Searcharxiv⌕ Search

arXiv subjects

Zhihui Li

Publications and source records attributed to Zhihui Li.

At least 19 recordsLinked to original sources
arXiv2026

Resonantly Scattered CIV Emission in Local Star-forming Galaxies: Radiative Transfer Constraints on High-ionization Gas in Reionization Analogs

Rest-frame ultraviolet emission lines are diagnostics of hard ionizing spectra and highly ionized gas in galaxies resembling the sources of cosmic reionization. Nebular CIV emission traces both intrinsic line production and resonant scattering through its velocity-resolved profile. Previous studies have treated CIV empirically, using integrated fluxes, equivalent widths, and line ratios rather than modeling the radiative transfer (RT) shaping the emergent profiles. We present the first systematic RT modeling of resonantly scattered CIV emission profiles in 18 local star-forming galaxies observed with HST/COS. Using the clumpy RT framework PEACOCK, idealized experiments show that profile morphology is governed by column density, gas kinematics, intrinsic equivalent width, and aperture-dependent recovery of scattered emission. We fit continuum-normalized spectra corrected for stellar contributions with a neural-network-accelerated Bayesian pipeline, reproducing P-Cygni-like, double-peaked, double-peaked with two absorption troughs, and blue-bump plus red-peak morphologies. Emission infilling connects these profile classes and complicates kinematic inferences, while aperture losses decouple observed net equivalent widths from intrinsic line production. Combining RT-inferred intrinsic equivalent widths with rest-UV diagnostics, we find that most galaxies are consistent with stellar photoionization, although some may require harder spectra or more extreme ionization conditions. Comparison with CLASSY reveals stronger associations of stellar mass and star formation rate with gas kinematics than with CIV column density. Our results establish CIV as both a tracer of hard ionizing radiation and a resonant-line probe of gas structure and kinematics, with velocity-resolved profiles providing information beyond integrated line strengths in local reionization analogs.

astro-ph.GA↗
arXiv2026

Construction of a Class of Communication-Efficient Quantum Secret Sharing Schemes

Quantum secret sharing is a fundamental technique in quantum cryptography. However, in practical quantum networks, it still faces several bottlenecks, such as high quantum communication cost and low transmission efficiency. To reduce the communication cost, ramp quantum secret sharing schemes have been proposed in existing studies. Nevertheless, intermediate sets in such schemes may leak partial information about the secret. To address this problem, this paper presents a method for detecting intermediate sets. Based on this method, we further propose a communication-efficient perfect quantum secret sharing scheme with eavesdropping detection capability.We analyze the communication cost under different numbers of participating parties and determine the range of participants that minimizes the reconstruction communication cost. The results verify the communication efficiency and security of the proposed scheme.

quant-ph↗
arXiv2026

Exploring Adaptive Masked Reconstruction for Self-Supervised Skeleton-Based Action Recognition

Recently, masked skeleton reconstruction models have emerged as strong action representation learners, driving significant progress in self-supervised skeleton-based action recognition. However, existing state-of-the-art methods must predict an exceedingly large number of spatiotemporal patches, significantly prolonging training time. Besides, by treating all spatiotemporal regions equally during reconstruction, these models are distracted from learning the critical motion patterns that underlie action semantics. To address these challenges, we propose Adaptive Masked Reconstruction (AMR), a faster and stronger pre-training framework. We first decouple the decoder from the encoder, enabling flexible prediction of larger spatiotemporal patches and dramatically reducing reconstruction complexity. Given that larger patches contain more complex information, which is challenging to predict and consequently degrades performance, we accordingly introduce an adaptive guidance module. This module identifies regions of high motion informativeness, guiding the model to focus on the most discriminative parts of each patch and alleviating reconstruction difficulty. Experiments on NTU RGB+D 60, NTU RGB+D 120, and PKU-MMD datasets demonstrate that AMR not only accelerates pre-training substantially but also improves downstream recognition accuracy, surpassing current state-of-the-art approaches.

