arXiv subjects

Ke Wang

Publications and source records attributed to Ke Wang.

At least 91 records · Page 5Linked to original sources

arXiv2026

Ego-Pi: VLA Fine-Tuning for Ego-Centric Human and Robot Data

Robotics faces a fundamental challenge of data scarcity. Unlike language or vision research, there is no internet-scale dataset for robotic manipulation. A promising path forward is to leverage egocentric human data, which can be collected more easily, with greater breadth, and at a larger scale. Towards this end, we investigate key design choices for learning across human and humanoid embodiments equipped with dexterous five-finger hands, using the $π_{0.5}$ model as a foundation. Our results show that human data enables robots to learn new task semantics and compose existing skills into novel behaviors without corresponding robot data. The paper website is here: https://egopipaper.github.io/

cs.RO↗

arXiv2026

Physics-Guided Policy Optimization with Self-Distillation

Self-distilled policy optimization (SDPO) has become a popular paradigm for LLM post-training, where a model learns from its own predictions conditioned on privileged information. SDPO, however, is sensitive to how much each update step should be trusted: corrections from a self-teacher can be highly informative on some batches and misleading on others, and applying them uniformly with a fixed step size can destabilize training. Drawing inspiration from viscous-fluid dynamics and formalizing the analogy at the SDE level, we propose Physics-Guided Policy Optimization (PGPO), which introduces an information-modulated step-size multiplier derived from a mutual-information estimate between the student's predictions and the feedback-conditioned teacher. We show that this modulation preserves the order-1 weak-approximation guarantees of vanilla SGD, and incurs negligible overhead per iteration. We evaluate PGPO on the Science-QA dataset, where it outperforms SDPO on 3 of the 4 domains with gains of up to +4.5 points, while remaining stable in a setting where SDPO collapses late in training.

cs.LG↗

arXiv2026

Hist2Style: Histogram-Guided Stylization with Bilateral Grids

Photorealistic style transfer aims to match the color and tone of an input image to that of a style target while preserving the content and details of the original scene. Although existing large image models can facilitate these kinds of appearance edits, their high computational demands, potential for hallucinations, and limited user control make them unsuitable for high-resolution, real-time workflows. We introduce Hist2Style, a bilateral-grid formulation for fast, edge-aware stylization that preserves visual fidelity by constraining operations to locally affine transforms in bilateral space. Our model distills a large image editing model into a lightweight network by training on a large supervised corpus generated with language and vision-language models, targeting spatially varying color edits. The network conditions on a histogram-based embedding of the style target to provide an interpretable interface for adjusting the output style by modifying the target color distribution. Overall, Hist2Style maintains content structure by construction, avoids hallucinations, and supports real-time, high-resolution photorealistic stylization with interactive user-controllable color and tone adjustments.

cs.CV↗

arXiv2026

Impact of Cu-Mn ratio on Structure and Defects in Layered Multiferroic Cu1-xMn1+ySiTe3

Multiferroic materials exhibit the coexistence of magnetic and ferroelectric order, enabling control of magnetism through electric fields and vice versa. These properties make them attractive for spintronic and memory device applications. Recent studies on Cu1-xMn1+ySiTe3 (0.04 \leq x \leq 0.26; 0.03 \leq y \leq 0.15) have revealed strong magnetoelectric coupling, with variations in Mn-to-Cu concentration leading to variations in optical, electronic, and magnetic responses. Despite these findings, the influence of nanoscale structure and defects on the observed properties remains poorly understood. In this study, we investigate the structure and nanoscale defects in Cu-deficient Cu1-xMn1+ySiTe3 (Cu:Mn ratio <1, i.e., with 0.04 \leq x \leq 0.26 and 0.03 \leq y \leq 0.15) and Cu-rich Cu1+xMn1-ySiTe3 (Cu:Mn ratio >1, i.e., with 0.04 \leq x \leq 0.3 and 0.13 \leq y \leq 0.31) crystals using scanning/transmission electron microscopy and single-crystal X-ray diffraction. Cu-deficient crystals exhibit extensive stacking faults correlated with chemical inhomogeneity between Mn and Cu, along with variations in Te stacking. In contrast, Cu-rich crystals show fewer stacking faults but contain other local structural variations, such as needle-shaped precipitates and loop-like features. These distinct local structural features between Cu-rich and Cu-deficient crystals can be correlated to variations in their observed properties. Complementary density functional theory calculations confirm that the Cu-rich structure is more polar than the Cu-deficient structure. Overall, this study provides a comprehensive understanding of how subtle changes in chemistry influence the nanoscale structure, defect distribution, and functional properties in Cu1-xMn1+ySiTe3, offering guidance for designing multiferroic materials with tailored performance.

