Source author record

Dongchan Kim

Dongchan Kim appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Catalog footprint

What is connected

2works
5topics
4close collaborators

Actions

Connect this record

Log in to claim

Research graph

See the researcher in context

Open full explorer

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

2 published item(s)

preprint2020arXiv

Multi-Head Attention based Probabilistic Vehicle Trajectory Prediction

This paper presents online-capable deep learning model for probabilistic vehicle trajectory prediction. We propose a simple encoder-decoder architecture based on multi-head attention. The proposed model generates the distribution of the predicted trajectories for multiple vehicles in parallel. Our approach to model the interactions can learn to attend to a few influential vehicles in an unsupervised manner, which can improve the interpretability of the network. The experiments using naturalistic trajectories at highway show the clear improvement in terms of positional error on both longitudinal and lateral direction.

preprint2014arXiv

Mid-Infrared Properties of Luminous Infrared Galaxies II: Probing the Dust and Gas Physics of the GOALS Sample

The Great Observatories All-Sky LIRG Survey (GOALS) is a comprehensive, multiwavelength study of luminous infrared galaxies (LIRGs) in the local universe. Here we present the results of a multi-component, spectral decomposition analysis of the low resolution mid-IR Spitzer IRS spectra from 5-38um of 244 LIRG nuclei. The detailed fits and high quality spectra allow for characterization of the individual PAH features, warm molecular hydrogen emission, and optical depths for silicate dust grains and water ices. We find that starbursting LIRGs, which make up the majority of GOALS, are very consistent in their MIR properties (i.e. tau_9.7um, tau_ice, neon line and PAH feature ratios). However, as their PAH EQW decreases, usually an indicator of an increasingly dominant AGN, LIRGs cover a larger spread in these MIR parameters. The contribution from PAHs to the total L(IR) in LIRGs varies from 2-29% and LIRGs prior to their first encounter show higher L(PAH)/L(IR) ratios on average. We observe a correlation between the strength of the starburst (IR8) and the PAH fraction at 8um but not with the 7.7 to 11.3 PAH ratio, suggesting the fractional PDR emission, and not the overall grain properties, is associated with the rise in IR8 for galaxies off the starburst main sequence. We detect crystalline silicate features in 6% of the sample but only in the most obscured sources (s_9.7um < -1.24). Ice absorption features are observed in 11% (56%) of GOALS LIRGs (ULIRGs). Most GOALS LIRGs have L(H2)/L(PAH) ratios elevated above those observed for normal star-forming galaxies and exhibit a trend for increasing L(H2)/L(PAH) ratio with increasing L(H2). While star formation appears to be the dominant process responsible for exciting the H2 in most of the GOALS galaxies, a subset of LIRGs (10%) show excess H2 emission that is inconsistent with PDR models and may be excited by shocks or AGN-induced outflows.