Natural Language Processing (NLP) is one of the most exciting fields in Machine Learning and AI. In this beginner-friendly workshop, I will introduce Keras as a Python library for applying NLP techniques on a multi-label emotion classification dataset and present how to use it to build a deep learning classifier. Besides, I will also compare and demonstrate how applying NLP techniques improve the predictability of a model and then, tune a neural network using Keras Tuner to boost the overall model performance. The complete agenda is as outlined below:
1. Introduction to Tokenization
- Keras tokenizer
- Padding and Truncating
2. Introduction to Neural Network and brief overviews of
- Fully Connected Layer
- Embedding Layer
- Dropout Layer
- Long Short Term Memory (LSTM)
3. Model Training
- Early Stopping
- Checkpoints and Callbacks
- Model Compilation
4. Model Evaluation
- Training and Validation Loss
- Reading a Multi-class Classification Report
- Confusion Matrix
5. Model Improvement
- Cleaning and Normalizing Text Data
- Model Compiling Strategy
- Data Usage Strategy
- Optimizing ANN hyperparameters using Keras Tuner

This workshop focuses more on using Keras for NLP, multi-class model evaluation, and model optimization using Keras Tuner than on building ANNs from scratch, however, the basic building blocks will be revised during the session.

Resources Available at: https://github.com/royn5618/Talks_Resources/tree/main/PyConPortugal2022