arXiv subjects

Wei Liu

Publications and source records attributed to Wei Liu.

At least 19 recordsLinked to original sources

arXiv2026

FOCUS: A Subspecialty Foundation Model for Intraoperative Central Nervous System Pathology

Intraoperative frozen-section diagnosis of central nervous system (CNS) lesions guides surgical decisions but is challenged by artefacts, diagnostic complexity and time constraints. Most pathology foundation models have limited exposure to CNS frozen tissue. Here we present FOCUS, a subspecialty foundation model pretrained on 27,930 frozen-section whole-slide images enriched for CNS tissue. FOCUS ranked first across all 11 diagnostic and 12 biomarker-prediction tasks in the internal test cohort, achieving mean AUCs of 0.969 and 0.877, respectively. This advantage was retained in the temporal and external test cohorts. We further organized a WHO-aligned diagnostic hierarchy, where FOCUS also outperformed comparator models. In a reader study, FOCUS assistance enabled junior readers to approach or exceed unassisted senior readers in hierarchical diagnostic accuracy and granularity. More importantly, FOCUS assistance enabled readers to correctly identify a mean of 63.0% of initially missed lymphoma cases (73.2% among junior readers), potentially avoiding unnecessary extensive resection. These findings highlight the potential of FOCUS to provide clinically actionable support for intraoperative CNS pathology and inform surgical decisions.

cs.CV↗

arXiv2026

Deep Bayesian Approaches for Interpretable Nonlinear Structural System Identification

System identification, namely the inference of mathematical representations of dynamical systems from measured data, is fundamental to engineering diagnostics, prognostics, and decision-making. Bayesian filtering offers a principled framework for probabilistic state estimation but traditionally relies on prescribed transition and observation models. Recent advances in deep Bayesian filtering address this limitation by embedding learnable representations within probabilistic state-space formulations, enabling data-driven identification of unknown dynamics while retaining physical structure and uncertainty-aware inference. This tutorial provides a systematic introduction to deep Bayesian filtering for nonlinear structural system identification, focusing on the interplay between learned dynamics, physical inductive biases, and posterior inference. The Deep Markov Model serves as a data-driven reference, while the Physics-guided Deep Markov Model (PgDMM) and Neural Extended Kalman Filter (Neural EKF) illustrate complementary approaches based on amortized variational inference and differentiable Kalman filtering, respectively. Their capabilities are examined through dynamic state estimation and joint dynamic state and parameter estimation on a three-dimensional shear-frame benchmark with hysteretic Bouc-Wen connections and sparse acceleration measurements. The results highlight important distinctions between measurement-conditioned reconstruction and open-loop forecasting, demonstrate how physical structure can improve latent-state interpretability without necessarily ensuring predictive fidelity, and reveal differences in the robustness of response and parameter estimation. Practical guidelines for model construction, training, evaluation, and deployment are provided, together with an outlook on remaining challenges in uncertainty calibration, identifiability, and real-world validation.

cs.LG↗

arXiv2026

AdaGround: Token-Normalized Adaptive Zooming with Verifier-in-the-Loop Refinement for GUI Grounding

GUI grounding on high-resolution screens fails mostly on small targets, and zooming in is a common training-free remedy. However, most existing zooming pipelines do not match the zoom to the target's size, accept the zoomed answer unconditionally and never revisit a wrong answer. To address these issues, we propose AdaGround, a training-free framework with three components: (1) Token-Normalized Adaptive Zooming, which sets the magnification, field of view and zoom depth of each sample from the model's own estimate of the target size; (2) Verifier-in-the-Loop Refinement, which uses one marked-click Yes/No verifier to keep or reject each zoom, choose between zoom chains, trigger re-grounding and, with an existence query, decide when to abstain; and (3) Compositional Structure Grounding, which composes spreadsheet cells from their headers. On Qwen3-VL models, AdaGround surpasses state-of-the-art training-free zooming methods and many trained GUI models, improving accuracy over the backbone by 7.4 to 20.6 points on OSWorld-G and ScreenSpot-Pro, and its gains carry over to held-out UI-Vision. Code is available at https://github.com/bcwang-sjtu/AdaGround

cs.CV↗

arXiv2026

GUI-JEPA: Candidate-Conditioned JEPA Latents as a Fast Action Policy for Mobile GUI Agents

