多媒體資訊學習與安全

Multimedia Information Learning and Security

學期

114-2

學分

3 學分

當期課號

639407

永久課號

AIIT30018

開課單位

智慧計算與科技研究所智慧物聯網產業碩士專班

授課教師

許志仲

校區

歸仁

類別

選修

上課時間表

節	週四
5 13:20–14:10	多媒體資訊學習與安全 CM216 3 節連堂
6 14:20–15:10
7 15:30–16:20

* 根據陽明交大上課時間表所列

概述

This course provides a comprehensive introduction to deep learning (DL) and its applications in multimedia security. The first part (Weeks 1-7) covers fundamental deep learning techniques, including image classification, detection, segmentation, restoration, and multi-dimensional image analysis. The second part (Weeks 8-15) explores security challenges in multimedia processing, such as Deepfake detection, adversarial attacks, multimedia forensics, and privacy-preserving AI techniques like federated learning. [Important] This is an implementation-heavy course. Each week includes hands-on assignments, requiring students to build and evaluate deep learning models. Students should anticipate significant computational workloads and plan their resources accordingly.

先修科目

Linear algebra, calculas

教學方式

Lecture

評分方式

Midterm 10%, Assignment 70%, Final Project 20%

週次計畫

週次	主題
第 1 週	Course Introduction: Overview of objectives, grading, and an introduction to deep learning and multimedia security.
第 2 週	Deep Learning Fundamentals: Neural networks, backpropagation, and core architecture concepts.
第 3 週	Convolutional Neural Networks (CNNs): Image processing and classification applications.
第 4 週	Optimization & Training Techniques: Loss functions, gradient descent, regularization, transfer learning.
第 5 週	Object Detection & Semantic Segmentation: Overview of Faster R-CNN, YOLO, FCN, DeepLabv3+, SAMv2/v2, UniDet, etc.
第 6 週	Image Restoration & Super-Resolution: Introduction to DIP, SRCNN, EDSR, GAN-based, and DRCT restoration.
第 7 週	Foundation Models: Overview of ViT, Swin, DINO, and their applications in computer vision.
第 8 週	Multi-Dimensional Image Analysis: Medical imaging (MRI/CT) and hyperspectral imaging techniques.
第 9 週	Midterm
第 10 週	Deepfake Detection: Methods to identify and prevent Deepfake media.
第 11 週	Adversarial Attacks & Defenses: Generating adversarial examples and implementing defensive strategies.
第 12 週	Multimedia Forensics: Hyperspectral image forensics and security analysis.
第 13 週	Robust Deep Learning: Techniques to improve model reliability and security against adversarial threats.
第 14 週	Federated Learning & Security: Privacy protection, collaborative learning, and distributed security.
第 15 週	Industry Expert Talk: Exploring cutting-edge developments in multimedia security.
第 16 週	Final Project Submission: Students present their research and security solutions.

教科書

n/a

Office Hours

地點: Office 208
時間: 1100am at 208 office
聯絡方式: chihchung [at] nycu.edu.tw