Hongsuk Yi

Deep Neural Networks for traffic flow prediction

Traffic flow prediction is an essential function of traffic information systems. Conventional approaches, using artificial neural networks with narrow network architecture and poor training samples for supervised learning, have been only partially successful. In this paper, a deep-learning neural-network based on TensorFlow™ is suggested for the prediction traffic flow conditions, using real-time traffic data. Until now, no research has applied the TensorFlow™ deep learning neural network model to the estimation of traffic conditions. The suggested supervised model is trained by a deep learning algorithm, which uses real traffic data aggregated every five minutes. Results demonstrate that the models accuracy rate is around 99%.

Atomic and electronic structures of doped grain boundaries in SrTiO3

The atomic and electronic structures of pristine, Mn- and Nb-doped grain boundaries in SrTiO3 are investigated by atomistic simulations and cluster calculations. The atomic structures of (310) symmetric tilt grain boundaries in SrTiO3 are determined by atomistic simulation using empirical potentials. The defect energies of Mn(Nb)-doped models are calculated and discussed in relation to the concentration profiles of Mn(Nb) in SrTiO3 grain boundaries. The local electronic structures near Mn(Nb)-doped grain boundaries in SrTiO3 are determined using embedded cluster calculations based on the density functional theory. The charge density of each system is calculated to elucidate the electronic structure of the grain boundary. The calculation results agree well with previous experimental observations of the atomic structures and grain boundary charges near the Mn(Nb)-doped grain boundary in SrTiO3.

Performance evaluation of GAIA supercomputer using NPB multi-zone benchmarks

GAIA is a recently developed IBM POWER6 supercomputer consisting of 24 SMP compute nodes with 64-way processors each, and it is currently ranked 393 on the Top500 supercomputer list published in November 2009. In this paper, we present the performance characteristics of GAIA evaluated by interconnecting the 24 computing nodes with low-latency InfiniBand network. We evaluate the performance of the new dual-core Power 595 system in terms of increased problem size using the multi-zone versions of the NAS Parallel Benchmarks.