synthetic tasks with analytically known relevance, ambient attribution assigns up to 72% of its mass to an irrelevant region. On MPIIGaze validation data, tangent Grad-CAM produces 44% greater prediction displacement under 10% pixel deletion. On PASCAL3D+ with a 6D rotation decoder, the full differential improves mean deletion displacement by 14.1 degrees over the ambient coordinate sum and by 13.8 degrees over a decoder-invariant scalar. Results for a two-polarity tangent LayerCAM variant are mixed: it exceeds ambient-sum LayerCAM on circular regression but no difference is detected on quaternion pose. These results separate two failure sources, radial contamination and scalarization loss. The approach uses existing attribution methods with decoder-aware targets, although some variants also require changes to attribution aggregation. An invariant attribution can still miss inputs that drive the prediction, so invariance and deletion performance should each be checked.
MAGIC-SSCIL: Manifold Anchoring and Geometric Incremental Calibration for Semi-Supervised Class Incremental Learning
Yousef Abdi, Mohammad Asadpour, Yousef Seyfari
AI 速读 基于摘要
MAGIC 通过在图传播伪标签并加权类别统计,提升无样本半监督类增量学习在极少标签下的性能。
cs.CV · cs.LG
查看原文摘要
Exemplar-free semi-supervised class-incremental learning (SSCIL) asks a model to learn new classes from a few labeled and many unlabeled images without storing past data. A common recipe replays old classes from stored Gaussian statistics and pseudo-labels unlabeled data with a confidence threshold. With one or two labels per class, however, the model's confidence is a poor guide to which pseudo-labels are right, so the statistics that replay depends on rest on few or noisy samples. We propose MAGIC (Manifold Anchoring and Geometric Incremental Calibration), which propagates labels over a k-nearest-neighbor graph of frozen features and uses the propagated confidences as weights in the unlabeled loss and, through Soft-Weighted Geometry Calibration (SWGC), in the class statistics from which phantom features replay old classes. With a frozen ResNet-18 on CIFAR-100, CUB-200, and ImageNet-R at 1%, 5%, and 10% labels, we compare MAGIC with nine supervised class-incremental methods equipped with FixMatch and two native SSCIL methods. MAGIC gains from unlabeled data in all nine settings, most at 1%, and leads on fine-grained CUB-200 at every ratio; at 1% it is level with the best on ImageNet-R and 2.2 points behind the native SSCIL method TACLE on CIFAR-100. Experiments inside MAGIC show that most of this gain flows through the weighted statistics, not the loss, that at 1% labels graph pseudo-labels are more accurate than the model's own at equal coverage, and that a fixed threshold of 0.95 removes most of the gain. At higher ratios, baselines that adapt the backbone or model full covariances overtake MAGIC on CIFAR-100 and ImageNet-R.
In the online stochastic matching problem on a bipartite graph, offline vertices are given in advance, and online requests arrive sequentially from known distributions, each matched immediately and irrevocably to a free neighbor or rejected. Computational hardness results of the optimal online strategy have been known only for instances where online vertices have a fixed arrival order [Papadimitriou et al., 2024]. For random-order arrivals or even i.i.d. arrivals, it has been long open since the problem was proposed [Feldman et al., 2009]. In this paper, we give the first computational hardness results for online stochastic matching with random-order arrivals. In particular, we prove that for known i.i.d. arrivals, computing the online optimum is APX-hard, ruling out a Polynomial-Time Approximation Scheme unless P = NP, even in unweighted graphs with two offline neighbors per online vertex. For random-order arrivals, even when there is only one offline node (i.e., the prophet secretary setting), computing the optimal online policy is #P-hard, resolving an open problem since [Dütting et al., 2023].