cs.CV↗
arXiv2026

CORE: Conflict-Oriented Reasoning for General Multimodal Manipulation Detection

The rapid rise of generative AI has made multimodal fake news increasingly realistic and pervasive, posing severe threats to public trust and social stability. Existing detection methods rely heavily on manipulation-specific models and large-scale labeled data, resulting in poor generalization to emerging manipulation types. We observed that the essence of manipulated misinformation lies in its intrinsic conflicts, \textbf{i.e.,} semantic or physical inconsistencies either across modalities or with common world knowledge. Inspired by this observation, we propose \textbf{C}onflict-\textbf{O}riented \textbf{RE}asoning (\textbf{CORE}) framework, an effective paradigm that learns to endows multimodal large language models (MLLMs) with explicit conflict-capturing capability. To this end, CORE first constructs the Conflict Attribution Corpus (CAC) with fine-grained annotations of conflict factors and sources, providing essential data support for subsequent conflict perception training. By performing conflict-oriented representation enhancement and reasoning based on CAC, CORE achieves robust and generalizable conflict detection, effectively and rapidly adapting to unseen manipulation types with a few samples or in even zero-shot settings. Extensive experiments demonstrate that CORE surpasses state-of-the-art models. The dataset and code are publicly available at https://github.com/shen8424/CORE.

cs.AI↗
arXiv2026

On the Scaling of PEFT: Towards Million Personal Models of Trillion Parameters

Parameter-efficient fine-tuning (PEFT) is usually treated as a cheaper alternative to full fine-tuning. We study a broader role: small trainable adapters as persistent local state on top of strong shared foundation models. In this framing, the base model provides shared competence while adapters carry instance-specific behavior such as preferences, skills, tool habits, and memory-like updates. We organize the problem around three scaling axes: Scale Up, where stronger shared priors make small local updates more useful; Scale Down, where we study how small adapters can be while remaining reliable; and Scale Out, where many persistent adapted instances coexist. MinT provides one infrastructure example for managing adapter identity, revision, provenance, evaluation, and serving residency. Together, the results suggest that PEFT can be a compact substrate for persistent personal models rather than only a budget substitute for full fine-tuning.

cs.LG↗
arXiv2026

Same Evidence, Different Answers: Canonical-Context On-Policy Distillation for Multi-Turn Language Models

Large language models (LLMs) often solve a task when all instructions are given in a single prompt, but fail when the same information is revealed gradually across turns. When a clean FULL prompt and a RAW-SHARDED conversation contain the same complete user evidence, the model should still arrive at the same answer. We argue that a key reason for this gap is self-anchored drift: responses produced under partial information introduce unsupported assumptions, and those assumptions later distort the final answer. To reduce this effect, we propose Canonical-Context On-Policy Distillation (CCOPD). During training, the same base model is used in two roles: a frozen teacher conditioned on the clean FULL prompt and a trainable student that receives the same evidence incrementally through a multi-turn conversation; CCOPD aligns the student's behavior on its own trajectories with the teacher's canonical full-context behavior. Trained only on math problem conversations, CCOPD yields a 32\% average relative improvement in RAW-SHARDED performance over the original base model across math and five zero-shot out-of-domain task families, while largely preserving full-context performance. Further analyses suggest that CCOPD strengthens grounding in user evidence and reduces sensitivity to contamination from earlier assistant turns.

cs.CL↗
arXiv2026

SRC-Flow: Compact Semantic Representations Enable Normalizing Flows for Image Generation

Normalizing flows (NFs) provide exact likelihoods and deterministic invertible sampling, but have historically lagged behind diffusion models for large-scale image generation. We identify a key obstacle: NFs are required to learn a single invertible transport over the full ambient space, making them highly sensitive to high-dimensional representations. This leads to a semantic-capacity mismatch in modern visual representation spaces, where semantic information is compact but encoded in overcomplete features. We propose SRC-Flow, which introduces a Semantic Representation Compressor (SRC) to compact high-dimensional RAE features into a low-dimensional semantic space before flow modeling and preserve reconstruction through the frozen RAE decoder. This compact space reduces the modeling burden of NFs and enables effective likelihood-based generation in semantic representation space. We further adopt constant noise regularization tailored to the fixed unconditional bijection learned by flows. On ImageNet $256 \times 256$ and $512 \times 512$, SRC-Flow achieves state-of-the-art generation quality among normalizing flow methods, with gFID scores of 1.65 and 2.07 under classifier-free guidance, while retaining exact likelihood computation in the compact semantic representation space and deterministic invertible sampling at the flow level. Codes and models will be available at https://github.com/longtaojiang/SRC-Flow.

