Searcharxiv⌕ Search

arXiv subjects

Chunjie Wang

Publications and source records attributed to Chunjie Wang.

27 records · Page 2Linked to original sources
arXiv2024

OpenSUN3D: 1st Workshop Challenge on Open-Vocabulary 3D Scene Understanding

This report provides an overview of the challenge hosted at the OpenSUN3D Workshop on Open-Vocabulary 3D Scene Understanding held in conjunction with ICCV 2023. The goal of this workshop series is to provide a platform for exploration and discussion of open-vocabulary 3D scene understanding tasks, including but not limited to segmentation, detection and mapping. We provide an overview of the challenge hosted at the workshop, present the challenge dataset, the evaluation methodology, and brief descriptions of the winning methods. For additional details, please see https://opensun3d.github.io/index_iccv23.html.

cs.CV↗
arXiv2022

GraphMAE: Self-Supervised Masked Graph Autoencoders

Self-supervised learning (SSL) has been extensively explored in recent years. Particularly, generative SSL has seen emerging success in natural language processing and other AI fields, such as the wide adoption of BERT and GPT. Despite this, contrastive learning-which heavily relies on structural data augmentation and complicated training strategies-has been the dominant approach in graph SSL, while the progress of generative SSL on graphs, especially graph autoencoders (GAEs), has thus far not reached the potential as promised in other fields. In this paper, we identify and examine the issues that negatively impact the development of GAEs, including their reconstruction objective, training robustness, and error metric. We present a masked graph autoencoder GraphMAE that mitigates these issues for generative self-supervised graph pretraining. Instead of reconstructing graph structures, we propose to focus on feature reconstruction with both a masking strategy and scaled cosine error that benefit the robust training of GraphMAE. We conduct extensive experiments on 21 public datasets for three different graph learning tasks. The results manifest that GraphMAE-a simple graph autoencoder with careful designs-can consistently generate outperformance over both contrastive and generative state-of-the-art baselines. This study provides an understanding of graph autoencoders and demonstrates the potential of generative self-supervised pre-training on graphs.

cs.LG↗
arXiv2022

One-Stage Deep Edge Detection Based on Dense-Scale Feature Fusion and Pixel-Level Imbalance Learning

Edge detection, a basic task in the field of computer vision, is an important preprocessing operation for the recognition and understanding of a visual scene. In conventional models, the edge image generated is ambiguous, and the edge lines are also very thick, which typically necessitates the use of non-maximum suppression (NMS) and morphological thinning operations to generate clear and thin edge images. In this paper, we aim to propose a one-stage neural network model that can generate high-quality edge images without postprocessing. The proposed model adopts a classic encoder-decoder framework in which a pre-trained neural model is used as the encoder and a multi-feature-fusion mechanism that merges the features of each level with each other functions as a learnable decoder. Further, we propose a new loss function that addresses the pixel-level imbalance in the edge image by suppressing the false positive (FP) edge information near the true positive (TP) edge and the false negative (FN) non-edge. The results of experiments conducted on several benchmark datasets indicate that the proposed method achieves state-of-the-art results without using NMS and morphological thinning operations.

cs.CV↗
arXiv2022

Kinetic investigation of the planar Multipole Resonance Probe under arbitrary pressure

Active plasma resonance spectroscopy (APRS) refers to a class of plasma diagnostic methods that use the ability of plasma to resonate at or near the electron plasma frequency for diagnostic purposes. The planar multipole resonance probe (pMRP) is an optimized realization of APRS. It has a non-invasive structure and allows simultaneous measurement of the electron density, electron temperature, and electron-neutral collision frequency. Previous work has investigated the pMRP through the Drude model and collision-less kinetic model. The Drude model misses important kinetic effects such as collision-less kinetic damping. The collision-less kinetic model is able to capture pure kinetic effects. However, it is only applicable to low-pressure plasma. To further study the behavior of the pMRP, we develop a collisional kinetic model in this paper, which applies to arbitrary pressure. In this model, the kinetic equation is coupled to the Poisson equation under the electrostatic approximation. The real part of the general admittance is calculated to describe the spectral response of the probe-plasma system. Both collision-less kinetic damping and collisional damping appear in the spectrum. This model provides a possibility to calculate the electron density, electron temperature, and electron-neutral collision frequency from the measurements.

physics.plasm-ph↗
arXiv2021

Kinetic investigation of the planar Multipole Resonance Probe in the low-pressure plasma

Active Plasma Resonance Spectroscopy (APRS) is a well-established plasma diagnostic method: a radio frequency signal is coupled into the plasma via a probe or antenna, excites it to oscillate, and the response is evaluated through a mathematical model. The majority of APRS probes are invasive and perturb the plasma by their physical presence. The planar Multipole Resonance Probe (pMRP) solves this problem: it can be integrated into the chamber wall and minimizes the perturbation. Previous work has studied the pMRP in the frame of the Drude model, but it misses important effects like collision-less damping. In this work, a collision-less kinetic model is developed to further investigate the behavior of the pMRP. This model consists of the Vlasov equation, which is coupled with the Poisson equation under electrostatic approximation. The spectral response of the probe-plasma system is found by calculating the complex admittance. This model covers the kinetic effects and overcomes the limitations of the Drude model.

physics.plasm-ph↗
arXiv2013

Gaussian Integral Means of Entire Functions

For an entire mapping $f:\mathbb C\mapsto\mathbb C$ and a triple $(p,α, r)\in (0,\infty)\times(-\infty,\infty)\times(0,\infty]$, the Gaussian integral means of $f$ (with respect to the area measure $dA$) is defined by $$ {\mathsf M}_{p,α}(f,r)=\Big({\int_{|z|<r}e^{-α|z|^2}dA(z)}\Big)^{-1}{\int_{|z|<r}|f(z)|^p{e^{-α|z|^2}}dA(z)}. $$ Via deriving a maximum principle for ${\mathsf M}_{p,α}(f,r)$, we establish not only Fock-Sobolev trace inequalities associated with ${\mathsf M}_{p,p/2}(z^m f(z),\infty)$ (as $m=0,1,2,...$), but also convexities of $r\mapsto\ln {\mathsf M}_{p,α}(z^m,r)$ and $r\mapsto {\mathsf M}_{2,α<0}(f,r)$ in $\ln r$ with $0<r<\infty$.

math.CV↗
arXiv2011

Logarithmic convexity of integral means for analytic functions

We show that the $L^2$ integral mean on $r\D$ of an analytic function in the unit disk $\D$ with respect to the weighted area measure $(1-|z|^2)^α\,dA(z)$, where $-3\leα\le0$, is a logarithmically convex function of $r$ on $(0,1)$. We also show that the range $[-3,0]$ for $α$ is best possible.

math.CV↗