Search arXiv⌕ Search

arXiv subjects

Yuxuan Zhou

Publications and source records attributed to Yuxuan Zhou.

At least 55 records · Page 3Linked to original sources
arXiv2026

PaperX: A Unified Framework for Multimodal Academic Presentation Generation with Scholar DAG

Transforming scientific papers into multimodal presentation content is essential for research dissemination but remains labor intensive. Existing automated solutions typically treat each format as an isolated downstream task, leading to redundant processing and semantic inconsistency. We introduce PaperX, a unified framework that models academic presentation generation as a structural transformation and rendering process. Central to our approach is the Scholar DAG, an intermediate representation that decouples the paper's logical structure from its final presentation syntax. By applying adaptive graph traversal strategies, PaperX generates diverse, high quality outputs from a single source. Comprehensive evaluations demonstrate that our framework achieves the state of the art performance in content fidelity and aesthetic quality while significantly improving cost efficiency compared to specialized single task agents.

cs.DL↗
arXiv2026

AnyTouch 2: General Optical Tactile Representation Learning For Dynamic Tactile Perception

Real-world contact-rich manipulation demands robots to perceive temporal tactile feedback, capture subtle surface deformations, and reason about object properties as well as force dynamics. Although optical tactile sensors are uniquely capable of providing such rich information, existing tactile datasets and models remain limited. These resources primarily focus on object-level attributes (e.g., material) while largely overlooking fine-grained tactile temporal dynamics during physical interactions. We consider that advancing dynamic tactile perception requires a systematic hierarchy of dynamic perception capabilities to guide both data collection and model design. To address the lack of tactile data with rich dynamic information, we present ToucHD, a large-scale hierarchical tactile dataset spanning tactile atomic actions, real-world manipulations, and touch-force paired data. Beyond scale, ToucHD establishes a comprehensive tactile dynamic data ecosystem that explicitly supports hierarchical perception capabilities from the data perspective. Building on it, we propose AnyTouch 2, a general tactile representation learning framework for diverse optical tactile sensors that unifies object-level understanding with fine-grained, force-aware dynamic perception. The framework captures both pixel-level and action-specific deformations across frames, while explicitly modeling physical force dynamics, thereby learning multi-level dynamic perception capabilities from the model perspective. We evaluate our model on benchmarks that covers static object properties and dynamic physical attributes, as well as real-world manipulation tasks spanning multiple tiers of dynamic perception capabilities-from basic object-level understanding to force-aware dexterous manipulation. Experimental results demonstrate consistent and strong performance across sensors and tasks.

cs.RO↗
arXiv2026

Beyond Closed-Pool Video Retrieval: A Benchmark and Agent Framework for Real-World Video Search and Moment Localization

Traditional video retrieval benchmarks focus on matching precise descriptions to closed video pools, failing to reflect real-world searches characterized by fuzzy, multi-dimensional memories on the open web. We present \textbf{RVMS-Bench}, a comprehensive system for evaluating real-world video memory search. It consists of \textbf{1,440 samples} spanning \textbf{20 diverse categories} and \textbf{four duration groups}, sourced from \textbf{real-world open-web videos}. RVMS-Bench utilizes a hierarchical description framework encompassing \textbf{Global Impression, Key Moment, Temporal Context, and Auditory Memory} to mimic realistic multi-dimensional search cues, with all samples strictly verified via a human-in-the-loop protocol. We further propose \textbf{RACLO}, an agentic framework that employs abductive reasoning to simulate the human ``Recall-Search-Verify'' cognitive process, effectively addressing the challenge of searching for videos via fuzzy memories in the real world. Experiments reveal that existing MLLMs still demonstrate insufficient capabilities in real-world Video Retrieval and Moment Localization based on fuzzy memories. We believe this work will facilitate the advancement of video retrieval robustness in real-world unstructured scenarios.

