publications

2024

1. 

   Stochastic Approximation with Biased MCMC for Expectation-Maximization

   Samuel Gruffaz,  Kyurae Kim, Alain Durmus, and Jacob R Gardner.

   In Proceedings of the International Conference on Artificial Intelligence and Machine Learning (AISTATS) (to be presented) May 2024
2. Linear Convergence of Black-Box Variational Inference: Should We Stick the Landing?

   Kyurae Kim, Yi-An Ma, and Jacob R. Gardner.

   In Proceedings of the International Conference on Artificial Intelligence and Machine Learning (AISTATS) (to be presented) May 2024

   Link

2023

1. 

   On the Convergence of Black-Box Variational Inference

   Kyurae Kim, Jisu Oh, Kaiwen Wu, Yi-An Ma, and Jacob R. Gardner.

   In Advances in Neural Information Processing Systems (to be presented) Dec 2023

   Link

2. 

   The Behavior and Convergence of Local Bayesian Optimization

   Kaiwen Wu,  Kyurae Kim, Roman Garnett, and Jacob R. Gardner.

   In Advances in Neural Information Processing Systems (to be presented) Dec 2023

   Link
3. Automatic calculation of myocardial perfusion reserve using deep learning with uncertainty quantification

   Yoon-Chul Kim,  Kyurae Kim, and Yeon Hyeon Choe.

   Quantitative Imaging in Medicine and Surgery Oct 2023

4. 

   Practical and Matching Gradient Variance Bounds for Black-Box Variational Bayesian Inference

   Kyurae Kim, Kaiwen Wu, Jisu Oh, and Jacob R. Gardner.

   In Proceedings of the International Conference on Machine Learning (ICML) Jul 2023

   Link

2022

1. 

   Markov Chain Score Ascent: A Unifying Framework of Variational Inference with Markovian Gradients

   Kyurae Kim, Jisu Oh, Jacob R. Gardner, Adji Bousso Dieng, and Hongseok Kim.

   In Advances in Neural Information Processing Systems Jul 2022

   Link Code

2021

1. 

   Fast Calculation Software for Modified Look-Locker Inversion Recovery (MOLLI) T1 Mapping

   Yoon-Chul Kim,  Kyurae Kim, Hyelee Lee, and Yeon Hyeon Choe.

   BMC Medical Imaging Jul 2021

   Abs Link

   Abstract Background The purpose of this study was to develop a software tool and evaluate different T1 map calculation methods in terms of computation time in cardiac magnetic resonance imaging. Methods The modified Look-Locker inversion recovery (MOLLI) sequence was used to acquire multiple inversion time (TI) images for pre- and post-contrast T1 mapping. The T1 map calculation involved pixel-wise curve fitting based on the T1 relaxation model. A variety of methods were evaluated using data from 30 subjects for computational efficiency: MRmap, python Levenberg–Marquardt (LM), python reduced-dimension (RD) non-linear least square, C++ single- and multi-core LM, and C++ single- and multi-core RD. Results Median (interquartile range) computation time was 126 s (98–141) for the publicly available software MRmap, 261 s (249–282) for python LM, 77 s (74–80) for python RD, 3.4 s (3.1–3.6) for C++ multi-core LM, and 1.9 s (1.9–2.0) for C++ multi-core RD. The fastest C++ multi-core RD and the publicly available MRmap showed good agreement of myocardial T1 values, resulting in 95% Bland–Altman limits of agreement of (- 0.83 to 0.58 ms) and (- 6.57 to 7.36 ms) with mean differences of - 0.13 ms and 0.39 ms, for the pre- and post-contrast, respectively. Conclusion The C++ multi-core RD was the fastest method on a regular eight-core personal computer for pre- or post-contrast T1 map calculation. The presented software tool (fT1fit) facilitated rapid T1 map and extracellular volume fraction map calculations.

2. 

   A GPU Scheduling Framework to Accelerate Hyper-Parameter Optimization in Deep Learning Clusters

   Jaewon Son, Yonghyuk Yoo,  Kyurae Kim, Youngjae Kim, Kwonyong Lee, and Sungyong Park.

   Electronics Jul 2021

   Abs Link

   This paper proposes Hermes, a container-based preemptive GPU scheduling framework for accelerating hyper-parameter optimization in deep learning (DL) clusters. Hermes accelerates hyper-parameter optimization by time-sharing between DL jobs and prioritizing jobs with more promising hyper-parameter combinations. Hermes’s scheduling policy is grounded on the observation that good hyper-parameter combinations converge quickly in the early phases of training. By giving higher priority to fast-converging containers, Hermes’s GPU preemption mechanism can accelerate training. This enables users to find optimal hyper-parameters faster without losing the progress of a container. We have implemented Hermes over Kubernetes and compared its performance against existing scheduling frameworks. Experiments show that Hermes reduces the time for hyper-parameter optimization up to 4.04 times against previously proposed scheduling policies such as FIFO, round-robin (RR), and SLAQ, with minimal time-sharing overhead.

2020

1. 

   A Probabilistic Machine Learning Approach to Scheduling Parallel Loops with Bayesian Optimization

   Kyurae Kim, Youngjae Kim, and Sungyong Park.

   IEEE Transactions on Parallel and Distributed Systems Jul 2020

   Link Code

2. 

   Automatic Myocardial Segmentation in Dynamic Contrast Enhanced Perfusion MRI Using Monte Carlo Dropout in an Encoder-Decoder Convolutional Neural Network

   Yoon-Chul Kim,  Kyurae Kim, and Yeon Hyeon Choe.

   Computer Methods and Programs in Biomedicine Jul 2020

   Link

2019

1. 

   EVCMR: A Tool for the Quantitative Evaluation and Visualization of Cardiac MRI Data

   Yoon-Chul Kim,  Kyurae Kim, Kwanghee Choi, Minwoo Kim, Younjoon Chung, and Yeon Hyeon Choe.

   Computers in Biology and Medicine Jul 2019

   Link
2. Towards Robust Data-Driven Parallel Loop Scheduling Using Bayesian Optimization

   Kyurae Kim, Youngjae Kim, and Sungyong Park.

   In Proceedings of the IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems Jul 2019

   Link Code