cond-mat.mtrl-sci↗

arXiv2026

Finitely generated Howson groups which are not strongly Howson

A group $G$ is called a Howson group if the intersection of any two finitely generated subgroups of $G$ is again finitely generated. It is called strongly Howson if, in addition, the rank of such an intersection is bounded only in terms of the ranks of the two subgroups. Strongly Howson groups are indeed Howson. Recently, Zhang and Zhao showed that the converse fails, by constructing the first examples of infinitely generated Howson groups which are not strongly Howson. In this note, we provide finitely generated examples with and without torsion elements, thereby answering a question posed by Zhang and Zhao.

math.GR↗

arXiv2026

SIREN: Unified Multi-Granularity Semantic Interaction for Multi-Modal Lifelong User Interest Modeling

Industrial recommender systems increasingly leverage lifelong user behavior histories and rich multi-modal content to capture evolving user preferences. However, effectively integrating multi-modal features into lifelong interest modeling remains challenging due to the inherent misalignment between multi-modal and collaborative spaces. Existing paradigms typically rely on separate modeling of multi-modal sequence and behavior sequence, and late fusion to alleviate the modality gap, which results in coarse-grained multi-modal representation and limited integration. In this paper, we propose SIREN, a unified multi-granularity semantic interaction framework for multi-modal lifelong user interest modeling. In the General Search Unit stage, we introduce two alternative retrieval strategies: multi-modal similarity-based soft retrieval for retrieval effectiveness, and Semantic ID (SemID)-based hard retrieval for efficient industrial serving. For the Exact Search Unit stage, we explicitly incorporate target-aware relevance via coarse similarity buckets and fine-grained prefix-encoded SemIDs, enabling unified interaction with collaborative ID features within the target-conditioned transformer architecture. Extensive experiments on the offline dataset demonstrate that SIREN achieves a state-of-the-art GAUC. Online A/B tests further demonstrate consistent GMV gains across multiple production scenarios, including +2.28% in Weixin Moments, +3.87% in Weixin Official Accounts, and +1.61% in Weixin Channels. From July 2025, SIREN has been fully launched for full-traffic serving in Tencent's advertising platform.

cs.IR↗

arXiv2026

SN 2023fyq: direct detection of a Type Ibn supernova progenitor and its multi-wavelength environmental constraints

Context. Type Ibn supernovae (SNe) are characterized by narrow helium emission lines arising from ejecta-circumstellar medium interaction, yet their progenitors remain debated, with both massive Wolf-Rayet stars and low-mass helium stars in binaries proposed. Aims. We aim to directly identify the progenitor of the Type Ibn SN 2023fyq and to characterize its environment in order to constrain the progenitor's nature and evolutionary channel. Methods. We search for the SN progenitor based on pre-explosion and late-time HST and JWST images and derive its properties by fitting the spectral energy distribution. We investigate the SN environment by probing the stars, dust, ionized gas and molecular gas with a multi-wavelength dataset including HST and JWST imaging, VLT/MUSE integral-field-unit spectroscopy and ALMA CO (2--1) radio interferometry. Results. We discover a pre-explosion source at the SN position, which is consistent with a hot ($T>$15000 K) and luminous (log($L$/$L_\odot$) $\gtrsim$ 5.5) SN progenitor and a possible host star cluster. The progenitor is confirmed to have disappeared after explosion. Analysis of the SN environment implies that the progenitor likely has an age of log($t$/yr) = 7.1--7.2. These phenomena disfavor a very massive single-star progenitor and instead support a binary scenario involving a low-mass helium star and a compact object; the observed progenitor emission likely arises from binary interaction that began at least $\sim$12 yr before the explosion. Conclusions. SN 2023fyq is the first Type Ibn SN with a directly detected progenitor and a possible host star cluster. It adds to the diversity of Type Ibn SNe in terms of their progenitor channels and mass-loss mechanisms.