cs.CV↗
arXiv2026

Quantum Charging Advantage in Superconducting Solid-State Batteries

Quantum battery, as a novel energy storage device, offers the potential for unprecedented efficiency and performance beyond the capabilities of classical systems, with broad implications for future quantum technologies. Here, we experimentally \RefC{demonstrate quantum charging advantage (QCA)} in a scalable solid-state quantum battery. More specifically, we show how double-excitation Hamiltonians for two-level systems promote scalable QCA \RefB{with standard methods.} We effectively implement the collective evolution of quantum systems with 2 up to 12 battery cells in a superconducting quantum processor, and study the performance of quantum charging compared to its uncorrelated classical counterpart. The model considered is a linear chain of superconducting transmon qubits with only \textit{nearest-neighbor} and \textit{pairwise} interactions, which constitute the simplest model of a multi-cell quantum battery. Our results empirically realize substantial QCA without the necessity of adopting long-range and many-body interactions \RefB{ and showcase the quantum features of the QB charging processes with measurements of non-zero coherent ergotropy, incoherent ergotropy and entanglement,} revealing a promising prospect for further developments of efficient and experimentally feasible protocols for QCA.

quant-ph↗
arXiv2026

MaxSup: Overcoming Representation Collapse in Label Smoothing

Label Smoothing (LS) is widely adopted to reduce overconfidence in neural network predictions and improve generalization. Despite these benefits, recent studies reveal two critical issues with LS. First, LS induces overconfidence in misclassified samples. Second, it compacts feature representations into overly tight clusters, diluting intra-class diversity, although the precise cause of this phenomenon remained elusive. In this paper, we analytically decompose the LS-induced loss, exposing two key terms: (i) a regularization term that dampens overconfidence only when the prediction is correct, and (ii) an error-amplification term that arises under misclassifications. This latter term compels the network to reinforce incorrect predictions with undue certainty, exacerbating representation collapse. To address these shortcomings, we propose Max Suppression (MaxSup), which applies uniform regularization to both correct and incorrect predictions by penalizing the top-1 logit rather than the ground-truth logit. Through extensive feature-space analyses, we show that MaxSup restores intra-class variation and sharpens inter-class boundaries. Experiments on large-scale image classification and multiple downstream tasks confirm that MaxSup is a more robust alternative to LS. Code is available at: https://github.com/ZhouYuxuanYX/Maximum-Suppression-Regularization

cs.LG↗
arXiv2026

Quantum-Enhanced Sensing Enabled by Scrambling-Induced Genuine Multipartite Entanglement

Quantum sensing leverages quantum resources to surpass the standard quantum limit, yet many existing protocols rely on the preparation of complex entangled states and Hamiltonian engineering, posing challenges for universality and scalability. Here, we report an experimental realization of a universal protocol, known as Butterfly Metrology, proposed in [arXiv:2411.12794], demonstrating a scrambling-based approach for quantum-enhanced sensing on a superconducting quantum processor. By exploiting many-body information scrambling, we observe quantum-enhanced sensitivity to an encoded phase beyond the standard quantum limit, with a scaling consistent with a factor-of-two of the Heisenberg limit for system sizes of up to 10 qubits. Importantly, we experimentally establish a connection between the enhanced sensitivity and the dynamics of the out-of-time-order correlator (OTOC), and show that the buildup of scrambling-induced genuine multipartite entanglement underlies the observed sensitivity enhancement. Our results demonstrate a scalable and practical approach for quantum-enhanced sensing in interacting many-body quantum systems.

quant-ph↗
arXiv2026

Sharp quantitative stability for the fractional Sobolev trace inequality

In this paper, we study the stability of fractional Sobolev trace inequality within both the functional and critical point settings. In the functional setting, we establish the following sharp estimate: $$C_{\mathrm{BE}}(n,m,α)\inf_{v\in\mathcal{M}_{n,m,α}}\left\Vert f-v\right\Vert_{D_α(\mathbb{R}^n)}^2 \leq \left\Vert f\right\Vert_{D_α(\mathbb{R}^n)}^2 - S(n,m,α) \left\Vertτ_mf\right\Vert_{L^{q}(\mathbb{R}^{n-m})}^2,$$ where $0\leq m< n$, $\frac{m}{2}<α<\frac{n}{2}, q=\frac{2(n-m)}{n-2α}$ and $\mathcal{M}_{n,m,α}$ denotes the manifold of extremal functions. Additionally, We find an explicit bound for the stability constant $C_{\mathrm{BE}}$ and establish a compactness result ensuring the existence of minimizers. In the critical point setting, we investigate the validity of a sharp quantitative profile decomposition related to the Escobar trace inequality and establish a qualitative profile decomposition for the critical elliptic equation \begin{equation*} Δu= 0 \quad\text{in }\mathbb{R}_+^n,\quad\frac{\partial u}{\partial t}=-|u|^{\frac{2}{n-2}}u \quad\text{on }\partial\mathbb{R}_+^n. \end{equation*} We then derive the sharp stability estimate: $$ C_{\mathrm{CP}}(n,ν)d(u,\mathcal{M}_{\mathrm{E}}^ν)\leq \left\Vert Δu +|u|^{\frac{2}{n-2}}u\right\Vert_{H^{-1}(\mathbb{R}_+^n)}, $$ where $ν=1,n\geq 3$ or $ν\geq2,n=3$ and $\mathcal{M}_{\mathrm{E}}^ν$ represents the manifold consisting of $ν$ weak-interacting Escobar bubbles. Through some refined estimates, we also give a strict upper bound for $C_{\mathrm{CP}}(n,1)$, which is $\frac{2}{n+2}$.

