深度學習於智慧汽車應用
Deep Learning for Autonomous Driving
| 節 | 週一 |
|---|---|
2 09:00–09:50 | 深度學習於智慧汽車應用 ED202（光復） 3 節連堂 |
3 10:10–11:00 | |
4 11:10–12:00 |
* 根據陽明交大上課時間表所列
課程目標:本課程模組目標為學生可自行開發以深度學習方法應用於智慧汽車系統,可訓練學生對於深度學習及智慧車的概念,在未來進一步實際運用於自走車及智慧交通系統等進階教學、實驗課程模組。 課程特色: PBL教學案例: 場景辨識與方向控制,目標95%正確率與30fps即時處理。 課程內容: □機器學習概論 □電腦視覺概論 □Convolutional neural network □Deep learning practices □Recurrent neural network □Object detection: 車輛與行人偵測/交通號誌偵測 □Motion planning: 車輛自動行進控制/ Reverse Reinforcement learning for driving learning 實習課程配合上述授課內容進行實際操作演練;期末專題則根據以上授課與實習內容,分小組進行設計與訓練測試,鼓勵學生以創意方式達成之課程目的。
教師未提供此項資料
無備註
all materials will be at e3.nctu.edu.tw
Lab x6 (60%) Final project (20%) Final exam (20%)
教師未提供此項資料
| 週次 | 主題 |
|---|---|
| 第 1 週 | course introduction 9/11 |
| 第 2 週 | overview of computer vision: image classification 9/18 |
| 第 3 週 | neural network: introduction and back propagation 1st lab 9/25 |
| 第 4 週 | Introduction of convolutional neural network 10/2 |
| 第 5 週 | CNN part I: setup the architecture: activation functions, neural net architecture CNN part II: setup the data and loss: proprocessing, weight initialization, batch normalization, regularization, loss function 10/9 |
| 第 6 週 | CNN part II: setup the data and loss: proprocessing, weight initialization, batch normalization, regularization, loss function CNN part III: learning and evaluation: gradient descent and variants 2nd lab 10/16 |
| 第 7 週 | advanced CNN architecture Object detection: 交通號誌偵測 10/23 |
| 第 8 週 | Guest lecture: Object detection: 車輛與行人偵測 3rd lab 10/30 |
| 第 9 週 | semantic segmentation 11/6 |
| 第 10 週 | recurrent neural network 11/13 |
| 第 11 週 | Motion planning: 車輛自動行進控制/ Reverse Reinforcement learning for driving learning 11/20 |
| 第 12 週 | model simplification and acceleration 11/27 |
| 第 13 週 | model simplification and acceleration 12/4 |
| 第 14 週 | Generative adversarial network and self driving security 12/11 |
| 第 15 週 | Guest lecture: patents for ADAS 12/18 |
| 第 16 週 | Generative adversarial network and self driving security 12/25 |
| 第 17 週 | 邀請演講 1/1 |
| 第 18 週 | 期末專題 1/8 |
Course slides and papers will be our major source. reference: Deep Learning, Ian Goodfellow, Yoshua Bengio and Aaron Courville
- 地點
- ED406
- 時間
- 1 EF, please make an appointment in advance.
- 聯絡方式
- e-mail: tschang@mail.nctu.edu.tw TEL: 03-5731925 (NCTU local extension number: 31925)