astro-ph.SR↗

arXiv2026

Repetitive Penrose process in Konoplya-Zhidenko rotating non-Kerr black holes

This paper investigates the repetitive Penrose process in Konoplya-Zhidenko rotating non-Kerr black hole, exploring the influence of the deformation parameter on the repetitive Penrose process. After a brief review of the Konoplya-Zhidenko rotating non-Kerr black hole, we study the fundamental equations of the Penrose process in this spacetime, examine the iterative stopping conditions required for the repetitive Penrose process, and obtain the corresponding numerical results. It is concluded that, in addition to previously observed phenomena, under the same decay radius, a larger initial dimensionless deformation parameter $\hatη$ leads to greater values of the energy return on investment and energy utilization efficiency, particularly at higher decay radii. Furthermore, a smaller initial $\hatη$ results in a larger maximum value of the energy return on investment. For energy utilization efficiency, the initial $\hatη$ should take an intermediate value to maximize its peak. Additionally, we find that a larger initial $\hatη$ corresponds to a smaller maximum value of the extracted energy.

gr-qc↗

arXiv2026

RoHIL: Robust Human-in-the-Loop Robotic Reinforcement Learning Against Illumination Variations

Human-in-the-loop reinforcement learning systems achieve near-perfect success on the workstation where they are trained, but collapse when the same robot is moved to a workstation a few meters away due to shifts in the visual input distribution caused by new lamp positions and window light. Re-collecting demonstrations and re-running HIL on every workstation is incompatible with deployment, and naively fine-tuning on shifted-light data triggers catastrophic forgetting of the source workstation. To close this cross-domain gap, we present RoHIL, an offline fine-tuning framework that uses no extra real-robot interaction. RoHIL combines (i) a world-model-based image relighter that re-synthesises the visual stream of source-workstation trajectories under multiple virtual HDRI environments, leaving actions and rewards real; (ii) Illumination-Retention Replay (IRR), a data-level anti-forgetting mechanism that interleaves relit adaptation transitions with original-light retention transitions to preserve source-workstation Bellman coverage; and (iii) an anchored Bellman-actor regulariser that constrains representation and policy drift from the original source-workstation policy. Across four real-robot manipulation tasks under significant cross-workstation illumination variations, RoHIL substantially improves shifted-light performance where standard HIL-RL collapses, while preserving source-workstation performance, eliminating the need to re-collect data and retrain for every new workstation and environment. Project page: https://anonymous4365.github.io/RoHIL/

cs.RO↗

arXiv2026

Quantum Algorithms for Nonlinear Differential Equations via Pivot-Shifted Carleman Linearization

We develop a pivot-shifted Carleman linearization framework for quantum algorithms solving quadratic nonlinear ordinary differential equations. By shifting the dynamics by a pivot state prior to Carleman lifting, and combining this with a Lyapunov transform and rescaling, we enlarge the class of nonlinear systems that can be efficiently simulated on quantum computers. For systems that exhibit stability in the shifted coordinates, we establish long time convergence of the truncated Carleman embedding. We prove that the truncation order scales only logarithmically with the simulation time and target precision, and we derive end-to-end quantum query complexity bounds for preparing a state proportional to the final solution. By introducing a modified nonlinearity condition, this framework entirely removes the conventional lower bound requirement on the initial condition. For more general systems that remain unstable after shifting, we provide short time convergence guarantees that are similarly free from the initial condition constraints. Numerical experiments on the logistic and the Lotka-Volterra equations demonstrate that an appropriate pivot choice improves stability and accuracy, and yields exponential error decay with truncation order. These results show that pivot shifting provides a practical and theoretically justified route for extending Carleman-based quantum algorithms to a broader class of nonlinear dynamical systems.

quant-ph↗

arXiv2026

Structured Labeling Enables Faster Vision-Language Models for End-to-End Autonomous Driving

Vision-Language Models (VLMs) offer a promising approach to end-to-end autonomous driving due to their human-like reasoning capabilities. However, troublesome gaps remains between current VLMs and real-world autonomous driving applications. One major limitation is that existing datasets with loosely formatted language descriptions are not machine-friendly and may introduce redundancy. Additionally, high computational cost and massive scale of VLMs hinder the inference speed and real-world deployment. To bridge the gap, this paper introduces a structured and concise benchmark dataset, NuScenes-S, which is derived from the NuScenes dataset and contains machine-friendly structured representations. Moreover, we present FastDrive, a compact VLM baseline with 0.9B parameters. In contrast to existing VLMs with over 7B parameters and unstructured language processing(e.g., LLaVA-1.5), FastDrive understands structured and concise descriptions and generates machine-friendly driving decisions with high efficiency. Extensive experiments show that FastDrive achieves competitive performance on structured dataset, with approximately 20% accuracy improvement on decision-making tasks, while surpassing massive parameter baseline in inference speed with over 10x speedup. Additionally, ablation studies further focus on the impact of scene annotations (e.g., weather, time of day) on decision-making tasks, demonstrating their importance on decision-making tasks in autonomous driving.