math.AP↗
arXiv2026

Classification of positive solutions to the Hénon-Sobolev critical systems

In this paper, we investigate positive solutions to the following Hénon-Sobolev critical system: $$ -\mathrm{div}(|x|^{-2a}\nabla u)=|x|^{-bp}|u|^{p-2}u+να|x|^{-bp}|u|^{α-2}|v|^βu\quad\text{in }\mathbb{R}^n,$$ $$ -\mathrm{div}(|x|^{-2a}\nabla v)=|x|^{-bp}|v|^{p-2}v+νβ|x|^{-bp}|u|^α|v|^{β-2}v\quad\text{in }\mathbb{R}^n,$$ $$u,v\in D_a^{1,2}(\mathbb{R}^n),$$ where $n\geq 3,-\infty< a<\frac{n-2}{2},a\leq b 0$ and $α>1,β>1$ satisfying $α+β=p$. Our findings are divided into two parts, according to the sign of the parameter $a$. For $a\geq 0$, we demonstrate that any positive solution $(u,v)$ is synchronized, indicating that $u$ and $v$ are constant multiples of positive solutions to the decoupled Hénon equation: \begin{equation*} -\mathrm{div}(|x|^{-2a}\nabla w)=|x|^{-bp}|w|^{p-2}w. \end{equation*} For $a<0$ and $b>a$, we characterize all nonnegative ground states. Additionally, we study the nondegeneracy of nonnegative synchronized solutions. This work also delves into some general $k$-coupled Hénon-Sobolev critical systems.

math.AP↗
arXiv2026

The stability on the Caffarelli-Kohn-Nirenberg and Hardy-type inequalities and beyond

In this paper, we establish several improved Caffarelli-Kohn-Nirenberg and Hardy-type inequalities. Our main results are divided into two parts. In the first part, we consider the following Caffarelli-Kohn-Nirenberg inequality: \begin{equation*} \left(\int_{\mathbb{R}^n}|x|^{-pa}|\nabla u|^pdx\right)^{\frac{1}{p}}\geq S(p,a,b)\left(\int_{\mathbb{R}^n}|x|^{-qb}|u|^qdx\right)^{\frac{1}{q}},\quad\forall\; u\in D_a^p(\mathbb{R}^n), \end{equation*} We establish gradient stability of this inequality in both functional and critical settings, and we derive some functional properties of the stability constant. Building on the gradient stability, we also obtain several refined Sobolev-type embeddings involving weak Lebesgue norms for functions supported in general domains. In the second part, we focus on various classical Hardy-type inequalities, including the standard Hardy inequality, the $L^p$-logarithmic Sobolev inequality with weights, the logarithmic Hardy inequality, the Hardy-Morrey inequality, the Hardy-Sobolev interpolation inequality, and the interpolated Caffarelli-Kohn-Nirenberg inequality. We investigate their weighted versions and derive corresponding extremal functions, refinements, new remaining terms and stability constants.

math.AP↗
arXiv2026

Degenerate stability of critical points of the Caffarelli-Kohn-Nirenberg inequality along the Felli-Schneider curve

