QLQuansheng Liu

Quansheng Liu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Quansheng Liu has a total of 22 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Contrastive Learning With Multiple Prototypes for Unsupervised Domain Adaptive Semantic Segmentation

    by Jun Yu, Guochen Xie, Quansheng Liu, Zhen Kan, Lei Wang, Tianyu Liu, Qiang Ling, Wei Xu, Fang Gao

    Unsupervised domain adaptive semantic segmentation aims to transfer knowledge from the annotated source domain to the unlabeled target domain. Recently, self-training methods have gained substantial attention, which leverage high-confidence predictions in the target domain as pseudo labels for supervision. However, limited exploration of intra-class variations across domains, including significant visual differences within each category, has led to misalignment between feature distribution across domains. In this paper, we present a unified non-parametric distance-based online clustering method to efficiently maintain multiple centroid-based prototypes within each category subspace instead of one prototype for each category subspace, which enables prototypes to possess the capacity for richer feature representation. Then, considering the variance across different dimensions of a feature representation, we then extend the prototypes from centroid-based ones to distribution-based ones. Specifically, each subspace is modeled using a Gaussian mixture model which includes several anisotropic Gaussian distributions, aimed at prioritizing discriminative dimensions and obtaining a finer measurement of the pixel-to-prototype similarity. Meanwhile, a category-aware feature space is achieved through pixel-to-prototype contrastive learning to ensure the compactness of pixel features in the same subcategory and drive the separation between pixel features of different subcategories. What's more, multi-resolution features are utilized to promote diversity and robustness among intra-class prototypes. Experiments validate the competitiveness of our two prototype-based methods against existing state-of-the-art methods, with a mIoU of $76.8\%$ on GTA $\rightarrow$ Cityscapes, $68.4\%$ on Synthia $\rightarrow$ Cityscapes, $54.5\%$ on Cityscapes $\rightarrow$ DarkZurich and $56.4\%$ on Cityscapes $\rightarrow$ ACDC. Notably, our method is able to seamlessly integrate with existing UDA methods.


  • Open Access

    Towards fully ab initio simulation of atmospheric aerosol nucleation

    by Shuai Jiang, Yi-Rong Liu, Teng Huang, Ya-Juan Feng, Chun-Yu Wang, Zhong-Quan Wang, Bin-Jing Ge, Quan-Sheng Liu, Wei-Ran Guang, Wei Huang

    Atmospheric aerosol nucleation contributes to approximately half of the worldwide cloud condensation nuclei. Despite the importance of climate, detailed nucleation mechanisms are still poorly understood. Understanding aerosol nucleation dynamics is hindered by the nonreactivity of force fields (FFs) and high computational costs due to the rare event nature of aerosol nucleation. Developing reactive FFs for nucleation systems is even more challenging than developing covalently bonded materials because of the wide size range and high dimensional characteristics of noncovalent hydrogen bonding bridging clusters. Here, we propose a general workflow that is also applicable to other systems to train an accurate reactive FF based on a deep neural network (DNN) and further bridge DNN-FF-based molecular dynamics (MD) with a cluster kinetics model based on Poisson distributions of reactive events to overcome the high computational costs of direct MD. We found that previously reported acid-base formation rates tend to be significantly underestimated, especially in polluted environments, emphasizing that acid-base nucleation observed in multiple environments should be revisited.


  • Open Access

    Berry–Esseen bound and precise moderate deviations for products of random matrices

    by Hui Xiao, Ion Grama, Quansheng Liu

    Let {(g_{n})_{n\geq 1}} be a sequence of independent and identically distributed (i.i.d.) {d\times d} real random matrices. For {n\geq 1} set {G_n = g_n \ldots g_1} . Given any starting point {x=\mathbb R v\in\mathbb{P}^{d-1}} , consider the Markov chain {X_n^x = \mathbb R G_n v } on the projective space {\mathbb P^{d-1}} and define the norm cocycle by {\sigma(G_n, x)= \log (|G_n v|/|v|)} , for an arbitrary norm {|\cdot|} on \smash{\mathbb R^{d}} . Under suitable conditions we prove a Berry–Esseen-type theorem and an Edgeworth expansion for the couple {(X_n^x, \sigma(G_n, x))} . These results are established using a brand new smoothing inequality on complex plane, the saddle point method and additional spectral gap properties of the transfer operator related to the Markov chain {X_n^x} . Cramér-type moderate deviation expansions as well as a local limit theorem with moderate deviations are proved for the couple {(X_n^x, \sigma(G_n, x))} with a target function {\varphi} on the Markov chain {X_n^x} .


  • Open Access

    Convergence theorems for the Non-Local Means filter

    by Qiyu Jin, Ion Grama, Quansheng Liu

    We introduce an oracle filter for removing the Gaussian noise with weights depending on a similarity function. The usual Non-Local Means filter is obtained from this oracle filter by substituting the similarity function by an estimator based on similarity patches. When the sizes of the search window are chosen appropriately, it is shown that the oracle filter converges with the optimal rate. The same optimal convergence rate is preserved when the similarity function has suitable errors-in measurements. We also provide a statistical estimator of the similarity which converges at a convenient rate. Based on our convergence theorems, we propose some simple formulas for the choice of the parameters. Simulation results show that our choice of parameters improves the restoration quality of the filter compared with the usual choice of parameters in the original algorithm.