Mobile GUI agents are typically fine-tuned 7B-scale vision-language models (VLMs) that generate each action as text. However, these VLMs decode every action token by token, and even agents that score candidate actions keep a large language model or VLM in the loop, making them slow and memory-hungry. To address these issues, we propose GUI-JEPA, to our knowledge the first decoder-free GUI agent policy built on a Joint-Embedding Predictive Architecture (JEPA), which selects actions in the latent space of a JEPA encoder pretrained on unlabelled GUI screenshots. Its action- and region-conditioned latent predictor turns the screen latent into one latent per candidate action, letting lightweight heads score all candidates against the instruction in a single batched pass, without decoding a token. We further show that the scorer relies on this latent pathway and that the predictor is trained for the decision itself: a randomly initialised predictor does just as well as one pretrained on screen dynamics. On AndroidControl, GUI-JEPA is 2.8-4.0x faster on the same GPU than a Qwen2.5-VL-7B agent we fine-tune with the same inputs, while reaching 86-92% of its in-distribution accuracy with 6% of its parameters. Code is available at https://github.com/bcwang-sjtu/GUI-JEPA

cs.AI↗

arXiv2026

AzeroS: Extending LLM to Speech with Self-Generated Instruction-Free Tuning

Extending large language models (LLMs) to the speech domain has recently gained significant attention. A typical approach connects a pretrained LLM with an audio encoder through a projection module and trains the resulting model on large-scale, task-specific instruction-tuning datasets. However, curating such instruction-tuning data for specific requirements is time-consuming, and models trained in this manner often generalize poorly to unseen tasks. In this work, we first formulate that the strongest generalization of a speech-LLM is achieved when it is trained with Self-Generated Instruction-Free Tuning (SIFT), in which supervision signals are generated by a frozen LLM using textual representations of speech as input. Our proposed SIFT paradigm eliminates the need for collecting task-specific question-answer pairs and yields the theoretically best generalization to unseen tasks. Building upon this paradigm, we introduce AZeroS (Auden Zero-instruction-tuned Speech-LLM), which is trained on speech-text pairs derived from publicly available corpora, including approximately 25,000 hours of speech with ASR transcripts and 3,000 hours of speech with paralinguistic labels. Built upon Qwen2.5-7B-Instruct, the model updates only two lightweight projection modules (23.8 million parameters each), while keeping both the LLM and audio encoders frozen. Despite the minimal training cost and modest data scale, AZeroS achieves state-of-the-art performance on both semantic and paralinguistic benchmarks, including VoiceBench, AIR-Bench Foundation (Speech), and AIR-Bench Chat (Speech).

cs.CL↗

arXiv2026

Optimal Linear Filtering of Markov Jump Systems

This paper deals with the optimal state estimation problem for discrete-time Markov jump linear systems in the sense of linear minimum mean-square error where the noises corrupting the system are mode-dependent and need not be Gaussian. An orthogonal projection extension-based equation (OPEBE) is established, which can decompose a random vector into several parts where the random vector contains conditional expectation of another random vector. Using the OPEBE and some results obtained in this paper, an optimal linear filter for the system under consideration is proposed. The computational and storage load of the proposed filter does not increase with time. Also, the proposed filter is finite-dimensionally computable and has a recursive structure. Computer simulations are provided to verify the effectiveness and performance of the proposed optimal linear filter.

eess.SY↗

arXiv2026

Retrospective Comparison of Radiation Oncology Patient Message Responses Generated by Epic and Institutionally Managed LLMs

We retrospectively compared Epic-generated draft replies with institutionally managed hybrid GPT-4.1 and Gemini 2.5 Pro responses to 51 selected single-turn radiation oncology messages. Six clinicians, blinded to response source, rated overall quality on a five-point scale and perceived editing effort on a four-point scale. Mixed-effects models accounted for repeated ratings by message and clinician. Adjusted overall-quality means were 2.95 for Epic, 4.27 for hybrid GPT-4.1, and 4.39 for hybrid Gemini 2.5 Pro. Corresponding editing-effort means were 2.38, 1.46, and 1.41, with lower scores indicating less perceived effort. Both hybrid models received more favorable ratings than Epic for both outcomes (all Holm-adjusted P < 0.001), without a statistically significant difference between the hybrid models. These findings support prospective evaluation of clinician-supervised draft generation, though editing time and patient outcomes were not measured, and clinical safety was not established.

physics.med-ph↗

arXiv2026

Physical AI for Intralogistics: Benchmarking the NVIDIA Sim-to-Real Ecosystem for Cosmos-Enhanced COMPASS Navigation on Industrial Mobile Robots

