###### tags: `主頁`

AIAcademy 課程共筆
===

### [主頁](https://hackmd.io/@aia-alumini-notes/AIA_Co-Note)
### [介紹](https://hackmd.io/cpFLWFfIQrilaxnJdBA3yw)
### [PrimeHub上常見問題](https://hackmd.io/uJ117wOOSKymphgC55IG5g)

## Machine Learning

### 宏觀問題
- [Regression and Classification](https://hackmd.io/YxCwYXrCSN60cGDiKO-cqg?view)
- (徵求中)[How Good Is a ML Model?]()
- **Data Pre-processing**
- Categorical Feature Encoding
- [One-hot Encoding](https://hackmd.io/kUiWFjktS1OXquv7RvM8RQ)
- (徵求中)[Encode by Number]()
- (徵求中)[Sin/Cosine Encoding]()
- Numerical Feature Engineering
- Normalization
- (徵求中)[Scaling and bias]()
- (徵求中)[Exponential,Log & Box-Cox Transformation]()
- (徵求中)Feature Synthesis
### Traditional Regression Model
- [Naive Forcasting](https://hackmd.io/EFjDwFsdRcWobVRknFvrJA)
- [Linear Regression(LR) and Multivariable Linear Regression(MLR)](https://hackmd.io/azV6MXWkTkmOgYxUkoawog)
- (徵求中)[Autoregression and Auto Regressive Integrated Moving Average(ARIMA)](https://hackmd.io/YSu3Qb0zTS6T_MxZK5bgeQ)
- (徵求中)[Close Form Solution for Linear Model](https://hackmd.io/Zi-tz49HRS6pA-3dtwmngg)
- **Estimation error**
- (徵求中)[Mean Square Error(MSE)](https://hackmd.io/djMngF2TRB6iZxSFf9AfQg)
- (徵求中)[Mean Average Error(MAE)]()
- **Weight Regularization**
- (徵求中)[Over-fitting]()
- [Ridge/Lasso Regression(L1/L2 Norm)](https://hackmd.io/pk-D63cIQyatymDm9kVqMw)
- (徵求中)[Soft and hard margin](https://hackmd.io/BYnX2qL1RZCtjF3QY3aZAg)
- **Updating Regression Model**
- (徵求中)[Gradient Descent & learning rate]()
- (徵求中)[Stochastic Gradient Descent(SGD) &Mini Batch]()
- (徵求中)[Momentum]()
- **Evaluation**
- (徵求中)MSE/MAE
- (徵求中)[SSE, SST & $R^2$ score]()
- **Data Splitting**
- (徵求中)[Train-Test Splitting](https://hackmd.io/mE4amcYoRCendfWZnnKqGw)
- (徵求中)[Grid Search & Cross Validation(CV)](https://hackmd.io/en-_Kbf7TT2Acmz_q5XVlg?view)
### Traditional Classification Model
- [**K-Nearest Neiborhood(KNN)**](https://hackmd.io/x-F-3I7eTtmZnWus6-Ot9Q)
- **Logistic regression Family**
- [Logistic Regression](https://hackmd.io/OExpqswaRNGNw11dmOlaEw) `10/15`
- (徵求中)[*Generative Models for Classification](https://hackmd.io/xfnGCeRmRE-Sg58qmFMf3g?view)
- (徵求中)Lagrange Multiplier
- [Support Vector Machine(SVM) &Kernel-SVM](https://hackmd.io/wlvAV98CQPGjgzw6y3kD1Q)`10/15`
- **Objective(Loss) function**
- (徵求中)[Hinge/Square Hinge loss]()
- [Cross-entropy VS Square-error](https://hackmd.io/reAXkE7cTVqcw5LJrcmJJg)
- [Categorical/Binary Cross Entropy(CCE/BCE) Loss](https://hackmd.io/J2nAR1kSRCi7z8237B3h2Q)
- **Evaluating Classification**
- (徵求中)[Precision, Recall & F1 Score]()
- (徵求中)[ROC]()
- **Tree-based model**
- (徵求中)[Decision Tree (CART)](https://hackmd.io/TX4WKinETViv8EjDdzb16w)
- (徵求中)[Root node, Leaf node](https://hackmd.io/faCknBBeTD-KcyLva9eH9g?view)
- (徵求中)[Gini-index and Gini-table](https://hackmd.io/cPx5ND6aQlSnVNzMIEA_vA)
- (徵求中)[Random Forest](https://hackmd.io/MWDt-6QZSJW3d88RXC5yZw)
- (徵求中)[Model Ensemble](https://hackmd.io/sa-UMrrpRnqEie3J8cy04g?view)
- (徵求中)[Bagging](https://hackmd.io/AlrhZhjDTVCTp68lUAASRw)
- (徵求中)[XGBoost](https://hackmd.io/jR0u3a_cSX64DFHp8JOAkQ)
- (徵求中)[Boosting](https://hackmd.io/95bUI-FhQ4SVh7N5FfeVlw?view)
- (徵求中)[Gradient Boosting](https://hackmd.io/LDyXI_FpSr-9pUaHMaSknQ)
### Traditional Unsupervised Learning
- (徵求中)[Unsupervised Learning](https://hackmd.io/iZlyHHP6T9yBZ5DuhRNYOQ?view)
- (徵求中)[Principle Component Analysis]()
- (徵求中)[*Independent Component Analysis]()
- (徵求中)[T-SNE]()
- (徵求中)[K-Means]()
- (徵求中)[Hierarchical Clustering]()
- (徵求中)[Semi-supervised, Transfer learning](https://hackmd.io/Q_hqEOomSL2GdkoVrZOAHw)

