arXiv subjects

Ziyan Chen

Publications and source records attributed to Ziyan Chen.

At least 19 recordsLinked to original sources

arXiv2026

DramaChain Bench: An End-to-End Benchmark for Short-Drama Generation

Commercial short-drama production follows a multi-stage chain: script, storyboard, keyframe imagery, shot-level video, and the finished short drama. Most existing benchmarks evaluate solely the video-generation stage using pre-authored inputs instead of real upstream pipeline outputs. This leaves two critical questions unanswerable: whether each stage adheres to the original script intent (rather than only its immediate input prompt), and whether disparate shots remain coherent after assembly into multi-episode releases. We present DramaChain Bench, the first short-drama benchmark that evaluates every stage of the complete production chain. It is built upon three in-house systems sharing one dimension system, DramaChain Dimensions: five evaluation axes instantiated at every stage, resolving into 63 leaf dimensions. DramaChain Agent is calibrated against commercial short-drama platforms in both workflow and finished short-drama quality, enabling stage-wise fair comparison across models. DramaChain Labeling System has each of the 5,785 items scored independently by three professional annotators, with all defects spatio-temporally localised and selected from a predefined defect list. This process produces 17,488 valid scores and 255,925 traceable attribution records. The human annotations confirm that upstream defects cascade across the pipeline, demonstrating that final episode quality is not governed by video generation alone. DramaChain Agentic Judge then scores every leaf dimension automatically, gathering evidence over multiple agentic rounds before judging against a per-item checklist; it reproduces the model ranking at a mean PLCC of 0.918, enough to admit new models at no annotation cost.

cs.AI↗

arXiv2026

Beyond Explicit Generators: Distribution-Free Linear-Decomposition Attacks on Public-Key Encryption

Linear-decomposition attacks can break public-key schemes without recovering the secret algebraic action: when a target public state lies in a known linear span, its decomposition coefficients transfer through the unknown action to reveal the shared value. We study a setting in which the adversary uses only the public sampling-and-evaluation oracle available to honest participants, the induced distribution is arbitrary, and the goal is to attack future ciphertexts rather than recover the full algebraic span. We model public paired samples under a fixed secret linear transport and define the sampled-orbit dimension as the effective dimension of the encryption distribution. We prove distribution-free one-shot recovery, a high-probability certificate for the future-ciphertext coverage of a sampled span, and the optimal sampled-span complexity $m^\star_{\mathrm{span}}(r,\varepsilon,\delta) =\Theta((r+\log(1/\delta))/\varepsilon)$. These results yield a generic impossibility theorem: publicly samplable linear key transport with polynomial sampled-orbit dimension is incompatible with IND--CPA security when the transported value determines the decryption payload. We apply the framework to the 2024 probabilistic PKE from twisted--skew group rings. Its underlying Computational Twisted--Skew Problem admits a sampler-only linear attack using independently generated public protocol samples, yielding plaintext recovery and constant IND--CPA advantage. Experiments verify the linear transport and end-to-end recovery, and show that high future-ciphertext coverage may precede recovery of the full algebraic span.

cs.CR↗

arXiv2026

Tensor--Action Ko--Lee Cryptography: A Framework and Structural Cryptanalysis of Commuting Subgroup Constructions

Tensor isomorphism has been studied as an algebraic problem relevant to post-quantum cryptography, while its use in public-key encryption remains open. In this paper, we formulate a Ko--Lee-style framework for public-key encryption from cubic tensor actions and prove its formal correctness. We then show that the framework is generically insecure when the commuting matrix subgroups are given by public finite generating sets. Viewing a cubic tensor as a vector in a $d^3$-dimensional space, a linear decomposition attack recovers the shared tensor from the public transcript in polynomial time without recovering either secret action. We also cryptanalyze three natural commuting-subgroup constructions---field-extension, block-diagonal, and tensor-product constructions---and give toy-scale experiments illustrating their specific structural leakage. Finally, we examine the lower-dimensional leakage caused by scaled-block structure. The contribution is therefore a framework proposal together with its cryptanalysis; it does not provide a secure public-key encryption scheme.

