國立中正大學數學系

暨應用數學碩士班、統計科學碩士班

學    術   演    講

Exploring Machine Learning Applications: The How and Why, From Industry to Healthcare and Scientific Research

呂秉澤 博士

The Oden Institute for Computational Engineering & Sciences, University of Texas at Austin

Abstract

Over the past two decades, machine learning has thrived, leading to a wide range of applications that have greatly improved the efficiency of routine tasks and opened a new era of technological innovation. In this talk, I will focus on two main topics related to the application and analysis of machine learning and deep learning techniques.

The first part will introduce random forest classification and regression, highlighting how these approaches enhance efficiency in industrial product design and support doctors in detecting bacterial infections. I will share a solution I implemented for a tech company that reduced 30x calibration time for their products. In addition, I will demonstrate how we can leverage medical expertise to extract patients’ vital signs and create a suggestion system to assist doctors in making a preliminary diagnosis—reducing the process from 48 hours to under a second.

In the second part, I will introduce and analyze a neural network architecture called neural ODE, which was proposed in 2018 and received over 5000 citations. This model incorporates numerical methods to learn dynamical systems for various purposes, including gaining insight into the mechanics of human movement for robotics. However, a qualitative analysis is necessary to identify the feasible design space for neural ODE. I will present our findings, showing how the choice of numerical methods influences learning outcomes and propose an optimal selection to guide the construction of effective neural networks.

日  期:113年12月18日(星期三) 12:10-13:00

google meet 網址: https://meet.google.com/cfs-rbzj-nyv

歡迎參加   敬請公佈