Classical robot navigation stacks are reliable but often require re-engineering for each new robot type and site. Learning-based navigation shifts this effort from hand-tuned parameters to data generation and training, where a policy adapts to each new embodiment or facility by fine-tuning in simulation. Whether policies adapted this way stay reliable on real industrial robots has seen little systematic testing. We study this for NVIDIA's X-Mobility and COMPASS workflow for training learned navigation policies on Idealworks robots. In simulation, we extend the COMPASS benchmark with two embodiments, the differential-drive iw$\mkern-1.5mu.\mkern-1.5mu$hub and an omnidirectional Humanoid Mobile Platform (HMP), and with a digital twin of the Idealworks R&D center. For the real world, we fine-tune X-Mobility on data from this digital twin, with and without sim-to-real variants generated by NVIDIA Cosmos, train COMPASS on top of it from scratch in the digital twin, deploy the resulting policies on the iw$\mkern-1.5mu.\mkern-1.5mu$hub's onboard Jetson AGX Xavier, and benchmark them in the Idealworks R&D center against COMPASS trained only on NVIDIA's environments and the existing Idealworks navigation stack on task success, efficiency, safety, smoothness and onboard compute. Running every static scenario both in a digital-twin replica and in the real facility measures the sim-to-real gap and how far Cosmos-augmented data closes it. Training in the digital twin with Cosmos sim-to-real augmentation raises the real-world success rate of learned navigation from 65.3% to 91.7% across four scenarios and nearly halves its weighted travel time compared with simulation-only training, bringing it within 8.3 points of the production navigation stack, fully onboard; with an integrated global planner, it also reaches far, occluded goals in every trial.

cs.RO↗

arXiv2026

EnvLunch: The Environment as a Free Lunch for Learning to Interact

For agentic data analysis, language-model agents often require a large number of annotated tasks to learn how to interact effectively with executable environments (i.e., databases). While high-quality annotated tasks require heavy manual effort, an executable environment already provides rich interaction structure and verifiable outcomes. In this sense, the environment itself offers a free lunch, an additional set of verifiable tasks that require no human annotation. We propose EnvLunch, a warm-up stage that turns this environmental signal into agent training without human-written or model-generated task descriptions, before downstream agentic reinforcement learning. EnvLunch samples valid action trajectories, executes them to obtain final observations, and trains an agent to recover a functionally equivalent trajectory given only its final observation. This label-free warm-up enables the model to outperform the baseline by over 9 points on average across all downstream training budgets, and with few labels, the baseline needs 8 to 32 times as many annotated examples to match the accuracy that the EnvLunch model reaches at its smallest training budget. Letting the environment alone define the task yields a larger gain than having the model write its own tasks over the same environments, and this warm-up also improves fine-tuning on benchmarks beyond data analysis. EnvLunch therefore saves annotated tasks and provides a better starting point for fine-tuning agentic models when such tasks are limited.

cs.AI↗

arXiv2026

Intervention-Certified Amplification: A Validity Gate for $σ$-Weighted Post-Training

Hybrid post-training often mixes supervised fine-tuning (SFT) and reinforcement learning (RL), using training-side signals to set each sample's balance. Such signals characterize the anchor data and policy, but not whether the emphasized behavior helps in the target task and evaluation setting. This mismatch can reactivate and amplify pretrained behavior even after its literal form is removed from supervision. We demonstrate this with DIAA (Dual-Implicit Adversarial Amplification), a per-sample $σ$-weighted SFT/RL hybrid. With all literal Python fences removed from its anchor, DIAA emits unexecuted Tool-Integrated Reasoning (TIR) on $29%$ of MATH-500 prompts, three times vanilla GRPO. Paired interventions show $10.6$ percentage points of accuracy recovery when TIR is suppressed and comparable contributions from SFT imitation and $σ$-amplification; a student trained on a $0%$-TIR teacher anchor emits it at $45.8%$. Intervention-Certified Amplification (ICA) gates $σ$ with paired-rollout target evidence. Frozen and online ICA reduce TIR to $13.8%$ and $11.2%$, halve residual drag, improve accuracy from $47.5%$ to $48.7%$ and $49.3%$, and transfer to held-out GSM8K without retuning.

cs.LG↗

arXiv2026

RadOnc-Agent: An Integrated Framework for AI Workflows Across the Radiotherapy Care Pathway