cs.CR↗

arXiv2026

A phase-field neural solver for moving contact line problems with dynamic boundary conditions

Phase-field models based on the Cahn--Hilliard equation coupled with dynamic boundary conditions provide a thermodynamically consistent framework for moving contact line (MCL) problems. Although physics-informed neural networks (PINNs) offer a mesh-free approach for solving partial differential equations, their direct application to MCL problems remains challenging due to long-time error accumulation, sharp interfacial profiles, localized contact line dynamics, and complex contact angle evolution. In this work, we propose MCL-PINNs, a specialized phase-field neural solver designed for MCL problems with dynamic boundary conditions. The method is built on a discrete-time formulation and incorporates several key techniques, including a multi-network time-marching scheme, a relaxed distribution constraint on the neural network outputs, variable scaling for sharply varying solution features, adaptive loss weighting, adaptive collocation sampling with interface extraction, and, when applicable, symmetry preservation through neural network inputs. These techniques improve the capability of the neural solver in resolving sharp interfacial profiles and contact line motion. The proposed method is validated through three numerical examples involving droplet coalescence, shear-induced droplet deformation, and dynamic wetting in a heterogeneous channel. The numerical results show that MCL-PINNs significantly improve prediction accuracy and robustness compared with standard PINNs formulations, enabling reliable resolution of complex interfacial evolution and moving contact line dynamics.

math.NA↗

arXiv2026

Electronic manipulation of polar order in electron crystal

When interaction among atoms or ions is strong enough, they often arrange periodically, forming a crystal. The arrangement patterns of atoms or ions can encode information, a concept that has enabled devices such as ferroelectric memories. It has been found that not only atoms or ions but also electrons in condensed matter can crystallize when Coulomb interaction is strong enough. Typical examples are charge-ordered states in solids, where different valences, or different electron numbers, of an ion spontaneously form a spatial pattern on the lattice. In such electron crystals, information is expected to be encoded into the electron-ordering patterns. Here, we demonstrate electronic manipulation and readout of charge-ordering directions in a paramagnetic semiconductor LuFe$_2$O$_4$. By applying current pulses at room temperature, we observed that the non-reciprocal resistivity of LuFe$_2$O$_4$ is modulated along with a sign reversal, which disappears above the charge-ordering temperature. A numerical calculation incorporating inter-band Berry curvature affected by the charge ordering is consistent with the experimental results. By applying the observed phenomenon, we also demonstrate a non-reciprocal resistance memory operation in the charge-ordered LuFe$_2$O$_4$. This result opens the door to realizing charge-ordering electronics.

cond-mat.mtrl-sci↗

arXiv2026

MMIR-TCM: Memory-Integrated Multimodal Inference and Retrieval for TCM Clinical Decision Support

Traditional Chinese Medicine (TCM) diagnosis, particularly through tongue inspection, faces persistent challenges in subjectivity and reproducibility. The application of multimodal artificial intelligence to TCM clinical tasks, such as syndrome differentiation and prescription generation, is significantly hampered by the semantic gap between visual tongue features and textual reasoning, as well as the lack of large-scale, standardized datasets. To address these challenges, we introduce MMIR-TCM, a novel framework that emulates the diagnostic process of TCM experts by integrating multimodal large language model(MLLM) with memory-augmented segmentation and retrieval-augmented generation (RAG). Employing a three-stage architecture, MMIR-TCM integrates a training-free Memory-SAM module for robust tongue extraction, a fine-tuned Qwen3-VL model for structured tongue diagnosis generation, and a Qwen3-based RAG component for evidence-grounded clinical decision support generation. The framework was developed and validated using MedTCM, a new large-scale multimodal dataset that we introduce specifically for advanced TCM research. To properly evaluate our framework's clinical accuracy, which existing metrics fail to capture, we also developed TDEU, a domain-specific evaluation metric incorporating semantic understanding and diagnostic importance. Our comprehensive experiments demonstrate that MMIR-TCM significantly outperforms leading models, including GPT-4o and Gemini 2.5 Flash.

