arXiv subjects

Hong Shu

Publications and source records attributed to Hong Shu.

4 recordsLinked to original sources

arXiv2022

One-Class Risk Estimation for One-Class Hyperspectral Image Classification

Hyperspectral imagery (HSI) one-class classification is aimed at identifying a single target class from the HSI by using only knowing positive data, which can significantly reduce the requirements for annotation. However, when one-class classification meets HSI, it is difficult for classifiers to find a balance between the overfitting and underfitting of positive data due to the problems of distribution overlap and distribution imbalance. Although deep learning-based methods are currently the mainstream to overcome distribution overlap in HSI multiclassification, few studies focus on deep learning-based HSI one-class classification. In this article, a weakly supervised deep HSI one-class classifier, namely, HOneCls, is proposed, where a risk estimator,the one-class risk estimator, is particularly introduced to make the fully convolutional neural network (FCN) with the ability of one class classification in the case of distribution imbalance. Extensive experiments (20 tasks in total) were conducted to demonstrate the superiority of the proposed classifier.

cs.CV↗

arXiv2022

The preliminary statistical analysis of LAMOST DR8 low resolution AFGK stars

We download the LAMOST DR8 low resolution catalog 6,478,063 AFGK tpye stars and plot the figures of effective temperature, gravitational acceleration, and metal abundance. Some small and medium mass stars are evolved from pre-main sequence or main sequence stage to planetary nebula stage or white dwarf stage by the stellar evolution code \texttt{MESA}. We analyze the observed statistical data and model calculation results, and then obtain some basic conclusions preliminarily. Most red giant and asymptotic giant stars with log$g$ less than 0.85 have poor metal abundance. Most hot A type main-sequence stars are metal rich stars with log$g$ from 3.5 to 4.5. The conclusions are reasonable within a certain error range. The theory of a gap area in the H-R diagram for stellar evolutions of medium mass stars is reflected in the statistical figures. The central core hydrogen burning stage and the central core helium burning stage correspond to the peak structures in the gravitational acceleration statistical figures respectively. The metal abundances among A, F, G, and K type stars have a wide distribution. We can not simply replace the metal abundances of these stars with the metal abundance of the Sun when doing a fine research work.

astro-ph.SR↗

arXiv2020

Asteroseismology of the DAV star R808

The DAV star R808 was observed by 13 different telescopes for more than 170 hours in April 2008 on the WET run XCOV26. 25 independent pulsation frequencies were identified by this data set. We assumed 19 $m$ = 0 modes and performed an asteroseismological study on those 19 modes. We evolve grids of DAV star models by \texttt{WDEC} adopting the element diffusion scheme with pure and screened Coulomb potential effect. The core compositions are from white dwarf models evolved by \texttt{MESA}, which are thermal nuclear burning results. Our best fitting model is from the screened Coulomb potential scenario, which has parameters of log($M_{\rm He}/M_{\rm *}$) = -2.4, log($M_{\rm H}/M_{\rm *}$) = -5.2, $T_{\rm eff}$ = 11100\,K, $M_{\rm *}$ = 0.710\,$M_{\odot}$, log$g$ = 8.194, and $\sigma_{RMS}$ = 2.86\,s. The value of $\sigma_{RMS}$ is the smallest among the four existing asteroseismological work. The average period spacing is 46.299\,s for $l$ = 1 modes and 25.647\,s for $l$ = 2 modes. The other 6 observed modes can be fitted by $m$ $\neq$ 0 components of some modes for our best fitting model. Fitting the 25 observed modes, we obtain a $\sigma_{RMS}$ value of 2.59\,s. Considering the period spacings, we also assume, that at least in one case, we detect an $l$ = 2 trapped mode.

astro-ph.SR↗

arXiv2020

Driving Tasks Transfer in Deep Reinforcement Learning for Decision-making of Autonomous Vehicles

Knowledge transfer is a promising concept to achieve real-time decision-making for autonomous vehicles. This paper constructs a transfer deep reinforcement learning framework to transform the driving tasks in inter-section environments. The driving missions at the un-signalized intersection are cast into a left turn, right turn, and running straight for automated vehicles. The goal of the autonomous ego vehicle (AEV) is to drive through the intersection situation efficiently and safely. This objective promotes the studied vehicle to increase its speed and avoid crashing other vehicles. The decision-making pol-icy learned from one driving task is transferred and evaluated in another driving mission. Simulation results reveal that the decision-making strategies related to similar tasks are transferable. It indicates that the presented control framework could reduce the time consumption and realize online implementation.

cs.AI↗