cs.CV↗
arXiv2026

MinT: Managed Infrastructure for Training and Serving Millions of LLMs

We present MindLab Toolkit (MinT), a managed infrastructure system for Low-Rank Adaptation (LoRA) post-training and online serving. MinT targets a setting where many trained policies are produced over a small number of expensive base-model deployments. Instead of materializing each policy as a merged full checkpoint, MinT keeps the base model resident and moves exported LoRA adapter revisions through rollout, update, export, evaluation, serving, and rollback, hiding distributed training, serving, scheduling, and data movement behind a service interface. MinT scales this path along three axes. Scale Up extends LoRA RL to frontier-scale dense and MoE architectures, including MLA and DSA attention paths, with training and serving validated beyond 1T total parameters. Scale Down moves only the exported LoRA adapter, which can be under 1% of base-model size in rank-1 settings; adapter-only handoff reduces the measured step by 18.3x on a 4B dense model and 2.85x on a 30B MoE, while concurrent multi-policy GRPO shortens wall time by 1.77x and 1.45x without raising peak memory. Scale Out separates durable policy addressability from CPU/GPU working sets: a tensor-parallel deployment supports 10^6-scale addressable catalogs (measured single-engine sweeps through 100K) and thousand-adapter active waves at cluster scale, with cold loading treated as scheduled service work and packed MoE LoRA tensors improving live engine loading by 8.5-8.7x. MinT thus manages million-scale LoRA policy catalogs while training and serving selected adapter revisions over shared 1T-class base models.

cs.LG↗
arXiv2026

Progressive Online Video Understanding with Evidence-Aligned Timing and Transparent Decisions

Visual agents operating in the wild must respond to queries precisely when sufficient evidence first appears in a video stream, a critical capability that is overlooked by conventional video LLMs evaluated in offline settings. The shift to an online, streaming paradigm introduces significant challenges: a lack of decision transparency, the difficulty of aligning response timing with visual evidence, and the need to maintain a global, causally consistent understanding under tight computational budgets. To address these issues, we propose a novel framework that decouples reasoning control from memory integration. We introduce \textbf{\model{}}, an instantiation of this framework with two core components. First, the \emph{Active Thinking Decision Maker (ATDM)} is a transparent reasoning controller that externalizes its decision process using observable progress ($\boldsymbol{\rho}$) and confidence ($\boldsymbol{c}$) metrics. This allows it to precisely time its response $t_r$ to match the first-sufficient-evidence timestamp $t^\star$ while streaming its reasoning to the user. Second, the \emph{Hierarchical Progressive Semantic Integration (HPSI)} module acts as an efficient memory system. It employs a set of learnable, multi-level aggregation tokens that are propagated across clips to build a rich, global cognitive state without exceeding token budgets. %Our approach sets a new standard on key online video understanding benchmarks, achieving strong performance of \textbf{71.6\%} on StreamingBench and \textbf{46.9\%} on OVOBench, demonstrating a robust solution for evidence-aligned and transparent online video analysis. Extensive experiments demonstrate the effectiveness of ATDM and HPSI, e.g., Thinking-QwenVL improves the accuracy of the previous state-of-the-art from 67.63\% to 71.60\% on the StreamingBench benchmark.

cs.CV↗
arXiv2026

LatentPilot: Scene-Aware Vision-and-Language Navigation by Dreaming Ahead with Latent Visual Reasoning