cs.AI↗

arXiv2026

Sketched Linear Contrastive Learning: Approximation, Optimization, and Statistical Scaling

Scaling laws describe how learning performance varies with model size, data size, and compute. While recent theoretical work has established scaling laws for sketched linear regression, much less is understood for contrastive representation learning. In this paper, we study a sketched linear model for contrastive learning under a paired Gaussian latent-variable setup. The learner observes only sketched views of two correlated variables and trains a bilinear contrastive score by full-batch empirical gradient descent. We analyze a Gaussian-negative quadratic contrastive surrogate under aligned power-law spectra and a contrastive source condition, where we derive a risk decomposition into irreducible risk, approximation error, GD bias, GD variance, and a cross term. The cross term is controlled by the bias and variance and therefore does not affect the upper-bound scaling. Our main theorem gives an explicit scaling law with respect to sketch dimension $M$, sample size $N$, and effective optimization horizon $L_{\mathrm{eff}}\gamma$. Compared with standard linear-regression scaling laws, the contrastive setting must learn interactions between two views, and this changes how optimization and finite-sample noise scale with model size, data, and training time. This provides a first theoretical step toward understanding scaling behavior in contrastive learning and gives guidance for balancing model size, data, and optimization compute.

cs.LG↗

arXiv2026

Scaling Laws for Dynamic Mini-Batch SGD in Sketched Linear Regression

Mini-batching is central to large-scale optimization, yet its role in statistical scaling laws remains limited. We study one-pass and multi-pass batch SGD for sketched linear regression under power-law spectral and source conditions. Our analysis reveals a two-horizon phenomenon induced by warmup--stable--decay schedules: deterministic learning is governed by the full optimization trajectory, while stochastic error retains only a shorter terminal memory. For dynamic batch schedules, the individual batch sizes enter through influence-weighted summaries that measure how strongly each update affects the final risk. Consequently, batching leaves the approximation and optimization-bias laws unchanged at a fixed update horizon, but controls the one-pass variance and the multi-pass fluctuation around full-batch gradient descent. We obtain matching one-pass variance bounds and nearly matching multi-pass fluctuation bounds, recover static-batch and full-batch behavior as special cases, and derive an oracle square-root rule for allocating a fixed iteration budget. These results identify WSD horizon separation and final-risk influence as the mechanisms governing dynamic mini-batch scaling.

cs.LG↗

arXiv2026

OSCAR: Offline Spectral Covariance-Aware Rotation for 2-bit KV Cache Quantization

INT2 KV-cache quantization is attractive for long-context LLM serving, but it remains difficult to make both accurate and deployable. Simple rotations such as Hadamard transforms reduce outliers, but still degrade at INT2 because they are not aligned with downstream attention. We propose OSCAR, an Ultra-low-bit KV Cache quantization method that estimates attention-aware covariance structures offline and uses them to derive fixed rotations and clipping thresholds for quantization. In this way, it aligns KV quantization with the covariance structures that attention actually consumes. More importantly, we not only provide theoretical justification but also develop a fully deployable OSCAR system with a custom INT2 attention kernel that remains compatible with paged KV-cache serving and fused kernel pipelines, enabling seamless integration into modern LLM serving frameworks such as SGLang and vLLM. We evaluate our methods on recent reasoning models with reasoning traces of up to 32k tokens across 5 tasks. On Qwen3-4B-Thinking-2507 and Qwen3-8B, OSCAR reduces the BF16 accuracy gap to 3.78 and 1.42 points, respectively, while naive rotation INT2 collapses to nearly zero. We further scale OSCAR to Qwen3-32B and GLM-4.7 (358B params), where it remains effectively on par with BF16. On long context - RULER-NIAH up to 128K, OSCAR remains robust on both Qwen3 models, while naive rotation INT2 collapses. System-wise, OSCAR reduces KV-cache memory by approximately 8x, improves throughput by up to 7x at large batch sizes under the same memory budget, and accelerates batch-size-1 decoding by up to 3x over BF16 due to reduced memory bandwidth overhead.

