For further information, please contact the General chair: Tohru Ishihara
Scalable Logic Architecture for Accelerating Gradient Boosted Tree Learning
Gradient Boosted Tree(GBT) is very efficient supervised learning algorithm. Due to it's availability and performance of efficienccy and accuracy, GBT is widely used for data analysis competition and also for real-world applications such as classification, web search ranking, and recommendation. Recently, learning tasks need to handle Big Data that is frequently updated over time, so training time became important. We proposed the scalable logic architecture to accelerate Gradient Boosted Tree training which can handle from small to big data-sets and a number of hyper parameters.
Tamon Sadasue is a senior research engineer at RICOH Company. He was engaged in research of image processing and software/hardware system architecture. And he has made research outcome implemented to variety of products with technologies of image compression, video processing, photogrammetry. Now he is working with embedded AI technologies.