In this paper, we investigate the validity of a quantitative version of stability for the critical Hardy-Hénon equation \begin{equation*} H(u):=÷(|x|^{-2a}\nabla u)+|x|^{-pb}|u|^{p-2}u=0,\quad u\in D_a^{1,2}(\R^n), \end{equation*} \begin{equation*} n\geq 2,\quad a<b<a+1,\quad a<\frac{n-2}{2},\quad p=\frac{2n}{n-2+2(b-a)}, \end{equation*} which is well known as the Euler-Lagrange equation of the classical Caffarelli-Kohn-Nirenberg inequality. Establishing quantitative stability for this equation amounts to finding a nonnegative function $F$ such that the estimate \begin{equation*} \inf_{\substack{U_i\in\mathcal{M} 1\leq i\leqν}}\norm*{u-\sum_{i=1}^νU_i}_{D_a^{1,2}(\R^n)}\leq C(a,b,n)F(\norm*{H(u)}_{D_a^{-1,2}(\R^n)}) \end{equation*} holds for any nonnegative function $u$ satisfying \begin{equation*} \left(ν-\frac{1}{2}\right)S(a,b,n)^{\frac{p}{p-2}}\leq\int_{\R^n}|x|^{-2a}|\nabla u|^2\mathrm{d}x\leq \left(ν+\frac{1}{2}\right)S(a,b,n)^{\frac{p}{p-2}}. \end{equation*} Here $ν\in\N_+$ and $\mathcal{M}$ denotes the set of positive solutions to this equation. When $(a,b)$ falls above the Felli-Schneider curve, Wei and Wu \cite{Wei} found an optimal $F$. Their proof relies heavily on the fact that $\mathcal{M}$ is non-degenerate. When $(a,b)$ falls on the Felli-Schneider curve, due to the absence of the non-degeneracy condition, it becomes complicated and technical to find a suitable $F$. In this paper, we focus on this case. When $ν=1$, we obtain an optimal $F$. When $ν\geq2$ and $u$ is not too degenerate, we also derive an optimal $F$. To our knowledge, the results in this paper provide the first instance of degenerate stability in the critical point setting. We believe that our methods will be useful in other works on degenerate stability.

math.AP↗
arXiv2026

DemoTuner: Automatic Performance Tuning for Database Management Systems Based on Demonstration Reinforcement Learning

The performance of modern DBMSs such as MySQL and PostgreSQL heavily depends on the configuration of performance-critical knobs. Manual tuning these knobs is laborious and inefficient due to the complex and high-dimensional nature of the configuration space. Among the automated tuning methods, reinforcement learning (RL)-based methods have recently sought to improve the DBMS knobs tuning process from several different perspectives. However, they still encounter challenges with slow convergence speed during offline training. In this paper, we mainly focus on how to leverage the valuable tuning hints contained in various textual documents such as DBMS manuals and web forums to improve the offline training of RL-based methods. To this end, we propose an efficient DBMS knobs tuning framework named DemoTuner via a novel LLM-assisted demonstration reinforcement learning method. Specifically, to comprehensively and accurately mine tuning hints from documents, we design a structured chain of thought prompt to employ LLMs to conduct a condition-aware tuning hints extraction task. To effectively integrate the mined tuning hints into RL agent training, we propose a hint-aware demonstration reinforcement learning algorithm HA-DDPGfD in DemoTuner. As far as we know, DemoTuner is the first work to introduce the demonstration reinforcement learning algorithm for DBMS knobs tuning. Experimental evaluations conducted on MySQL and PostgreSQL across various workloads demonstrate that DemoTuner achieves performance gains of up to 44.01% for MySQL and 39.95% for PostgreSQL over default configurations. Compared with three representative baseline methods, DemoTuner is able to further reduce the execution time by up to 10.03%, while always consuming the least online tuning cost. Additionally, DemoTuner also exhibits superior adaptability to application scenarios with unknown workloads.

cs.LG↗
arXiv2025

Entanglement protection induced by mixed noise

Contrary to the conventional view that noise is detrimental, we show that mixed noise can protect entanglement in a two-atom-cavity system. Specifically, the leakage of the cavity and the stochastic atom-cavity couplings are modeled as two types of noises. From the analytical derivation of the dynamical equations, the mechanism of the entanglement protection is revealed as the high-frequency(HF) noise in the atom-cavity couplings could suppress the decoherence caused by the cavity leakage, thus protect the entanglement. We investigate the entanglement protection induced by mixed noise constructed from diverse noise types, including the Ornstein-Uhlenbeck noise, flicker noise, and telegraph noise. Numerical simulations demonstrate that entanglement protection depends critically on the proportion of HF components in the power spectral density of the mixed noise. Our work establishes that enhanced HF components are essential for effective noise-assisted entanglement protection, offering key insights for noise engineering in practical open quantum systems.

