數據所(Institute of Data Science)
Artificial intelligence (deep learning) has achieved great success in many fields. For example, image recognition, medical image analysis, and self-driving cars. Each application has a specific neural network model and a learning strategy in the field of deep learning. This course will explore the deep learning architecture details and use computer vision-related applications as examples to introduce learning strategies. This course contains not only these essential expositions of deep Learning but also many practical programming tasks. The goal is to give students the ability to build a corresponding deep neural network based on their problems.
Artificial intelligence (deep learning) has achieved great success in many fields. For example, image recognition, medical image analysis, and self-driving cars. Each application has a specific neural network model and a learning strategy in the field of deep learning. This course will explore the deep learning architecture details and use computer vision-related applications as examples to introduce learning strategies. This course contains not only these essential expositions of deep Learning but also many practical programming tasks. The goal is to give students the ability to build a corresponding deep neural network based on their problems.


週次 Week進度說明 Progress Description
1Course Introduction
2Machine learning: The data-driven approach
3Loss Functions and Optimization
4Introduction to Neural Networks / Multilayer Perceptron
5Convolutional Neural Networks
6Intro to Pytorch and Tensorflow under Ubuntu System
7CNN Architectures: AlexNet, VGG / Final Project Proposal due
8Training Neural Networks
9In-class midterm
10CNN Architectures: GoogLeNet, ResNet, DenseNet, EfficientNet, etc
11Proposal presentation
12Object Detection
13Semantic Segmentation
14Unsupervised Learning: Generative Models
15Recurrent Neural Networks
16Semi-Supervised learning
17Invited talk
18Final Project Presentation