Artificial intelligence has advanced radiotherapy tasks, but these capabilities remain fragmented across the care pathway. Here we present RadOnc-Agent, an integrated framework connecting 17 phase-specific functions and 9 cross-cutting utilities across four phases through a conversational interface. A language-model controller maps clinical intent to structured calls, maintains longitudinal context and validates dispatch. We evaluated request-expression diversity, function coverage, cross-stage composition, information transfer and failure recovery using 2,600 single-function requests (7,800 executions), 200 cross-stage scenarios (600 executions), and 120 workflows from 60 de-identified patient records (360 executions). RadOnc-Agent selected the intended function in 98.79% of single-function executions and completed 96.50% of scripted and 96.67% of real-patient workflows. Removing longitudinal state reduced cross-stage completion from 96.50% to 84.00%; bypassing schema and identity validation increased mismatched dispatch from 0% to 95.28% in replay/test execution. Single-function performance did not by itself predict workflow-level capability. Information continuity and cross-stage dependencies emerged as central constraints. These results establish the technical feasibility of longitudinal integration across radiotherapy workflows and provide a foundation for prospective clinical evaluation.

cs.AI↗

arXiv2026

Group Contrastive Policy Optimization: Learning Beyond the Self-Exploration Limit with Golden Answers

Reinforcement learning with verifiable rewards (RLVR) has become a powerful paradigm for improving the reasoning abilities of large language models, yet group-based methods lose their learning signal when every sampled response receives zero reward. We propose Group Contrastive Policy Optimization (GCPO), a hybrid objective that preserves standard on-policy optimization for informative rollout groups and activates a proximal contrastive objective only when exploration completely fails. The auxiliary objective pairs an offline golden reasoning trajectory with the current policy's failed responses, thereby providing a reliable positive reference and prompt-specific hard negatives without treating the golden trajectory as an on-policy rollout. Its old-policy token probabilities are evaluated through teacher forcing and used solely to constrain the update, rather than as an importance-sampling correction for the teacher distribution. A first-order analysis shows that the resulting update increases a length-normalized likelihood margin between reliable reasoning and observed failures. Experiments across two model scales and 10 core mathematical reasoning benchmarks demonstrate that GCPO improves training efficiency and final reasoning performance over competitive supervised and RLVR baselines.

cs.AI↗

arXiv2026

PROACT-Agent: Progressive Runtime Oversight and Active Circuit-breaking for Real-Time Safety

The transition from Large Language Models (LLMs) to agents shifts safety stakes from toxic text to irreversible environmental harm. While current defenses remain largely retrospective, proactive runtime intervention is bottlenecked by the lack of large-scale, causally-consistent data. We propose PROACT-Agent, a framework for synthesizing high-fidelity trajectories to enable real-time guardrails. We identify a critical "safety drift" in prior benchmarks, where lenient annotation paradigms fail to enforce temporal consistency. PROACT-Agent addresses this through: (1) Progressive Trajectory Unrolling to reveal risks hidden in long-context interactions; (2) Reasoning-Augmented Causal Rectification to enforce monotonic causal consistency; and (3) Culturally-Aware Data Localization for cross-border robustness. We introduce PROACT-Bench, a bilingual safety benchmark with 155,780 states labeled through multi-model adjudication. Evaluating updated context before the next LLM inference, the trained guard achieves 91.46% unsafe-class F1 and 90.63% exact-boundary detection under complete source holdout. In AgentDojo, it reduces non-DoS targeted attack success from 20.82% to 0.40%.

cs.LG↗

arXiv2026

PortraitAes: Intent-Conditioned Structured Portrait Aesthetic Assessment

Portrait aesthetic assessment assigns comparable scores according to how effectively human-centered images fulfill their photographic intent. These scores support data filtering, candidate selection, and preference modeling in image-generation pipelines. Existing methods typically predict a single aesthetic score or use general-purpose MLLMs without conditioning on photographic intent. This omission matters because the same blur, pose, lighting, or framing choice may serve one photographic intent but undermine another. These models thus learn context-agnostic aesthetic priors and yield inconsistent, inaccurate, misleading judgments for portraits with distinct photographic objectives. We introduce PortraitAes-Bench, an 11K-scale benchmark that decomposes this task into intent-conditioned subjudgments. Expert-authored rubrics define nine photographic intents, six first-level dimensions, and 22 secondary criteria. They support a structured pipeline for intent routing, specialist assessment, verification, and score fusion. Following this structure, we train PortraitAes with multi-task supervision. We then improve score comparability through Gaussian score calibration and within-dimension cross-image ranking. On the standard benchmark, PortraitAes achieves a Pearson correlation of 0.924 and a Spearman rank correlation of 0.934. On the hard-case set, its Pearson correlation is 0.829 and its Spearman rank correlation is 0.795. Across both sets, PortraitAes outperforms the evaluated general-purpose MLLMs and specialized aesthetic baselines.