Recent advances in text-to-image generation have substantially improved image quality, particularly through diffusion-based methods that require multiple inference steps. In contrast, Sphere Encoder offers an efficient generative framework capable of producing images with fewer forward steps. However, extending it to free-form text is nontrivial, because its class-conditioned design lacks cross-modal alignment. To address this challenge, we propose TextSphere, an efficient text-to-image generative framework that extends Sphere Encoder with language conditioning and cross-modal alignment. We incorporate pretrained CLIP text embeddings through adaptive layer normalization (AdaLN) to guide image generation while retaining the original spherical sampling mechanism. To bridge the cross-modal representation gap, we introduce a Spherical Semantic Alignment and Uniformity (SSAU) objective that jointly promotes semantic alignment and representation uniformity on the hypersphere. Experiments on CUB-200-2011, MS-COCO-2014, and Flickr30k demonstrate that TextSphere consistently reduces generation FID (gFID) over the text-conditioned baseline while achieving reconstruction FID (rFID) below 1.0. With only 4 and 6 sampling steps, TextSphere achieves lower gFID than the compared diffusion- and GAN-based models trained from scratch on CUB-200-2011 and MS-COCO-2014, respectively. We further observe that improved reconstruction fidelity does not necessarily yield better generation quality. In addition, we use Partial Least Squares Singular Value Decomposition (PLS-SVD) to analyze cross-modal alignment and modality gap, offering insight into the effectiveness of our approach.
Inference-time scaling improves the performance of large language models by sampling multiple responses and evaluating their collective utility. However, many existing approaches obtain these samples by repeatedly sampling from a single policy. In this work, we study if the sample budget at inference time can instead be allocated across a population of policies. We introduce stratified policies, where each of $k$ learned policies contributes one response to a $k$-sample inference-time objective. We realize this policy population using separate LoRA adapters on a shared language model backbone and optimize it for pass@$k$. We theoretically motivate the approach by analyzing the linear setting; a population of LoRAs can achieve zero best-of-$k$ approximation error in settings where every single LoRA has strictly positive prediction error. Empirically, we demonstrate that stratified policies consistently improve pass@$k$ over a single-policy baseline across multiple reasoning tasks. Overall, our results highlight how distributing samples across complementary policies can be a more effective use of sample budget than repeated i.i.d. sampling from a single policy at inference time.
Video question answering under tight frame budgets requires preserving localized motion and event order, yet current video multimodal large language models (MLLMs) reduce comprehension to static multi-image reasoning by discarding unselected frames and encoding keyframes independently: (1) Spatially, whole-frame pooling dilutes small moving targets into static backgrounds; (2) Temporally, dropping intermediate frames or averaging tokens erases inter-frame motion, confusing direction, speed, and order. We propose VIVID (Visual Inter-frame Variations and Individual Dynamics), which moves beyond static snapshots like human vision: keyframes capture core states while inter-frame Patch Diffs preserve neighboring changes within $K_f$ keyframes. Autonomous Delta-Driven Keyframe Planning (ADKP) selects anchors from a $3\text{ fps}$ pool and feeds surrounding patch differences into Dual-Road Change Memory (DRCM), forming Dynamic Tokens at zero extra anchor tokens alongside a camera-stabilized Active Motion Region of Interest (ROI) and chronological Entity State Chains. Adding $0.22\%$ trainable parameters to a frozen $27\text{B}$ backbone, VIVID-Small ($50$ examples, $\sim 1.5$ GPU hours) beats all eleven baselines in overall accuracy across $8,083$ questions and trails only VIVID-Large ($3,600$ examples), which leads on MotionBench, Video-MME, and TempCompass.
Answer continuation often returns a diagnosis already stated in unfinished clinical reasoning. We study this behavior in MamaReason, adapted from dense Gemma 4 31B, on MamaBench's 217 paired maternal and child-health presentations. An explicit-commitment extractor identifies a diagnosis in 75 of 94 failed RL first passes and 75 of 93 failed SFT first passes. The continuation matches that diagnosis in 67/75 RL cases and 58/75 SFT cases, using diagnosis strings alone. Reading these commitments without additional generation recovers 30 of the 40 RL pairs restored by continuation under the archived correctness labels. Recurrence controls distinguish routine mentions from repeated answer selection. An offline second-commitment trigger measures the tradeoff between earlier readout and paired accuracy. The resulting decomposition separates diagnostic content already present in a trace, completion of an accepted answer, and paired diagnostic correctness. For clinical language-model evaluation, these are distinct outcomes that become visible when the same reasoning trace is examined through multiple readouts.