cs.CV↗

arXiv2026

Edit-Based Refinement for Parallel Masked Diffusion Language Models

Masked diffusion language models enable parallel token generation and offer improved decoding efficiency over autoregressive models. However, their performance degrades significantly when generating multiple tokens simultaneously, due to a mismatch between token-level training objectives and joint sequence consistency. In this paper, we propose ME-DLM, an edit-based refinement framework that augments diffusion generation with lightweight post-editing steps. After producing an initial complete response, the model refines it through minimal edit operations, including replacement, deletion, and insertion, conditioned on the full sequence. Training supervision is derived from edit distance, providing a deterministic signal under a fixed canonicalization scheme for learning minimal corrections. This approach encourages sequence-level consistency through globally conditioned edits while preserving the efficiency benefits of parallel diffusion decoding. Extensive experiments demonstrate that ME-DLM improves the quality and robustness of multi-token parallel generation. In particular, when built upon LLaDA, our method achieves consistent gains of 11.6 points on HumanEval and 33.6 points on GSM8K while using one-eighth of the total diffusion steps. Code is available at https://github.com/renhouxing/ME-DLM.

cs.CL↗

arXiv2026

Analytic Solution for the Motion of Spinning Particles in Plane Gravitational Wave Spacetime

The interaction between spin and gravitational waves causes spinning bodies to deviate from their geodesics. In this work, we obtain the analytic solution of the Mathisson--Papapetrou--Dixon equations at linear order in the spin for plane gravitational wave spacetimes. Our approach combines a parallel-transported tetrad with the translational Killing symmetries of plane wave spacetimes, yielding six conserved quantities that fully determine the momentum, spin evolution, and worldline. The resulting transverse and longitudinal motions are expressed in closed form as single integrals of the retarded time, providing a unified and model-independent framework for computing spin--curvature-induced deviations. This analytic solution offers a versatile tool for studying spin-dependent effects in gravitational memory, Penrose-limit geometries, and high-energy scattering regimes.

gr-qc↗

arXiv2026

R3-VAE: Reference Vector-Guided Rating Residual Quantization VAE for Generative Recommendation

Generative Recommendation (GR) has gained traction for its merits of superior performance and cold-start capability. As the vital role in GR, Semantic Identifiers (SIDs) represent item semantics through discrete tokens. However, current techniques for SID generation based on vector quantization face two main challenges: (i) training instability, stemming from insufficient gradient propagation through the straight-through estimator and sensitivity to initialization; and (ii) inefficient SID quality assessment, where industrial practice still depends on costly GR training and A/B testing. To address these challenges, we propose Reference Vector-Guided Rating Residual Quantization VAE (R3-VAE). This framework incorporates three key innovations: (i) a reference vector that functions as a semantic anchor for the initial features, thereby mitigating sensitivity to initialization; (ii) a dot product-based rating mechanism designed to stabilize the training process and prevent codebook collapse; and (iii) two SID evaluation metrics, Semantic Cohesion and Preference Discrimination, serving as regularization terms during training. Empirical results on six benchmarks demonstrate that R3-VAE outperforms state-of-the-art methods, achieving an average improvement of 14.5% in Recall@10 and 15.5% in NDCG@10 across three public datasets (Beauty, Sports, and Toys). Furthermore, we perform GR training and online A/B tests on Toutiao. Our method achieves a 1.62% improvement in MRR and a 0.83% gain in StayTime/U versus baselines. Additionally, we employ R3-VAE to replace the item ID of CTR model, resulting in significant improvements in content cold start by 15.36%, corroborating the strong applicability and business value in industry-scale recommendation scenarios.