cs.CV↗

arXiv2026

IQuest-Q1: Advancing Foundation Capabilities for Agentic CLI Systems

We present IQuest-Q1, a sparse Mixture-of-Experts (MoE) agentic foundation model with 320B total parameters (15B activated). IQuest-Q1 is developed through a three-stage process: (1) stable and efficient pre-training, combining Stable SplitNorm, Muon, and muP with a two-stage curriculum whose second stage increases the coding-data mixture, preserving a wide stability margin between safe and optimal learning rates; (2) fine-grained mid-training, expanding the data mixture across domains and formats and progressively extending the context; and (3) multi-scaffold with multi-version post-training, combining SFT, domain-specialized RL, and MOPD across various scaffolds to integrate complementary training signals and strengthen agentic coding capabilities. Furthermore, model merging combines different experts and checkpoints from different stages while mitigating conflicts among their parameter updates. The resulting IQuest-Q1 demonstrates capabilities in both coding and general tasks. We release IQuest-Q1 as an open-source model to support community research and development.

cs.CL↗

arXiv2026

Explicit quasi-periodic solutions to the discrete derivative nonlinear Schrödinger equation

Quasi-periodic solutions to the discrete derivative nonlinear Schrödinger equation are investigated. A Lax pair of the discrete integrable equation is derived, by which we obtain a nonlinear integrable symplectic map. Resorting to this map, the Riemann theta function expressions for discrete potentials are calculated. Consequently, we get the explicit solutions for the discrete derivative nonlinear Schrödinger equation, whose quasi-periodic properties are illustrated through the cases of genus one and genus two.

nlin.SI↗

arXiv2026

Making AI-Assisted Claims Independently Challengeable: Publication Authority and a Protocol for Falsifiable Publication Records

AI-assisted claims can appear authoritative when evidence, analysis, human authorization, presentation, and correction history describe different states. We develop Publication Authority: an exact-state, non-transferable, single-use publication capability within PAC-2026 (Publication-Accountability Calculus), an AIJIM Protocol candidate. The evaluated specification version (SF-4) defines six non-compensating obligations: evidence, runs and artifacts, measurement disclosure, authorization, surface correspondence, and lifecycle continuity. Each yields a target-bound witness, localized counterexample, or localized unverifiability. Only a fresh, complete all-pass record derives a permit consumed by one atomic publication transition. Ten models explored 110,764 distinct safe reachable states; 76 unsafe configurations yielded the expected violation or observer countermodel. A surface passing its correspondence check cannot authorize publication unless admitted by the accepted record. SF-4 separates evidence horizon from verification time and rejects authentic but causally invalid authorization. An in-house, instance-blind study (G7-B) matched 183 scored expectations after shared integration. Three author-led native comparisons identify gaps in a constructed in-toto/OPA profile: measurement-state admission, proof-expiry evaluation, and required-run identity. Repairs restore parity on the tested cases. An exploratory extension constructs exact claim-edit regions under existing e-closure semantics: a minimum retained core qualifies every permitted selection. For the full hypothesis universe (U=H), a quadratic frontier computes all strongest-core sizes. Enumeration and native controls distinguish reusable statistical qualification from the fresh publication authority each candidate requires. Results yield transferable protection rules; independent conformance and field evaluation remain open.

cs.AI↗

arXiv2026

Tamed Euler Approximation of Invariant Measures for Weakly Interacting Particle Systems and McKean--Vlasov Equations

We analyze explicit tamed Euler approximations for weakly interacting particle systems and McKean--Vlasov diffusions with additive noise and polynomially weighted locally Hölder, possibly superlinear drift.Under dissipativity at infinity and suitable stability assumptions, we establish time-uniform $W_1$ bounds by reflection coupling, separating relaxation, time-discretization, and finite-particle errors.For small step sizes, we prove existence and uniqueness of numerical invariant laws and bound their error relative to the continuous-time particle invariant law uniformly in $N$ in a normalized joint $W_1$ distance. We further bound the error of the one-particle numerical invariant marginal relative to the McKean--Vlasov invariant law by $O(θ^{\min\{α,β/2\}}+N^{-1/2})$, where $θ$ is the step size, $N$ the particle number, $α\in(0,1/2]$ the taming exponent, and $β\in(0,1]$ the Hölder exponent. Analytical examples illustrate the framework for smooth and non-Lipschitz drifts.

math.PR↗