國立暨南國際大學

學年

學期

開課號

課號

課程名稱

英文課名

學院

開課系所

班級

任課教師

學分

授課時數

授課類型

課程類別

修別

排課時間

排課地點

全英語授課

任課教師電子郵件信箱

核心能力

職涯類別

教學內涵

程式設計

性別平等

品德教育

人權教育

永續環境

一、課程目標Course objectives

This course serves as the advanced sequel to Machine Learning I, transitioning from classical statistical learning to the architectural design of Deep Learning. Students will build upon foundational concepts, such as model training, classification-regression tasks, and cross-validation, to address high-dimensional data challenges. A primary goal is to bridge the gap between theoretical neural networks and physical hardware implementation. Consequently, this course emphasizes Edge AI deployment using Raspberry Pi platforms for optoelectronic applications.

二、師生晤談時間及地點Instructor office hours

時間

三c,d

地點

科四312

電話

(049)2910960#4903

三、授課方式Teaching approach

The instructional strategy for this course combines theoretical frameworks with intensive technical application. Each module begins with instructor-led lectures utilizing comprehensive PPT presentations to establish a conceptual foundation. These sessions focus on the physical mechanisms of optoelectronic data and the underlying logic of deep learning architectures. Practical development is facilitated through a multi-tiered computing environment. Students will initially perform rapid prototyping and exploratory coding using Google Colab. As model complexity increases, the curriculum transitions to the Computer Center’s GPU-equipped workstations for large-scale training and optimization tasks. A distinctive feature of this course is the integration of Edge AI hardware. Students will engage in specialized laboratory exercises involving Raspberry Pi platforms to explore the constraints of real-time inference. This hands-on approach ensures that students can move beyond simulation to deploy functional models on physical semiconductor sensing systems. Frequent practical sessions are embedded throughout the semester to reinforce technical proficiency. Rather than relying solely on passive listening, students will participate in iterative coding workshops. As a result, participants gain the necessary skills to troubleshoot hardware-software interfaces in professional optoelectronic research settings.

四、評量方式Grading criteria:

The final grade is based on your performance in both theoretical understanding and hands-on implementation:Model Training & Logs (25%):
You need to submit your Google Colab training logs and loss/accuracy curves. We focus on your ability to tune hyperparameters and ensure model convergence.  Hardware Implementation & Rubrics (25%):
This part evaluates your skills in setting up the Raspberry Pi 5. Grading is based on correct hardware connection, model conversion (e.g., ONNX), and real-time inference performance.  Edge AI Performance Analysis (20%):
You will write a brief report comparing model performance on the Cloud vs. Edge. You should analyze the trade-offs between speed, accuracy, and power consumption.  Final Project Demo & Q&A (30%):
At the end of the semester, you will demonstrate a real-time Edge AI application. The grade depends on the technical depth of your project and your ability to answer technical questions during the demo.

五、參考書目Textbook & references:

六、教學進度Course schedule

七、彈性教學

To facilitate a seamless transition into advanced topics, this course includes two weeks of flexible teaching via pre-recorded instructional videos focusing on a comprehensive review of Machine Learning I. These asynchronous modules revisit essential concepts such as statistical regression, classification frameworks, and the mathematical foundations of cross-validation. This format allows students to reinforce their understanding of model evaluation and data preprocessing at a personalized pace before advancing to deep learning implementations. By consolidating these foundational principles, the course ensures that all participants possess the technical baseline required for complex optoelectronic research applications. This approach also provides students with the flexibility to revisit critical classical algorithms whenever necessary throughout the semester.

八、TA協助事項Teaching Assistant tasks

Teaching Assistants are essential for managing the technical and logistical complexity of this practical course. Their primary responsibility is to maintain the functionality of the computing environment and provide real-time support during hardware deployment.

九、備註Remarks

1. Prerequisites: Machine Learning I is a mandatory prerequisite.
2. To ensure access to specialized equipment, enrollment is strictly limited to students within the Department of Applied Materials and Optoelectronic Engineering. This course is not open to students from other departments.
3. Students requesting an add-code or manual registration must contact the instructor via email for prior authorization.
4. To ensure everyone has enough hands-on time and computing power, we have arranged the following support measures:Lab Resource Booking: We provide 5 sets of Raspberry Pi 5 and authorized high-speed Cloud accounts. Please use our online calendar to book your practice sessions in advance to ensure hardware availability.  Flexible Lab Sessions: Besides the Thursday lectures, you can choose a specific time slot on Wednesday afternoons for lab work. This "split-group" system ensures that each student gets direct guidance from the professor or TA.  Technical Support: A dedicated TA will be present during all lab sessions to help with hardware setup and debugging. If the cloud server is unstable, pre-trained models are available so you can continue your work without delay.  Digital Learning Hub: We use a dedicated online group to track your progress and share debugging tips. This keeps everyone connected, even when you are working in different time slots.

十、本課程可培養學生之核心能力與教學活動、核心能力與評量方法之對應表

◎請同學們遵守智慧財產權及不得不法影印。
Course participants should respect intellectual property rights. Illegal copying of copyrighted course materials is strictly prohibited.
◎請任課教師在教學過程中適當引導學生使用正版教科書，並適時提醒或制止學生使用非法影印教科書，或通報學校予以輔導。
Instructors should ensure that students purchase licensed textbooks and preventstudents from using illegally copied texts.

學年

學期

開課號

課號