Updated overview

Author: amitkaps

008-Data_Aug-DosaNoDosa.ipynb renamed to 008-Data-Aug-DosaNoDosa.ipynb

@@ -792,7 +792,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.6"
+   "version": "3.6.4"
   }
  },
  "nbformat": 4,

011-RNN-LSTM-Toxic.ipynb renamed to 011-RNN-LSTM-Toxic-Copy1.ipynb

@@ -1155,7 +1155,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.6"
+   "version": "3.6.4"
   }
  },
  "nbformat": 4,

012-CNN-1D-Toxic.ipynb

@@ -21,8 +21,11 @@ Notebook Overviews
 - **008**: Data Augmentation - Dosa/No Dosa
 - **009**: Transfer Learning - Dosa/No Dosa
 - **010**: Theory & Concept - Natural Language Processing
-- **011**: Pre-Trained Embedding - NLP
-- **012**: Recurrent Neural Network - Toxic Classification
+- **011**: Recurrent Neural Network - Toxic Classification
+- **012**: Convolution - 1D - Toxic Classification
+- **013**: Pre-Trained Embedding - Words - Toxic Classification
+- **014**: Pre-Trained Embedding - Sentences - Toxic Classification
+- **015**: Generating Text
 
 ## Resources
 

012-PreTrained-Toxic.ipynb renamed to 013-PreTrained-Words-Toxic.ipynb

@@ -800,7 +800,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.6"
+   "version": "3.6.4"
   }
  },
  "nbformat": 4,

014-PreTrained-Sentence-Toxic.ipynb

@@ -31,7 +31,7 @@ _**Getting Started with Image & Text**_
 
 - **Supervised**: Regression, Classification, ...
 - Unsupervised: Dimensionality Reduction, Clustering, ...
-- Semi-supervised
+- Self-supervised
 - Reinforcement
 
 ---
@@ -126,6 +126,8 @@ Core Layers
 
 # Best Practices
 
+---
+
 ## Pre-processing
 - **Normalize** / **Whiten** your data
 - **Scale** your data appropriately (for outlier)
@@ -134,13 +136,17 @@ Core Layers
 - **Stratified** split for multi-class data
 - **Shuffle** data for non-sequence data. Careful for sequence!!
 
+---
+
 ## General Architecture
 - Use **ADAM** Optimizer
 - Use **RELU** for non-linear activation
 - Add **Bias** to each layer
 - Use **Xavier** or **Variance-Scaling** initialisation
 - Refer to output layers activation & loss function guidance
 
+---
+
 ## Dense Architecture
 - No. of units reduce in deeper layer
 - Units are typically 2^n
@@ -150,6 +156,7 @@ Core Layers
 - Increase **Convoluton filters** as you go deeper in 32 -> 64 -> 128
 - Use **Pooling** to subsample: Makes image robust from translation, scaling, rotation
 
+---
 
 ## Learning Process
 - **Validation Process**