cs.LG↗

arXiv2026

Beyond Accuracy: Unveiling Inefficiency Patterns in Tool-Integrated Reasoning

In real-world Tool-Integrated Reasoning (TIR) scenarios, where LLMs interleave reasoning with external tool calls, a major source of inefficiency is that the toolcalls create pauses between LLM requests and cause KV-Cache eviction, forcing recomputation. Also, the long, unfiltered response returned by external tools inflates the KV-Cache, so each decode step spends more time loading the growing cache and thus becomes steadily slower as context length increases. However, existing efficiency metrics like token counts and toolcall counts fail to capture the real model inference latency. To address this, we introduce PTE (Prefill Token Equivalents), a hardware-aware TIR-efficiency metric that unifies internal reasoning and external tool-use costs while explicitly accounting for non-reusable KV-Cache and long-tool-response scenarios. Validation in a high-concurrency industrial setting indicates that PTE aligns significantly better with wall-clock latency than standard token counts, while maintaining consistent efficiency rankings across diverse hardware profiles. We conduct extensive experiments across five TIR benchmarks, quantify their PTE costs, and identify four inefficiency patterns that appear in TIR. We also discover that trajectories with higher PTE costs tend to have lower reasoning correctness, indicating that simply using more tools does not improve the quality of the answer.

cs.PF↗

arXiv2026

CARE: Covariance-Aware and Rank-Enhanced Decomposition for Enabling Multi-Head Latent Attention

Converting pretrained attention modules such as grouped-query attention (GQA) into multi-head latent attention (MLA) can improve expressivity without increasing KV-cache cost, making it attractive for efficient inference. However, many practical conversion baselines rely on weight-only low-rank approximations (e.g., SVD-style initializations) and uniform rank allocation. They focus on minimizing the difference between weight matrices rather than on how those weights affect input activations, ignore the covariance structure of activations, and enforce uniform rank across layers, causing activation drift and degraded attention fidelity. To address these issues, we propose CARE, a Covariance-Aware, Rank-Enhanced MLA conversion pipeline under a fixed KV width. CARE introduces three key steps: (i) activation-preserving factorization, which aligns the approximation with the actual input activations rather than just the weights; (ii) adjusted-rank allocation, which spreads a fixed KV budget across layers by giving more capacity to layers that need it most; and (iii) KV-parity mapping, which reparameterizes the converted K and V to fit the MLA format while keeping the KV-cache size unchanged. Our method outperforms a uniform-rank SVD baseline on Qwen3-4B/30B-A3B-Instruct-2507 and Llama-3.1-8B/70B-Instruct, reducing one-shot perplexity by up to 215x and improving mean accuracy by up to 1.70x at matched KV budgets. With a brief post-SVD healing fine-tune, we fully recover the original model's accuracy.

cs.LG↗

arXiv2025

Magnetic electron-hole asymmetry in cuprates: a computational revisit

In this work, we revisit the electron-hole asymmetry of antiferromagnetism in cuprates by studying the three-band Emery model. Using parameters relevant to La$_2$CuO$_4$, we benchmark the anti-ferromagnetic response for a large range of dopings with variational Monte Carlo, determinant quantum Monte Carlo, constrained-path auxiliary-field quantum Monte Carlo, density-matrix embedding theory, and the Gutzwiller approximation. Across methods and accessible sizes/temperatures, we find no significant electron-hole asymmetry if we consider only Neel anti-ferronagnetic response and ignore other possible orders such as stripe state. This result is robust to a moderate oxygen-site repulsion $U_p$ and to parameter sets of Nd$_2$CuO$_4$. Incorporating dopant-induced local potentials reveals an extrinsic route to asymmetry: Cu-site defects enhance AFM on the electron-doped side, whereas O-site defects suppress it on the hole-doped side. These results indicate that dopant-driven effects make a non-negligible contribution to apparent electron-hole asymmetry in the general phase diagram of cuprates and should be included when analyzing competing orders in cuprates.