Existing vision-and-language navigation (VLN) models primarily reason over past and current visual observations, while largely ignoring the future visual dynamics induced by actions. As a result, they often lack an effective understanding of the causal relationship between actions and how the visual world changes, limiting robust decision-making. Humans, in contrast, can imagine the near future by leveraging action-dynamics causality, which improves both environmental understanding and navigation choices. Inspired by this capability, we propose LatentPilot, a new paradigm that exploits future observations during training as a valuable data source to learn action-conditioned visual dynamics, while requiring no access to future frames at inference. Concretely, we propose a flywheel-style training mechanism that iteratively collects on-policy trajectories and retrains the model to better match the agent's behavior distribution, with an expert takeover triggered when the agent deviates excessively. LatentPilot further learns visual latent tokens without explicit supervision; these latent tokens attend globally in a continuous latent space and are carried across steps, serving as both the current output and the next input, thereby enabling the agent to dream ahead and reason about how actions will affect subsequent observations. Experiments on R2R-CE, RxR-CE, and R2R-PE benchmarks achieve new SOTA results, and real-robot tests across diverse environments demonstrate LatentPilot's superior understanding of environment-action dynamics in scene. Project page:https://abdd.top/latentpilot/

cs.CV↗
arXiv2026

RiskProp: Collision-Anchored Self-Supervised Risk Propagation for Early Accident Anticipation

Accident anticipation aims to predict impending collisions from dashcam videos and trigger early alerts. Existing methods rely on binary supervision with manually annotated "anomaly onset" frames, which are subjective and inconsistent, leading to inaccurate risk estimation. In contrast, we propose RiskProp, a novel collision-anchored self-supervised risk propagation paradigm for early accident anticipation, which removes the need for anomaly onset annotations and leverages only the reliably annotated collision frame. RiskProp models temporal risk evolution through two observation-driven losses: first, since future frames contain more definitive evidence of an impending accident, we introduce a future-frame regularization loss that uses the model's next-frame prediction as a soft target to supervise the current frame, enabling backward propagation of risk signals; second, inspired by the empirical trend of rising risk before accidents, we design an adaptive monotonic constraint to encourage a non-decreasing progression over time. Experiments on CAP and Nexar demonstrate that RiskProp achieves state-of-the-art performance and produces smoother, more discriminative risk curves, improving both early anticipation and interpretability.

cs.CV↗
arXiv2026

Beyond Dense Futures: World Models as Structured Planners for Robotic Manipulation

Recent world-model-based Vision-Language-Action (VLA) architectures have improved robotic manipulation through predictive visual foresight. However, dense future prediction introduces visual redundancy and accumulates errors, causing long-horizon plan drift. Meanwhile, recent sparse methods typically represent visual foresight using high-level semantic subtasks or implicit latent states. These representations often lack explicit kinematic grounding, weakening the alignment between planning and low-level execution. To address this, we propose StructVLA, which reformulates a generative world model into an explicit structured planner for reliable control. Instead of dense rollouts or semantic goals, StructVLA predicts sparse, physically meaningful structured frames. Derived from intrinsic kinematic cues (e.g., gripper transitions and kinematic turning points), these frames capture spatiotemporal milestones closely aligned with task progress. We implement this approach through a two-stage training paradigm with a unified discrete token vocabulary: the world model is first trained to predict structured frames and subsequently optimized to map the structured foresight into low-level actions. This approach provides clear physical guidance and bridges visual planning and motion control. In our experiments, StructVLA achieves strong average success rates of 75.0% on SimplerEnv-WidowX and 94.8% on LIBERO. Real-world deployments further demonstrate reliable task completion and robust generalization across both basic pick-and-place and complex long-horizon tasks.

cs.RO↗
arXiv2026

BLUEPRINT: Blue-dominant Lyman-$\alpha$ emission as evidence of gas inflow in ultra-low-mass galaxies at z = 3