---
## Deep Learning

### Dense Neural Network
- [**MLP**](https://hackmd.io/fgeG3CvMTVSKM41_QOTAKg)
- [***Tensors**](https://hackmd.io/uKBuTiXiTOOWhfW1ZybPGg)
- (徵求中)[**Gradient and Gradinet Vanishing**]()
- (徵求中)[**Gradient descent**]()
- **Activation**
- (徵求中)[Sigmoid]()
- (徵求中)[Softmax]()
- ***Initialization**
- (徵求中)[*Xavier]()
- (徵求中)[*HeNormal]()
- **Loss**
- MSE/MAE/CCE/BCE,...
- (徵求中)[Margin Loss](https://hackmd.io/efcpPL0mSJisycGtSfpdMA)
- [Focal Loss](https://hackmd.io/@aia-alumini-notes/H1YMR8lmq)
- (徵求中)[Gradient Descent & Backpropagation for DNN]()
- **Optimization**
- SGD, momentum
- (徵求中)[Adagrad]()
- (徵求中)[RMSProp]()
- (徵求中)[Adam](https://hackmd.io/5lUtu33wSPeX7VOlL12gng)
- **Normalization**
- (徵求中)[Batch Normalization]()
- (徵求中)[*Instance Normalization]()
- **Training Control**
- (徵求中)[Early Stopping]()
- (徵求中)[*Learning Rate Schedular]()
- **Discussion(陳弘軒)**
- (徵求中)[Better Model or Larger Data Size]()

### Convolution Neural Network
- (徵求中)[Convolution/Cross-correlation/Autocorrelation](https://hackmd.io/GRoOOqwvQpWFyd4ObdDHkA)
- (徵求中)[2D convolution](https://hackmd.io/SlR5JwG8RpGWGZufucdNsA)
- **Classical Model**
- (徵求中)[2012 AlexNet](https://)
- (徵求中)[2014 VGGNet](https://)
- (徵求中)[2014 GoogleNet](https://)
- (徵求中)[2015 ResNet](https://)
- [2016 SqueezeNet](https://hackmd.io/yxqT6e3MTCmcQt5BVVhmvw?view)
- (徵求中)[2017 DenseNet](https://)
- (徵求中)[*2019 EfficientNet](https://)
- (徵求中)[*2021 EfficientNetv2](https://)
- **XAI**
- [Expainable AI](https://)
- **Object Detection**
- [Problems of Object detection](https://hackmd.io/WEvIjq8XSGOC9dbsalddXg)
- RCNN Family
- (徵求中)[2014 RCNN](https://)
- (徵求中)[2015 Fast RCNN](https://)
- (徵求中)[2016 Faster RCNN](https://)
- YOLO Family
- [2016 YOLOv1](https://hackmd.io/DZ0IpjPRRh2a8e_7Y8lXRw)
- [2017 YOLOv2](https://hackmd.io/_-_bGj2GQViFrlQoob7MUw)
- [2018 YOLOv3](https://hackmd.io/3B18HqCDSzyIhNcZhcAM4A)
- [2020 YOLOv4](https://hackmd.io/TUmzWBUDRK2g7GIy_7Gydg)
- (徵求中)[*2021 Scaled-YOLOv4](https://)
- Anchor free
- (徵求中)[*2019 CenterNet](https://)
- (徵求中)[*2019 FCOS](https://)
- Roboflow
- [New Object Detection Project](https://hackmd.io/VuKT3cbyS8aqfVj3Dhmy-g)
- **Sementic Segmentation**
- DeepLab Family
- (徵求中)[*2014 DeepLab v1](https://)
- (徵求中)[*2016 DeepLab v2](https://)
- (徵求中)[*2017 DeepLab v3](https://)
- (徵求中)[*2018 DeepLab v3+](https://)
- (徵求中)[2015 U-Net](https://)
- (徵求中)[2019 HRNet](https://)
- (徵求中)[*2019 BASNet](https://)
- (徵求中)[*2020 OCRNet](https://)
- (徵求中)[*2020 $U^2$-Net](https://)
- **Instance Segmentation**
- 物件偵測 top-down 方法
- (徵求中)[*2017 Mask RCNN](https://)
- (徵求中)[*2019 Yolact](https://)
- (徵求中)[*2019 Yolact++](https://)
- (徵求中)[*2020 YolactEdge](https://)
- 基於語義分割 bottom-up 方法
- (徵求中)[*2017 SGN](https://)
- 綜合 top-down 及 bottom-up
- 直接分割方法

#### 其他電腦視覺任務:
- Pose Estimation
- Action Recognition
- Object Tracking
- Face Recognition
- Optical Character Recognition
- Person Re-Identification
- Emotion Recognition

### RNN
- [Vanilla RNN](https://hackmd.io/c9c1MmN9TyCLxVOXCioOjQ)
- LSTM
- GRU
- Seq2seq

### Transfer Learning
- [Model Fine-tuning](https://)
- **Multitask Learning**
- [Weight Sharing](https://)
- **Domain Adaptation**
- [ADDA](https://)
- **Zero-shot learning**

### Generative AI
- [Variational Auto Encoder(VAE)](https://hackmd.io/BICj4wwPQge7_3SlQNGYog)


- [GAN](https://hackmd.io/EZV9ZRLxSA2cwrS6LU6lPA)

### NLP
- [Transformer](https://hackmd.io/4t_Dn-8-QqqTxLQrBlEDHA)