cond-mat.str-el↗

arXiv2025

From Perception to Reasoning: Deep Thinking Empowers Multimodal Large Language Models

With the remarkable success of Multimodal Large Language Models (MLLMs) in perception tasks, enhancing their complex reasoning capabilities has emerged as a critical research focus. Existing models still suffer from challenges such as opaque reasoning paths and insufficient generalization ability. Chain-of-Thought (CoT) reasoning, which has demonstrated significant efficacy in language models by enhancing reasoning transparency and output interpretability, holds promise for improving model reasoning capabilities when extended to the multimodal domain. This paper provides a systematic review centered on "Multimodal Chain-of-Thought" (MCoT). First, it analyzes the background and theoretical motivations for its inception from the perspectives of technical evolution and task demands. Then, it introduces mainstream MCoT methods from three aspects: CoT paradigms, the post-training stage, and the inference stage, while also analyzing their underlying mechanisms. Furthermore, the paper summarizes existing evaluation benchmarks and metrics, and discusses the application scenarios of MCoT. Finally, it analyzes the challenges currently facing MCoT and provides an outlook on its future research directions.

cs.CL↗

arXiv2025

EndoWave: Rational-Wavelet 4D Gaussian Splatting for Endoscopic Reconstruction

In robot-assisted minimally invasive surgery, accurate 3D reconstruction from endoscopic video is vital for downstream tasks and improved outcomes. However, endoscopic scenarios present unique challenges, including photometric inconsistencies, non-rigid tissue motion, and view-dependent highlights. Most 3DGS-based methods that rely solely on appearance constraints for optimizing 3DGS are often insufficient in this context, as these dynamic visual artifacts can mislead the optimization process and lead to inaccurate reconstructions. To address these limitations, we present EndoWave, a unified spatio-temporal Gaussian Splatting framework by incorporating an optical flow-based geometric constraint and a multi-resolution rational wavelet supervision. First, we adopt a unified spatio-temporal Gaussian representation that directly optimizes primitives in a 4D domain. Second, we propose a geometric constraint derived from optical flow to enhance temporal coherence and effectively constrain the 3D structure of the scene. Third, we propose a multi-resolution rational orthogonal wavelet as a constraint, which can effectively separate the details of the endoscope and enhance the rendering performance. Extensive evaluations on two real surgical datasets, EndoNeRF and StereoMIS, demonstrate that our method EndoWave achieves state-of-the-art reconstruction quality and visual accuracy compared to the baseline method.

cs.CV↗

arXiv2025

Generalizable Hierarchical Skill Learning via Object-Centric Representation

We present Generalizable Hierarchical Skill Learning (GSL), a novel framework for hierarchical policy learning that significantly improves policy generalization and sample efficiency in robot manipulation. One core idea of GSL is to use object-centric skills as an interface that bridges the high-level vision-language model and the low-level visual-motor policy. Specifically, GSL decomposes demonstrations into transferable and object-canonicalized skill primitives using foundation models, ensuring efficient low-level skill learning in the object frame. At test time, the skill-object pairs predicted by the high-level agent are fed to the low-level module, where the inferred canonical actions are mapped back to the world frame for execution. This structured yet flexible design leads to substantial improvements in sample efficiency and generalization of our method across unseen spatial arrangements, object appearances, and task compositions. In simulation, GSL trained with only 3 demonstrations per task outperforms baselines trained with 30 times more data by 15.5 percent on unseen tasks. In real-world experiments, GSL also surpasses the baseline trained with 10 times more data.

cs.RO↗

arXiv2025

Dissecting Embodied Abilities in Multimodal Language Models through Skill-level Evaluation and Diagnosis

