arXiv subjects

Yue Zhong

Publications and source records attributed to Yue Zhong.

24 records · Page 2Linked to original sources

arXiv2021

Deep Sketch-Based Modeling: Tips and Tricks

Deep image-based modeling received lots of attention in recent years, yet the parallel problem of sketch-based modeling has only been briefly studied, often as a potential application. In this work, for the first time, we identify the main differences between sketch and image inputs: (i) style variance, (ii) imprecise perspective, and (iii) sparsity. We discuss why each of these differences can pose a challenge, and even make a certain class of image-based methods inapplicable. We study alternative solutions to address each of the difference. By doing so, we drive out a few important insights: (i) sparsity commonly results in an incorrect prediction of foreground versus background, (ii) diversity of human styles, if not taken into account, can lead to very poor generalization properties, and finally (iii) unless a dedicated sketching interface is used, one can not expect sketches to match a perspective of a fixed viewpoint. Finally, we compare a set of representative deep single-image modeling solutions and show how their performance can be improved to tackle sketch input by taking into consideration the identified critical differences.

cs.CV↗

arXiv2020

Event-Independent Network for Polyphonic Sound Event Localization and Detection

Polyphonic sound event localization and detection is not only detecting what sound events are happening but localizing corresponding sound sources. This series of tasks was first introduced in DCASE 2019 Task 3. In 2020, the sound event localization and detection task introduces additional challenges in moving sound sources and overlapping-event cases, which include two events of the same type with two different direction-of-arrival (DoA) angles. In this paper, a novel event-independent network for polyphonic sound event localization and detection is proposed. Unlike the two-stage method we proposed in DCASE 2019 Task 3, this new network is fully end-to-end. Inputs to the network are first-order Ambisonics (FOA) time-domain signals, which are then fed into a 1-D convolutional layer to extract acoustic features. The network is then split into two parallel branches. The first branch is for sound event detection (SED), and the second branch is for DoA estimation. There are three types of predictions from the network, SED predictions, DoA predictions, and event activity detection (EAD) predictions that are used to combine the SED and DoA features for on-set and off-set estimation. All of these predictions have the format of two tracks indicating that there are at most two overlapping events. Within each track, there could be at most one event happening. This architecture introduces a problem of track permutation. To address this problem, a frame-level permutation invariant training method is used. Experimental results show that the proposed method can detect polyphonic sound events and their corresponding DoAs. Its performance on the Task 3 dataset is greatly increased as compared with that of the baseline method.

eess.AS↗

arXiv2019

The instability of a black hole with $f(R)$ global monopole under extended uncertainty principle

We consider the evolution of black hole involving an $f(R)$ global monopole based on the Extended Uncertainty Principle (EUP). The black hole evolutions refer to the instability due to the Parikh-Kraus-Wilczeck tunneling radiation or fragmentation. It is found that the EUP corrections make the entropy difference larger to encourage the black hole to radiate more greatly. We also show that the appearance of the EUP effects result in the black hole's division. The influence from global monopole and the revision of general relativity can also adjust the black hole evolution simultaneously, but can not change the final result that the black hole will not be stable because of the EUP's effects.

hep-th↗

arXiv2018

Diagnosing the magnetic field structure of a coronal cavity observed during the 2017 total solar eclipse

We present an investigation of a coronal cavity observed above the western limb in the coronal red line Fe X 6374 Å using a telescope of Peking University and in the green line Fe XIV 5303 Å using a telescope of Yunnan Observatories, Chinese Academy of Sciences during the total solar eclipse on 2017 August 21. A series of magnetic field models are constructed based on the magnetograms taken by the Helioseismic and Magnetic Imager onboard the Solar Dynamics Observatory (SDO) one week before the eclipse. The model field lines are then compared with coronal structures seen in images taken by the Atmospheric Imaging Assembly on board SDO and in our coronal red line images. The best-fit model consists of a flux rope with a twist angle of 3.1$π$, which is consistent with the most probable value of the total twist angle of interplanetary flux ropes observed at 1 AU. Linear polarization of the Fe XIII 10747 Å line calculated from this model shows a "lagomorphic" signature that is also observed by the Coronal Multichannel Polarimeter of the High Altitude Observatory. We also find a ring-shaped structure in the line-of-sight velocity of Fe XIII 10747 Å, which implies hot plasma flows along a helical magnetic field structure, in the cavity. These results suggest that the magnetic structure of the cavity is a highly twisted flux rope, which may erupt eventually. The temperature structure of the cavity has also been investigated using the intensity ratio of Fe XIII 10747 Å and Fe X 6374 Å.

astro-ph.SR↗

arXiv2018

The further analytical discussions on the $U(1)$ gauged Q-balls with $N$-power potential

We discuss the $U(1)$ gauged Q-balls with $N$-power potential to examine their properties analytically. More numerical descriptions and some analytical consideration have been contributed to the models governed by four-power potential. We also demonstrate strictly some new limitations that the stable $U(1)$ gauged Q-balls should accept instead of estimating those with only some specific values of model variables numerically. Having derived the explicit expressions of radius, the Noether charge and energy of the gauged Q-balls, we find that these models under the potential of matter field with general power and the boundary conditions will exist instead of dispersing and decaying. The Noether charge of the large gauged Q-balls must be limited. The mass parameter of the model can not be tiny.

hep-th↗

arXiv2015

The virial relation for the Q-balls in the thermal logarithmic potential revisited analytically

We study the properties of Q-balls dominated by the thermal logarithmic potential analytically instead of estimating the characters with only some specific values of model variables numerically. In particular the analytical expressions for radius and energy of this kind of Q-ball are obtained. According to these explicit expressions we demonstrate strictly that the large Q-balls enlarge and the small ones become smaller in the background with lower temperature. The energy per unit charge will not be divergent if the charge is enormous. We find that the lower temperature will lead the energy per unit charge of Q-ball smaller. We also prove rigorously the necessary conditions that the model parameters should satisfy to keep the stability of the Q-balls. When one of model parameters of Q-balls $K$ is positive, the Q-balls will not form or survive unless the temperature is high enough. In the case of negative $K$, the Q-balls are stable no matter the temperature is high or low.

hep-th↗