Qualcomm AI Training Webinar // 高通AI線上訓練課程
近年AI已進入人類的生活，跨領域的應用也蓬勃興起。為協助台灣產業因應AI帶來的挑戰，台北市電腦商業同業公會(TCA)與Qualcomm高通公司合作辦理AI訓練課程，針對台灣中小企業與新創團隊提供一系列的三天的AI免費線上訓練課程，2021年12月21日至12月23日，推廣AI技術的基本知識應用，協助台灣中小企業與新創企業領先於這波先進技術中的浪潮！
課程資訊
時間：9:30AM to 5:30PM，2021年12月21日-12月23日(共三天)
對象：中小企業、新創團隊
費用：免費
語言：中文
主辦單位：Qualcomm高通公司
協辦單位：台北市電腦商業同業公會
聯絡人：台北市電腦公會 02-25774249 分機847 羅小姐
註1. 主辦單位將審核您的報名，您需收到報名確認信才算報名成功。主辦單位會再寄發線上課程網址。未入選學員不另行通知。
註2. 本次共開放35個名額參加，主辦單位保有學員篩選與培訓內容調整之權利。
註3. 學員需至少參與1次報到點名，並填寫問卷，
註4. 報名時請務必填寫您的中文姓名、公司名稱以及Email。
12月21-12月23日 三日議程:
| Date | Agenda | Time | Duration | |
| Dec. 21st Day 1 | Opening Session | Opening Speech delivered by Qualcomm | 9:30-9:35 am | 5 mins |
| Agenda Brief | 9:35-9:40 am | 5 mins | ||
| 1.0 Self-introduction by teachers and students | Teachers introduce themselves | 9:40-10:00 am | 20 mins | |
| Attendees give briefs of who they are and what they are working on. | ||||
| 1.1 Introduction to AI | 1.1.1 AI: What? How? Where? | 10:00 - 10:30 am | 30 mins | |
| 1.1.2 Qualcomm AI | ||||
| 1.1.3 AI vs. Machine learning vs. Deep Learning | ||||
| 1.1.4 Different types of Machine Learning | ||||
| 1.1.5 Basic concepts in Machine Learning & Deep Learning | ||||
| 1.2 Quick Tour of Deep Learning | 1.2.1 From ML to DL: What is deep learning? | 10:30 - 11:10 am | 20 mins | |
| 1.2.2 Datasets of Deep learning: From public to custom | 20 mins | |||
| Tea break 11:10 - 11:20 am | ||||
| 1.2.3 Infrastructure of Deep learning: From hardware to software | 11:20 - 12:00 pm | 40 mins | ||
| Lunch 12:00-1:00 pm | ||||
| Q&A, Open Discussion 1:00-1:30 pm | ||||
| 1.2.4 History, Present and Future | 1:30 - 2:30 pm | 10 mins | ||
| 10 mins | ||||
| 20 mins | ||||
| 20 mins | ||||
| Tea break 2:30 - 2:40 p.m. | ||||
| 1.3.1 Review | Review | 2:40-2:50 pm | 10 mins | |
| 1.3.2 Model Conversion and Demo | Hardware Preparation: AI Kit | 2:50-3:00 pm | 10 mins | |
| SoftWare Preparation: Part 1 SNPE SDK Development Environment Setup | 3:00-3:25 pm | 25 mins | ||
| SoftWare Preparation: Part 2 SNPE Application Development Tools | 3:25-3:45 pm | 20 mins | ||
| Tea break 3:45 - 3:55 pm | ||||
| AI Demos - Object Detector Demo Converting Model | 3:55-4:15 pm | 20 mins | ||
| AI Demos - Object Detector Run Demo | 4:15-4:30 pm | 15 mins | ||
| 1.3.3 Solve AI Problem Skills | Solve AI Problem Skills | 4:30-4:40 pm | 10 mins | |
1.3.4 Homework, Q & A | Homework, Q & A | 4:40 pm~ | - | |
| Date | Agenda | Time | Duration | |
| Dec. 22nd Day2 | 2.1 Foundation of Deep Learning | 2.1.1 Perceptron & Multilayer Perceptron | 9:30 - 10:00 am | 10 mins |
| 10 mins | ||||
