364 research outputs found
Accelerating Deep Reinforcement Learning via Phase-Level Parallelism for Robotics Applications
No full textDeep Reinforcement Learning (DRL) plays a critical role in controlling future intelligent machines like robots and drones. Constantly retrained by newly arriving real-world data, DRL provides optimal autonomous control solutions for adapting to ever-changing environments. However, DRL repeats inference and training that are computationally expensive on resource-constraint mobile/embedded platforms. Even worse, DRL produces a severe hardware underutilization problem due to its unique execution pattern. To overcome the inefficiency of DRL, we propose Train Early Start, a new execution pattern for building the efficient DRL algorithm. Train Early Start parallelizes the inference and training execution, hiding the serialized performance bottleneck and improving the hardware utilization dramatically. Compared to the state-of-the-art mobile SoC, Train Early Start achieves 1.42x speedup and 1.13x energy efficiency.
sj-docx-2-tam-10.1177_17588359211068732 – Supplemental material for Predicting treatment outcomes using <sup>18</sup>F-FDG PET biomarkers in patients with non-small-cell lung cancer receiving chemoimmunotherapy
No full textSupplemental material, sj-docx-2-tam-10.1177_17588359211068732 for Predicting treatment outcomes using 18F-FDG PET biomarkers in patients with non-small-cell lung cancer receiving chemoimmunotherapy by Chang Gon Kim, Sang Hyun Hwang, Kyung Hwan Kim, Hong In Yoon, Hyo Sup Shim, Ji Hyun Lee, Yejeong Han, Beung-Chul Ahn, Min Hee Hong, Hye Ryun Kim, Byoung Chul Cho, Arthur Cho and Sun Min Lim in Therapeutic Advances in Medical Oncology</p
sj-pptx-1-tam-10.1177_17588359211068732 – Supplemental material for Predicting treatment outcomes using <sup>18</sup>F-FDG PET biomarkers in patients with non-small-cell lung cancer receiving chemoimmunotherapy
No full textSupplemental material, sj-pptx-1-tam-10.1177_17588359211068732 for Predicting treatment outcomes using 18F-FDG PET biomarkers in patients with non-small-cell lung cancer receiving chemoimmunotherapy by Chang Gon Kim, Sang Hyun Hwang, Kyung Hwan Kim, Hong In Yoon, Hyo Sup Shim, Ji Hyun Lee, Yejeong Han, Beung-Chul Ahn, Min Hee Hong, Hye Ryun Kim, Byoung Chul Cho, Arthur Cho and Sun Min Lim in Therapeutic Advances in Medical Oncology</p
OptimStore: In-Storage Optimization of Large Scale DNNs with On-Die Processing
No full textTraining deep neural network (DNN) models is a resource-intensive, iterative process. For this reason, nowadays, complex optimizers like Adam are widely adopted as it increases the speed and efficiency of training. These optimizers, however, employ additional variables and raise the memory demand 2× to 3× of model parameters, worsening the memory capacity bottleneck. Moreover, as the size of DNN models is projected to grow even further, it is not practical to assume that the future models will fit in accelerator memory. This has triggered various efforts to offload models to flash-based storage. However, when the model, especially the optimizer, is offloaded to flash, the limited I/O bandwidth severely slows down the overall training process. To this end, we present OptimStore, a solid-state drive (SSD) system with on-die processing (ODP) architectures for gradient descent-based machine learning models. OptimStore accelerates the training process of such large-scale models by processing model optimization in the storage device, specifically inside the flash dies. ODP capability of OptimStore eliminates the heavy data movement over external interconnect and internal flash channels. Overall, OptimStore achieves, on average, a 2.8× speedup and a 3.6× improved energy efficiency in the weight update stage over baseline SSD offloading
200MHz sampling rate를 갖는 Low-glitch energy 10bit D/A converter의 설계
No full text학위논문(석사) - 한국과학기술원 : 전기및전자공학과, 1999.2, [ v, 63 p. ]한국과학기술원 : 전기및전자공학과
Author Correction: Immune cell composition in normal human kidneys (Scientific Reports, (2020), 10, 1, (15678), 10.1038/s41598-020-72821-x)
No full text© 2021, The Author(s).This Article contains an error in Figure 6a, where the data shown does not correlate with the Article. The correct Figure 6a appears below as Figure 1. (Figure presented.).N
Association Between Intimate Partner Violence and Suicidal Ideation Among Female and Male Adults in South Korea: A Nationally Representative Longitudinal Study
No full text© The Author(s) 2022.This study sought to assess the prevalence of intimate partner violence (IPV) victimization and to examine the sex-stratified associations between IPV and suicidal ideation in South Korea. This study used nationally representative longitudinal survey data from 2011 to 2020 of the Korean Welfare Panel Study of 9732 married adults at baseline. Participants were annually asked about IPV experiences of verbal abuse, physical threat, or physical assault over the past year in 2011–2019. Suicidal ideation in the past year was assessed in the following year 2012–2020 of IPV experiences. Using multilevel logistic regression models, we examined the associations between IPV and suicidal ideation among married adults. The prevalence of verbal abuse, physical threat, and physical assault was respectively 19.8%, 3.0%, and 1.4% in females (n = 5056); and 18.8%, 2.7%, and 1.0% in males (n = 4676). After adjusting for potential confounders (e.g., lifetime suicidal ideation and random effects of individuals and households) in the same year when participants had IPV experiences, all types of IPV experiences were statistically associated with suicidal ideation in the following year among both females and males, except for physical threat among males. We found a statistically significant association between physical threat and suicidal ideation among females (adjusted odds ratio: 2.22, 95% confidence interval: 1.62, 3.04), whereas no association was observed among males. Our findings suggest that public health interventions related to suicidal behaviors that can be attributable to experiencing IPV are necessary among both females and males in South Korea.N
Effect of K<sup>+</sup>/Ca<sup>2+</sup> ratios on the uptake level of ionic solution.
No full text<p>Temporal variations in the increase in ionic solutions with three K<sup>+</sup>/Ca<sup>2+</sup> ratios in the excised laurel stems and curve-fitted with 2-D diffusion-convection model. Variations in the increased K<sup>+</sup>/Ca<sup>2+</sup> ratios were quantified using a neutron imaging technique for 64 h. Orange circles denote K<sup>+</sup>/Ca<sup>2+</sup> = 0.01; green rotated triangles denote K<sup>+</sup>/Ca<sup>2+</sup> = 1; purple rectangles denote K<sup>+</sup>/Ca<sup>2+</sup> = 100.</p
- …
