arXiv subjects

Hongsuk Yi

Publications and source records attributed to Hongsuk Yi.

3 recordsLinked to original sources

arXiv2020

StRDAN: Synthetic-to-Real Domain Adaptation Network for Vehicle Re-Identification

Vehicle re-identification aims to obtain the same vehicles from vehicle images. This is challenging but essential for analyzing and predicting traffic flow in the city. Although deep learning methods have achieved enormous progress for this task, their large data requirement is a critical shortcoming. Therefore, we propose a synthetic-to-real domain adaptation network (StRDAN) framework, which can be trained with inexpensive large-scale synthetic and real data to improve performance. The StRDAN training method combines domain adaptation and semi-supervised learning methods and their associated losses. StRDAN offers significant improvement over the baseline model, which can only be trained using real data, for VeRi and CityFlow-ReID datasets, achieving 3.1% and 12.9% improved mean average precision, respectively.

cs.CV↗

arXiv1999

Suppression of ferromagnetic ordering in doped manganites: Effects of the superexchange interaction

From a Monte Carlo study of the ferromagnetic Kondo lattice model for doped manganites, including the antiferromagnetic superexchange interaction ($J_{AF}$), we found that the ferromagnetic ordering was suppressed as $J_{AF}$ increased. The ferromagnetic transition temperature $T_c$, as obtained from a mean field fit to the calculated susceptibilities, was found to decrease monotonically with increasing $J_{AF}$. Further, the suppression in $T_c$ scales with the bandwidth narrowing induced by the antiferromagnetic frustration originating from $J_{AF}$. From these results, we propose that the change in the superexchange interaction strength between the $t_{2g}$ electrons of the Mn ions is one of the mechanisms responsible for the suppression in $T_c$ observed in manganites of the type (La$_{0.7-y}$Pr$_{y}$)Ca$_{0.3}$MnO$_3$.

cond-mat.str-el↗

arXiv1999

Anomalous spin susceptibility and magnetic polaron formation in the double exchange systems

The magnetic susceptibility and spin-spin correlation of the double-exchange model for doped manganites are investigated through the Monte Carlo calculations on the three-dimensional lattice model. Deviations of the susceptibility from the Curie-Weiss behavior above the ferromagnetic ordering temperature $T_c$ seem to indicate a formation of local ferromagnetic clusters in the vicinity of $T_c$, which is consistent with recent electron paramagnetic resonance experiments for La$_{2/3}$Ca$_{1/3}$MnO$_3$. A further analysis of the spin-spin correlations show the ferromagnetic cluster size to be three-to-four lattice spacings, suggesting that the charge carriers may form magnetic polarons.

cond-mat.str-el↗