# 議題 1. 從無到有建立一個機器學習 API 挑選 Kaggle 現有資料集,以表格型資料為主。 資料集: - 共享單車需求預測 - 改善生產線效能 Agenda: - 機器學習建模 - 建置 API: Flask、FastAPI - 模型部署:GCP、AWS、Heroku(free) - 應用串接(擇一):streamlit APP 製作、網頁前後端串屆(HTML、JS) # 議題 2. huggingface 應用 挑選幾個有趣的應用,讓學員了解該領域知識以及如侯使用 pretrain 好的模型直接 inference。 # 議題 3. XAI ## XAI for ML - Interpretable Models (可解釋的模型) - linear model - Decision Tree 透過資料解釋模型 - Global Model-Agnostic Methods (全局模型解釋) - Partial Dependence Plot (PDP) - Local Model-Agnostic Methods - LIME - SHAP  ## XAI for DL - XAI on deep learning - Perturbation-Based - Gradient-Based - Propagation-Based - CAM-Based - Evaluation of explanations - Limits of explanations  # 議題 4. 注意力機制於電腦視覺 ## Attention for CNN - Convolutional Block Attention Module - 通道注意力 (Channel Attention) - 空間注意力 (Spatial Attention) - 通道空間注意力 (Channel & Spatial Attention) - 分支注意力 (Branch Attention) - Object Detection using attention for CNN - YOLOv4 - Prime Sample Attention in Object Detection - Attention-RPN - Segmentation using attention for CNN - Attention U-Net - DANet  ## Transformer for CV - Transformer overview - ViT - DeiT - Swin Transformer  # 議題5. 智慧製造 ## 製程參數優化 - 機器學習建模 - 啟發式演算法:基因演算法最佳參數搜尋搜尋 
×
Sign in
Email
Password
Forgot password
or
By clicking below, you agree to our
terms of service
.
Sign in via Facebook
Sign in via Twitter
Sign in via GitHub
Sign in via Dropbox
Sign in with Wallet
Wallet (
)
Connect another wallet
New to HackMD?
Sign up