cs.IR↗

arXiv2026

Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Training

Have you ever post-trained a generalist vision-language-action (VLA) policy on a small demonstration dataset, only to find that it stops responding to new instructions and is limited to behaviors observed during post-training? We identify this phenomenon as lock-in: after low-data, supervised fine-tuning (SFT), the policy becomes overly specialized to the post-training data and fails to generalize to novel instructions, manifesting as concept lock-in (fixation on training objects/attributes) and spatial lock-in (fixation on training spatial targets). Many existing remedies introduce additional supervision signals, such as those derived from foundation models or auxiliary objectives, or rely on augmented datasets to recover generalization. In this paper, we show that the policy's internal pre-trained knowledge is sufficient: DeLock mitigates lock-in by preserving visual grounding during post-training and applying test-time contrastive prompt guidance to steer the policy's denoising dynamics according to novel instructions. Across eight simulation and real-world evaluations, DeLock consistently outperforms strong baselines and matches or exceeds the performance of a state-of-the-art generalist policy post-trained with substantially more curated demonstrations.

cs.RO↗

arXiv2026

MOA: Multi-Objective Alignment for Role-Playing Agents

Role-playing agents (RPAs) require balancing multiple objectives, such as instruction following, persona consistency, and stylistic fidelity, which are not always perfectly aligned across different dimensions. While prior work has primarily relied on supervised fine-tuning or reinforcement learning with scalarized rewards, these approaches do not explicitly address the coordination of multiple reward dimensions during optimization. We present \textbf{MOA} (\textbf{M}ulti-\textbf{O}bjective \textbf{A}lignment), a reinforcement-learning framework that enables multi-dimensional, fine-grained rubric optimization for general RPAs. MOA introduces a novel multi-objective optimization strategy that trains simultaneously on multiple fine-grained rubrics to boost optimization performance. Additionally, to improve both output diversity and generation quality, we employ thought-augmented rollouts with off-policy guidance. Experiments on PersonaGym and RoleMRC show that MOA consistently improves multi-dimensional role-playing performance over supervised and standard RL baselines. Under identical evaluation protocols, an 8B model trained with MOA reaches performance competitive with strong closed-source models across multiple evaluation dimensions. These results suggest that MOA provides a practical framework for training more capable general-purpose role-playing agents.

cs.CL↗

arXiv2026

PhotoFramer: Multi-modal Image Composition Instruction

Composition matters during the photo-taking process, yet many casual users struggle to frame well-composed images. To provide composition guidance, we introduce PhotoFramer, a multi-modal composition instruction framework. Given a poorly composed image, PhotoFramer first describes how to improve the composition in natural language and then generates a well-composed example image. To train such a model, we curate a large-scale dataset. Inspired by how humans take photos, we organize composition guidance into a hierarchy of sub-tasks: shift, zoom-in, and view-change tasks. Shift and zoom-in data are sampled from existing cropping datasets, while view-change data are obtained via a two-stage pipeline. First, we sample pairs with varying viewpoints from multi-view datasets, and train a degradation model to transform well-composed photos into poorly composed ones. Second, we apply this degradation model to expert-taken photos to synthesize poor images to form training pairs. Using this dataset, we finetune a model that jointly processes and generates both text and images, enabling actionable textual guidance with illustrative examples. Extensive experiments demonstrate that textual instructions effectively steer image composition, and coupling them with exemplars yields consistent improvements over exemplar-only baselines. PhotoFramer offers a practical step toward composition assistants that make expert photographic priors accessible to everyday users.

cs.CV↗

arXiv2026

TabEmb: Joint Semantic-Structure Embedding for Table Annotation

Table annotation is crucial for making web and enterprise tables usable in downstream NLP applications. Unlike textual data where learning semantically rich token or sentence embeddings often suffice, tables are structured combinations of columns wherein useful representations must jointly capture column's semantics and the inter-column relationships. Existing models learn by linearizing the 2D table into a 1D token sequence and encoding it with pretrained language models (PLMs) such as BERT. However, this leads to limited semantic quality and weaker generalization to unseen or rare values compared to modern LLMs, and degraded structural modeling due to 2D-to-1D flattening and context-length constraints. We propose TabEmb, which directly targets these limitations by decoupling semantic encoding from structural modeling. An LLM first produces semantically rich embeddings for each column, and a graph-based module over columns then injects relationships into the embeddings, yielding joint semantic-tructural representations for table annotation. Experiments show that TabEmb consistently outperforms strong baselines on different table annotation tasks. Source code and datasets are available at https://github.com/hoseinzadeehsan/TabEmb

cs.LG↗