We report the detection of a clumpy, blue-dominated Ly$\alpha$ emission at z = 3.066 located in the heart of a cosmic web filament in the MUSE eXtremely Deep Field (MXDF), spatially associated with the formation of two compact star-forming regions revealed by deep JWST/NIRCam imaging. Gas accretion in these regions is indicated by the blue-dominated Ly$\alpha$ profiles, spectral signatures that are rarely observed. Radiative transfer simulations of the Ly$\alpha$ profile using a clumpy multiphase model suggest a radial inflow of gas clumps with a velocity of 100 km/s. Embedded in this Ly$\alpha$ structure, we find that the associated main galaxy dominates the stellar mass budget, while the two compact ultra-low-mass systems ($\log(M_\star/M_\odot) = 6.3\text{-}6.9$) have formed the bulk of their stellar mass in less than 7 Myr. These two components also have high specific star-formation rates, and elevated ionisation parameters, consistent with recent bursty star formation. This system provides compelling observational evidence that how gas accretion, most likely from the cosmic web, can induce starburst in ultra-low-mass galaxies.

astro-ph.GA↗
arXiv2026

See, Plan, Rewind: Progress-Aware Vision-Language-Action Models for Robust Robotic Manipulation

Measurement of task progress through explicit, actionable milestones is critical for robust robotic manipulation. This progress awareness enables a model to ground its current task status, anticipate verifiable intermediate states, and detect and recover from failures when progress stalls. To embody this capability, we introduce \textbf{S}ee, \textbf{P}lan, \textbf{R}ewind (SPR), a progress-aware vision-language-action framework that dynamically grounds language instructions into a sequence of spatial subgoals. SPR operates through a continuous core cycle, Seeing the current state and upcoming milestone, Planning a trajectory towards the next 2D waypoint, and Rewinding to a recoverable state upon failure by monitoring progress against the expected sequence. This closed-loop approach enables robust error correction without requiring additional training data or auxiliary models. Extensive experiments demonstrate the framework's effectiveness, generalization and robustness: SPR outperforms the MolmoAct baseline by 5\% on the LIBERO benchmark. On the challenging LIBERO-Plus benchmark with unseen instructions and initial states, SPR achieves state-of-the-art robustness with the smallest performance drop, surpassing OpenVLA-OFT and UniVLA, demonstrating superior out-of-distribution robustness.

cs.RO↗
arXiv2026

Meta-PINNs: Meta-Learning Enhanced Physics-Informed Machine Learning Framework for Turbomachinery Flow Predictions under Varying Operation Conditions

Coupling physics with machine learning models has shown great potential for solving fluid dynamics problems governed by partial differential equations. However, conventional methods, such as physics-informed neural networks, often suffer from slow convergence, unstable training, and limited generalization across different flow conditions. To overcome these challenges, this study proposes a novel meta-learning en- hanced physics-informed neural networks (Meta-PINNs) framework, which integrates a meta-optimization strategy into the training process. The approach allows the model to automatically adapt its learning process to varying physical regimes, thereby substantially improving both training efficiency and predictive robustness. The proposed Meta-PINNs model is evaluated on two representative flow problems: (1) unsteady flow around a circular cylinder at multiple inlet Reynolds numbers, and (2) steady turbulent flow within a compressor cascade passage at various angles of attack. In both cases, the extrapolation performance of the developed framework is comprehensively tested by predicting the flow fields at Reynolds numbers and angles of attack that are not included in the training set. The results demonstrate that Meta-PINNs achieve a 1-2 order-of-magnitude improvement in accuracy over vanilla physics-informed neural networks and standard neural networks, while reducing computational cost by up to 95.7 % and 92.1 %, respectively. It successfully captures the sequential patterns of key flow features such as pressure and velocity distributions under unseen conditions. Thus, the findings confirm that the Meta-PINNs framework offers a notable improvement in convergence and generalization over existing machine learning approaches, providing a promising pathway toward smart simulations of complex turbomachinery flows.

physics.flu-dyn↗
arXiv2026

Kinematically Coherent Multiphase Galactic Winds in Star-Forming Galaxies Revealed by Unified Radiative Transfer Modeling of UV Emission and Absorption Lines

