Machine Learning (AI) and Semiconductors
24th July 2024, 2:00 p.m., UTAR Kampar, Block E, EDK5
Lead Staff : Dr. Nuraidayani
Prof. Jen-Shiun Chiang delivered an insightful lecture on the intersection of machine learning and semiconductors, highlighting how AI technologies are revolutionizing various fields. The lecture explored applications of machine learning, its core principles, and its implications for the semiconductor industry.
Prof. Chiang began by discussing the broad range of AI applications, including image processing, audio and speech recognition, medical diagnostics, defense systems, agriculture, gaming, and lifestyle enhancements. He explained that AI, primarily driven by machine learning, involves teaching machines to learn from data, similar to neurons in a neural network.
Machine learning seeks to identify functions based on examples, such as speech and image recognition, or strategic games like Go. These functions can be categorized into regression (predicting continuous outcomes) and classification (categorizing discrete items).
Structured learning was illustrated through an example of predicting future YouTube video views. Prof. Chiang explained how machine learning models use functions with unknown parameters, define a loss function based on training data, and optimize to find the best fit.
He noted that linear models are often too simplistic for complex data, necessitating more sophisticated models like piecewise linear curves and sigmoid functions. This led to discussions on neural networks with fully connected layers and the use of matrices to represent these connections.
The lecture delved into the transition from sigmoid activation functions to ReLU (Rectified Linear Unit) functions, which are crucial for deep learning models. Prof. Chiang highlighted popular deep learning architectures such as AlexNet, VGG, and GoogleNet.
Prof. Chiang addressed several challenges in machine learning, particularly the need for large models and extensive datasets to achieve high accuracy. He stressed the importance of balancing model complexity with computational efficiency and the ongoing need for innovation in this rapidly evolving field.
Prof. Chiang's lecture introduced machine learning and provided an overview of how it is transforming various industries, including semiconductors.
A total of 68 undergraduate students attended the talk.