Understanding the capability bottlenecks of embodied multimodal large language models (MLLMs) is crucial for improving embodied agents. However, existing embodied benchmarks mainly focus on task-level evaluation and fail to provide actionable insights into the underlying causes of model failures. To address this limitation, we introduce BEAR, a benchmark that decomposes embodied tasks into 14 atomic skills for fine-grained skill-level evaluation. BEAR comprises 4,469 interleaved image-video-text samples spanning 14 skills across 6 categories, ranging from low-level perception to high-level planning. We evaluate 20 MLLMs on BEAR under a hierarchical skill-level diagnosis framework and uncover two key findings: (1) perceptual capabilities are major bottlenecks behind reasoning failures, and (2) current models suffer from unstable spatiotemporal modeling that remains largely unexposed in prior benchmarks. Motivated by these findings, we further propose BEAR-Agent, a multimodal conversational agent that augments MLLMs with visual and spatial reasoning tools. BEAR-Agent substantially improves performance across embodied skills, achieving a relative improvement of 17.5% on GPT-5 over the base model on BEAR, while also outperforming strong baselines in both simulation and real-world robotic experiments. Project page: https://bear-official66.github.io/

cs.CV↗

arXiv2025

Imitate Optimal Policy: Prevail and Induce Action Collapse in Policy Gradient

Policy gradient (PG) methods in reinforcement learning frequently utilize deep neural networks (DNNs) to learn a shared backbone of feature representations used to compute likelihoods in an action selection layer. Numerous studies have been conducted on the convergence and global optima of policy networks, but few have analyzed representational structures of those underlying networks. While training an optimal policy DNN, we observed that under certain constraints, a gentle structure resembling neural collapse, which we refer to as Action Collapse (AC), emerges. This suggests that 1) the state-action activations (i.e. last-layer features) sharing the same optimal actions collapse towards those optimal actions respective mean activations; 2) the variability of activations sharing the same optimal actions converges to zero; 3) the weights of action selection layer and the mean activations collapse to a simplex equiangular tight frame (ETF). Our early work showed those aforementioned constraints to be necessary for these observations. Since the collapsed ETF of optimal policy DNNs maximally separates the pair-wise angles of all actions in the state-action space, we naturally raise a question: can we learn an optimal policy using an ETF structure as a (fixed) target configuration in the action selection layer? Our analytical proof shows that learning activations with a fixed ETF as action selection layer naturally leads to the AC. We thus propose the Action Collapse Policy Gradient (ACPG) method, which accordingly affixes a synthetic ETF as our action selection layer. ACPG induces the policy DNN to produce such an ideal configuration in the action selection layer while remaining optimal. Our experiments across various OpenAI Gym environments demonstrate that our technique can be integrated into any discrete PG methods and lead to favorable reward improvements more quickly and robustly.

cs.LG↗

arXiv2025

Unveiling the landscape of Mottness and its proximity to superconductivity in 4Hb-TaS$_2$

Mott physics is at the root of a plethora of many-body quantum phenomena in quantum materials. Recently, the stacked or twisted structures of van der Waals (vdW) materials have emerged as a unique platform for realizing exotic correlated states in the vicinity of the Mott transition. However, the definitive feature of Mottness and how it rules the low-energy electronic state remain elusive and experimentally inaccessible in many interesting regimes. Here, we quantitatively describe a filling-controlled Mott state and its interplay with superconductivity by scanning tunnelling spectroscopy in a vdW bulk heterostructure, 4Hb-TaS$_2$, that interleaves strongly correlated 1T-TaS$_2$ layers with superconducting 1H-Ta$_2$ layers. The fine tunability of electron doping induced by interlayer charge transfer allows us to continuously track the spectral function with unsurpassed energy resolution from a depleted narrow band (0.2 electrons per site) toward a Mott transition at half filling. The gradually emerging Mott-Hubbard bands, followed by the sharpening and vanishing of the central quasiparticle peak as predicted in the Brinkman-Rice scenario, unambiguously demonstrate the Mott physics at play. Importantly, the renormalization of the low-energy electrons acts destructively on the superconducting pairing potential, leaving behind nonsuperconducting, paramagnetic puddles at the nanoscale. Our results reveal a seminal system near the border of the Mott criterion that enables us to illustrate the predictive power of the Hubbard model, and set such heterostructures as promising ground for realizing new correlated states in the heavily doped Mott regime.

cond-mat.supr-con↗