arXiv subjects

Weixiao Wang

Publications and source records attributed to Weixiao Wang.

5 recordsLinked to original sources

arXiv2026

Training-Free Token-Level Steering for LLM Personalized Co-Writing

While Large Language Models (LLMs) show great promise for personalization, they often lack specialized domain knowledge. Conventional solutions like fine-tuning struggle with high computational costs and rapid data updates, while Retrieval-Augmented Generation fails to provide fine-grained, token-level steering. Furthermore, chat-based interfaces remain dominant, whereas productive co-writing paradigms have not yet been well exploited beyond the coding domain. To this end, we introduce SteerWrite, a training-free framework designed for personalized co-writing. Our method effectively adapts the base model to specialized domains without gradient updates, with specific designs tailored to small datasets. Experiments demonstrate that SteerWrite achieves state-of-the-art performance across diverse datasets, metrics, and models, significantly reducing human editing effort.

cs.CL↗

arXiv2023

Vicinal Feature Statistics Augmentation for Federated 3D Medical Volume Segmentation

Federated learning (FL) enables multiple client medical institutes collaboratively train a deep learning (DL) model with privacy protection. However, the performance of FL can be constrained by the limited availability of labeled data in small institutes and the heterogeneous (i.e., non-i.i.d.) data distribution across institutes. Though data augmentation has been a proven technique to boost the generalization capabilities of conventional centralized DL as a "free lunch", its application in FL is largely underexplored. Notably, constrained by costly labeling, 3D medical segmentation generally relies on data augmentation. In this work, we aim to develop a vicinal feature-level data augmentation (VFDA) scheme to efficiently alleviate the local feature shift and facilitate collaborative training for privacy-aware FL segmentation. We take both the inner- and inter-institute divergence into consideration, without the need for cross-institute transfer of raw data or their mixup. Specifically, we exploit the batch-wise feature statistics (e.g., mean and standard deviation) in each institute to abstractly represent the discrepancy of data, and model each feature statistic probabilistically via a Gaussian prototype, with the mean corresponding to the original statistic and the variance quantifying the augmentation scope. From the vicinal risk minimization perspective, novel feature statistics can be drawn from the Gaussian distribution to fulfill augmentation. The variance is explicitly derived by the data bias in each individual institute and the underlying feature statistics characterized by all participating institutes. The added-on VFDA consistently yielded marked improvements over six advanced FL methods on both 3D brain tumor and cardiac segmentation.

eess.IV↗

arXiv2022

Magnetically driven accretion disc winds: the role of gas thermodynamics and comparison to ultra-fast outflows

Winds are commonly observed in luminous active galactic nuclei (AGNs). A plausible model of those winds is magnetohydrodynamic (MHD) disc winds. In the case of disc winds from a thin accretion disc, isothermal or adiabatic assumption is usually adopted in such MHD models. In this work we perform two-dimensional MHD simulations implementing different thermal treatments (isothermal, adiabatic and radiative) to study their effects on winds from a thin accretion disc. We find that both the isothermal model and the adiabatic model overestimate the temperature, underestimate the power of disc winds, and cannot predict the local structure of the winds, compared to the results obtained by solving the energy equation with radiative cooling and heating. Based on the model with radiative cooling and heating, the ionization parameter, the column density and the velocity of the disc winds have been compared to the observed ultrafast outflows (UFOs). We find that in our simulations the UFOs can only be produced inside hundreds of Schwarzschild radius. At much larger radii, no UFOs are found. Thus, the pure MHD winds cannot interpret all the observed UFOs.

astro-ph.HE↗

arXiv2022

Multimodality in Meta-Learning: A Comprehensive Survey

Meta-learning has gained wide popularity as a training framework that is more data-efficient than traditional machine learning methods. However, its generalization ability in complex task distributions, such as multimodal tasks, has not been thoroughly studied. Recently, some studies on multimodality-based meta-learning have emerged. This survey provides a comprehensive overview of the multimodality-based meta-learning landscape in terms of the methodologies and applications. We first formalize the definition of meta-learning in multimodality, along with the research challenges in this growing field, such as how to enrich the input in few-shot learning (FSL) or zero-shot learning (ZSL) in multimodal scenarios and how to generalize the models to new tasks. We then propose a new taxonomy to discuss typical meta-learning algorithms in multimodal tasks systematically. We investigate the contributions of related papers and summarize them by our taxonomy. Finally, we propose potential research directions for this promising field.

cs.LG↗

arXiv2022

Magnetic Spirals in Accretion Flows Originated from Misaligned Magnetic Field

Misalignment between rotation and magnetic field has been suggested to be one type of physical mechanisms which can easen the effects of magnetic braking during collapse of cloud cores leading to formation of protostellar disks. However, its essential factors are poorly understood. Therefore, we perform a more detailed analysis of the physics involved. We analyze existing simulation data to measure the system torques, mass accretion rates and Toomre Q parameters. We also examine the presence of shocks in the system. While advective torques are generally the strongest, we find that magnetic and gravitational torques can play substantial roles in how angular momentum is transferred during the disk formation process. Magnetic torques can shape the accretion flows, creating two-armed magnetized inflow spirals aligned with the magnetic field. We find evidence of an accretion shock that is aligned according to the spiral structure of the system. Inclusion of ambipolar diffusion as explored in this work has shown a slight influence in the small scale structures but not in the main morphology. We discuss potential candidate systems where some of these phenomena could be present.

astro-ph.SR↗