We present PEACOCK, a three-dimensional Monte Carlo radiative transfer (RT) framework designed to self-consistently model rest-frame ultraviolet emission and absorption lines arising from multiphase, clumpy galactic winds. Applied to deep HST/COS spectra of 50 nearby star-forming galaxies, PEACOCK reproduces 220 observed profiles of Ly-alpha, Si II, C II, Si III, Si IV, and C IV spanning absorption, emission, and P-Cygni-like morphologies within a single CGM model. By combining Monte Carlo RT with deep-learning acceleration and nested sampling, the framework enables fully converged multi-line inference at a small fraction of the cost of traditional RT grids. Systematic experiments show that ion column densities, bulk outflow velocities, and turbulent motions leave distinct imprints on line profiles, allowing the underlying gas properties to be constrained with minimal degeneracy. Purely radial accelerating flows often fail to reproduce the observed absorption morphologies, whereas macroscopic velocity dispersion naturally produces the broad asymmetric troughs seen in the data, indicating that turbulent motions are a key component of outflow kinematics. The inferred kinematics reveal strong coherence among low- and high-ionization metal lines in both bulk and turbulent velocities, consistent with a dynamically coupled multiphase wind. In contrast, neutral hydrogen shows weaker correspondence with metals, suggesting incomplete mixing and a distinct kinematic structure. By unifying emission and absorption diagnostics across multiple ions, PEACOCK provides a physically grounded bridge between UV observations and theoretical models of galactic winds.

astro-ph.GA↗
arXiv2026

The Prevalence of Turbulence-Regulated Multiphase Galactic Winds in Star-Forming Galaxies

We build upon our previously developed multi-ion radiative transfer (RT) framework, PEACOCK, to investigate the kinematic and energetic structure of cool-to-warm galactic winds in a sample of 50 nearby star-forming galaxies. Using self-consistent constraints derived from joint modeling of Ly-alpha and multiple ultraviolet metal lines, we analyze how bulk outflows and turbulent motions contribute to the dynamics and energy budget of galactic winds in the circumgalactic medium (CGM). We find that macroscopic turbulent velocities are often comparable to, and sometimes exceed, the coherent bulk outflow velocity. The associated turbulent pressure frequently dominates over both microscopic pressure and ram pressure, indicating that turbulence is a major contributor to the kinetic energy budget of the CGM wind. Wind kinematics, ionic column densities, and metal mass outflow rates all scale systematically with stellar mass and star formation rate, demonstrating a strong coupling between stellar feedback and CGM structure. Including turbulent motions strengthens these CGM-galaxy scaling relations and favors an energy-driven feedback regime. The total kinetic energy flux of the cool-to-warm CGM correlates tightly with the mechanical energy injection rate from star formation, implying that stellar feedback provides sufficient power to sustain both coherent outflows and turbulence. Comparisons with phenomenological line-profile fitting methods further show that simplified treatments can introduce systematic biases in inferred wind properties. Together these results support a turbulence-regulated picture of galactic winds in which a substantial fraction of feedback energy is stored in turbulent motions within a multiphase CGM.

astro-ph.GA↗
arXiv2025

Which Layer Causes Distribution Deviation? Entropy-Guided Adaptive Pruning for Diffusion and Flow Models

Large-scale vision generative models, including diffusion and flow models, have demonstrated remarkable performance in visual generation tasks. However, transferring these pre-trained models to downstream tasks often results in significant parameter redundancy. In this paper, we propose EntPruner, an entropy-guided automatic progressive pruning framework for diffusion and flow models. First, we introduce entropy-guided pruning, a block-level importance assessment strategy specifically designed for generative models. Unlike discriminative models, generative models require preserving the diversity and condition-fidelity of the output distribution. As the importance of each module can vary significantly across downstream tasks, EntPruner prioritizes pruning of less important blocks using data-dependent Conditional Entropy Deviation (CED) as a guiding metric. CED quantifies how much the distribution diverges from the learned conditional data distribution after removing a block. Second, we propose a zero-shot adaptive pruning framework to automatically determine when and how much to prune during training. This dynamic strategy avoids the pitfalls of one-shot pruning, mitigating mode collapse, and preserving model performance. Extensive experiments on DiT and SiT models demonstrate the effectiveness of EntPruner, achieving up to 2.22$\times$ inference speedup while maintaining competitive generation quality on ImageNet and three downstream datasets.

cs.CV↗