| 2.1.2 Basic neuron layers | 10:00 - 10:30 am | 20 mins | ||
| 2.1.3 Loss functions | 10:30 - 11:00 am | 15 mins | ||
| 2.1.4 Optimizer | 20 mins | |||
| 10 mins | ||||
| 2.1.5 Prevent Over-fitting in Deep Learning | 11:00 - 12:00 pm | 20 mins | ||
| 30 mins | ||||
| Lunch 12:00-13:00 pm | ||||
| Q&A, Open Discussion 13:00-13:30 pm | ||||
| 2.2 Building Deep Learning Model | 2.2.1 Classic models | 1:30 - 2:20 pm | 10 mins | |
| 30 mins | ||||
| 10 mins | ||||
| 2.2.2 Hyper-parameters & Tuning Tricks | 2:20 - 2:50 pm | 10 mins | ||
| 10 mins | ||||
| 2.2.3 Fine-tune & Transfer Learnin | 10 mins | |||
| 2.2.4 Data pre-process & Data augmentation | 10 mins | |||
| 2.2.5 Hands-on - Keras_MNIST | 2:50-3:00 pm | 10 mins | ||
| Tea break 3:00 - 3:15 p.m | ||||
| 2.3.1 Review | Review | 3:15-3:55 pm | 10 mins | |
| 2.3.2 SNPE Training Part 2 | An Image Classifiers Demo | 10 mins | ||
| SNPE Introduction | 10 mins | |||
| SNPE Workflow | 10 mins | |||
| Supported Chipsets / Supported Network Layers | 3:55-4:40 pm | 5 mins | ||
| User-defined Operations (UDO) Workflow | 30 mins | |||
| Limitations | 5 mins | |||
| CPU vs GPU vs DSP | 5 mins | |||
| Tea break 4:40 - 4:50 p.m | ||||
| Run SNPE on Linux Machine | 4:50-5:20 pm | 10 mins | ||
| Building and Running the C++ Application on ARM Android | 10 mins | |||
| Thermal Measurement | 5 mins | |||
| Solve Problem with SNPE | 5 mins | |||
| 2.3.3 Q & A | Q & A | 5:20 pm~ | - | |
| Date | Agenda | Time | Duration | |
Dec. 23rd Day 3 | 3.1 Getting Started with TensorFlow | 3.1.1 TensorFlow Overview | 9:30 - 10:30 am | 10 mins |
| 3.1.2 Low level API | 10 mins | |||
| 3.1.3 Middle level API | 10 mins | |||
| 3.1.4 High level API: Keras | 10 mins | |||
| 20 mins | ||||
| 3.2 Basic Knowledge of Object Detection | 3.2.1 Overview | 10:30 - 11:00 am | 10 mins | |
| 3.2.2 Performance metrics | 20 mins | |||
| 3.2.3 Traditional methods for Object Detection | 11:00 - 12:00 pm | 20 mins | ||
| 3.2.4 Two-stage detection | 20 mins | |||
| 3.2.5 One-stage detection | 20 mins | |||
| Lunch 12:00-13:00 pm | ||||
| Q&A, Open Discussion 13:00-13:30 pm | ||||
| 3.2.6 Hands-on Object Detection | 1:30 - 2:30 p.m | 60 mins | ||
| 3.3 SNPE Training Part 3 | Review | 2:30 - 3:00 p.m | 5 mins | |
| SNPE Benchmarking | 25 mins | |||
| AI demos – Object Detector Demo | 3:00 - 3:15 p.m | 15 mins | ||
| Part 1 : Get Camera Stream by Camera2 | ||||
| Part 2 : JNI Interface | ||||
| Part 3 : Integrating SNPE | ||||
| Tea break 3:15 - 3:25p.m | ||||
| AI demos – Face Recognition | 3:25 - 4:00 pm | 25 mins | ||
| Solve SNPE Problem Skills | 10 mins | |||
| Q & A, Others | 4:00 pm ~ | - | ||