quant-ph↗
arXiv2025

Kitaev interaction and possible spin liquid state in CoI2 and Co2/3Mg1/3I2

Kitaev materials are of great interest due to their potential in realizing quantum spin liquid (QSL) states and applications in topological quantum computing. In the pursuit of realizing Kitaev QSL, a Mott insulator with strong bond-dependent frustration and weak geometric frustration is highly desirable. Here we explore Kitaev physics in the van der Waals triangular antiferromagnet (AF) CoI$_2$, through the spin-orbital states and Wannier function analyses, exact diagonalization and density matrix renormalization group study of the electronic structure and magnetic properties. We find that the high-spin Co$^{2+}$ ion is in the $J_\mathrm{eff}=1/2$ state because of strong spin-orbit coupling, and the weak trigonal elongation and crystal field contribute to the observed weak in-plane magnetic anisotropy. The strong $t_{2g}$-$e_g$ hopping via the strong Co 3$d$-I 5$p$ hybridization gives rise to a strong Kitaev interaction ($K_1$) at the first nearest neighbors (1NN), and the long Co-Co distance and the weak $t_{2g}$-$t_{2g}$ hoppings determine a weak Heisenberg interaction $J_1$. The resultant $|K_1/J_1|$ = 6.63 confirms a strong bond-dependent frustration, while the geometric frustration due to the 3NN Heisenberg interaction $J_3$ gets involved, and they all together result in the experimental helical AF order in CoI$_2$. We then propose to suppress the $J_3$ using a partial Mg substitution for Co, and indeed we find that Co$_{2/3}$Mg$_{1/3}$I$_2$ has the much reduced geometric frustration but hosts the robust bond-dependent frustration, and thus it would be a promising Kitaev material being so far closest to the QSL state.

cond-mat.mtrl-sci↗
arXiv2025

Logical operations with a dynamical qubit in Floquet-Bacon-Shor code

Quantum error correction (QEC) protects quantum systems against inevitable noises and control inaccuracies, providing a pathway towards fault-tolerant (FT) quantum computation. Stabilizer codes, including surface code and color code, have long been the focus of research and have seen significant experimental progress in recent years. Recently proposed time-dynamical QEC, including Floquet codes and generalized time-dynamical code implementations, opens up new opportunities for FT quantum computation. By employing a periodic schedule of low-weight parity checks, Floquet codes can generate additional dynamical logical qubits, offering enhanced error correction capabilities and potentially higher code performance. Here, we experimentally implement the Floquet-Bacon-Shor code on a superconducting quantum processor. We encode a dynamical logical qubit within a $3\times 3$ lattice of data qubits, alongside a conventional static logical qubit. We demonstrate FT encoding and measurement of the two-qubit logical states, and stabilize these states using repeated error detection. We showcase universal single-qubit logical gates on the dynamical qubit. Furthermore, by implementing a logical CNOT gate, we entangle the dynamical and static logical qubits, generating an error-detected logical Bell state with a fidelity of 75.9\%. Our results highlight the potential of Floquet codes for resource-efficient FT quantum computation.

quant-ph↗
arXiv2025

SF-Loc: A Visual Mapping and Geo-Localization System based on Sparse Visual Structure Frames

For high-level geo-spatial applications and intelligent robotics, accurate global pose information is of crucial importance. Map-aided localization is a universal approach to overcome the limitations of global navigation satellite system (GNSS) in challenging environments. However, current solutions face challenges in terms of mapping flexibility, storage burden and re-localization performance. In this work, we present SF-Loc, a lightweight visual mapping and map-aided localization system, whose core idea is the map representation based on sparse frames with dense but compact depth, termed as visual structure frames. In the mapping phase, multi-sensor dense bundle adjustment (MS-DBA) is applied to construct geo-referenced visual structure frames. The local co-visbility is checked to keep the map sparsity and achieve incremental mapping. In the localization phase, coarse-to-fine vision-based localization is performed, in which multi-frame information and the map distribution are fully integrated. To be specific, the concept of spatially smoothed similarity (SSS) is proposed to overcome the place ambiguity, and pairwise frame matching is applied for efficient and robust pose estimation. Experimental results on the cross-season dataset verify the effectiveness of the system. In complex urban road scenarios, the map size is down to 3 MB per kilometer and stable decimeter-level re-localization can be achieved. The code will be made open-source soon (https://github.com/GREAT-WHU/SF-Loc).

cs.RO↗
arXiv2025

MASt3R-Fusion: Integrating Feed-Forward Visual Model with IMU, GNSS for High-Functionality SLAM

