Abstract

Non-volatile memory (NVM) based computing systems have gained significant attention in recent years due to their low power consumption and fast access times. However, building machine learning algorithms on such systems requires careful consideration of their unique properties. In this talk, we focus on the popular machine learning algorithm, random forest, and discuss how to efficiently build it on NVM-based computing systems. We will first introduce the challenges of building random forest on NVM-based systems, including the limited write endurance of NVM and the high memory access costs. We will demonstrate the effectiveness of our approach through experiments on several benchmark datasets. Our results show that our proposed approach can achieve a significant improvement in both energy consumption and performance over existing methods.

Bio.

Tseng-Yi Chen has received his Ph.D degree in Computer Science from National Tsing Hua University. Currently, Dr. Chen is an assistant professor at Department of Computer Science and Information Engineering, National Central University. His research interests lay in embedded and large-scale storage system designs, emerging memory and storage technologies, non-volatile memory systems for machine learning. Dr. Chen has published more than 40 research papers , which were mainly published in prestige journals (e.g., IEEE TC, IEEE TVLSI, IEEE TCAD, ACM TECS, ACM TODAES, and ACM TOS) and premier conferences (e.g., ACM/IEEE DAC, ACM/IEEE ISLPED, and ACM/IEEE CODES+ISSS). His works received best paper awards from a premier conference (i.e., ACM/IEEE ISLPED 2020) and an important conference (i.e., IEEE NVMSA 2019).

All faculty and students are welcome to join.