Visual SLAM is a cornerstone technique in robotics, autonomous driving and extended reality (XR), yet classical systems often struggle with low-texture environments, scale ambiguity, and degraded performance under challenging visual conditions. Recent advancements in feed-forward neural network-based pointmap regression have demonstrated the potential to recover high-fidelity 3D scene geometry directly from images, leveraging learned spatial priors to overcome limitations of traditional multi-view geometry methods. However, the widely validated advantages of probabilistic multi-sensor information fusion are often discarded in these pipelines. In this work, we propose MASt3R-Fusion,a multi-sensor-assisted visual SLAM framework that tightly integrates feed-forward pointmap regression with complementary sensor information, including inertial measurements and GNSS data. The system introduces Sim(3)-based visualalignment constraints (in the Hessian form) into a universal metric-scale SE(3) factor graph for effective information fusion. A hierarchical factor graph design is developed, which allows both real-time sliding-window optimization and global optimization with aggressive loop closures, enabling real-time pose tracking, metric-scale structure perception and globally consistent mapping. We evaluate our approach on both public benchmarks and self-collected datasets, demonstrating substantial improvements in accuracy and robustness over existing visual-centered multi-sensor SLAM systems. The code will be released open-source to support reproducibility and further research (https://github.com/GREAT-WHU/MASt3R-Fusion).

cs.RO↗
arXiv2025

Enhancing the Medical Context-Awareness Ability of LLMs via Multifaceted Self-Refinement Learning

Large language models (LLMs) have shown great promise in the medical domain, achieving strong performance on several benchmarks. However, they continue to underperform in real-world medical scenarios, which often demand stronger context-awareness, i.e., the ability to recognize missing or critical details (e.g., user identity, medical history, risk factors) and provide safe, helpful, and contextually appropriate responses. To address this issue, we propose Multifaceted Self-Refinement (MuSeR), a data-driven approach that enhances LLMs' context-awareness along three key facets (decision-making, communication, and safety) through self-evaluation and refinement. Specifically, we first design a attribute-conditioned query generator that simulates diverse real-world user contexts by varying attributes such as role, geographic region, intent, and degree of information ambiguity. An LLM then responds to these queries, self-evaluates its answers along three key facets, and refines its responses to better align with the requirements of each facet. Finally, the queries and refined responses are used for supervised fine-tuning to reinforce the model's context-awareness ability. Evaluation results on the latest HealthBench dataset demonstrate that our method significantly improves LLM performance across multiple aspects, with particularly notable gains in the context-awareness axis. Furthermore, by incorporating knowledge distillation with the proposed method, the performance of a smaller backbone LLM (e.g., Qwen3-32B) surpasses its teacher model, achieving a new SOTA across all open-source LLMs on HealthBench (63.8%) and its hard subset (43.1%). Code and dataset will be released at https://muser-llm.github.io.

cs.AI↗
arXiv2025

Improving Deepfake Detection with Reinforcement Learning-Based Adaptive Data Augmentation

The generalization capability of deepfake detectors is critical for real-world use. Data augmentation via synthetic fake face generation effectively enhances generalization, yet current SoTA methods rely on fixed strategies-raising a key question: Is a single static augmentation sufficient, or does the diversity of forgery features demand dynamic approaches? We argue existing methods overlook the evolving complexity of real-world forgeries (e.g., facial warping, expression manipulation), which fixed policies cannot fully simulate. To address this, we propose CRDA (Curriculum Reinforcement-Learning Data Augmentation), a novel framework guiding detectors to progressively master multi-domain forgery features from simple to complex. CRDA synthesizes augmented samples via a configurable pool of forgery operations and dynamically generates adversarial samples tailored to the detector's current learning state. Central to our approach is integrating reinforcement learning (RL) and causal inference. An RL agent dynamically selects augmentation actions based on detector performance to efficiently explore the vast augmentation space, adapting to increasingly challenging forgeries. Simultaneously, the agent introduces action space variations to generate heterogeneous forgery patterns, guided by causal inference to mitigate spurious correlations-suppressing task-irrelevant biases and focusing on causally invariant features. This integration ensures robust generalization by decoupling synthetic augmentation patterns from the model's learned representations. Extensive experiments show our method significantly improves detector generalizability, outperforming SOTA methods across multiple cross-domain datasets.

cs.CV↗