Learning Outcome:

Understand and code CNN models for image recognition
Diagnose overfitting issues in image recognition and propose methods to overcome the issues.
Perform functional API coding based on a selected model.
Implement transfer learning to fine tune the image recognition models.
Understand code RNN models and word embedding for text recognition

Course Outline:

Topic 1 Image Recognition with CNN

Introduction to Convolutional Neural Network (CNN)

Convolution & Pooling

Build a CNN Model for Image Recognition

Topic 2 Overfitting for Small Datasets

Overfitting and Underfitting

Methods to Solve Overfitting

Small Dataset Overfitting Issue

Data Augmentation & Dropout

Topic 3 Functional Keras API

What is Functional API

Create Sequential Model with Functional API

Create Non-Sequential Models with Functional API

Topic 4 Transfer Learning for Small Datasets

Introduction to Transfer Learning

Pre-trained Models

Transfer Learning on Small Dataset

Topic 5 Text Classification with RNN

Introduction to Recurrent Neural Network (RNN)

Types of RNN Architectures

LSTM and GRU

Word Embedding

Build a RNN Model for Text Classification

Knowledge and Skills

Able to operate computer functions with minimum Computer Literacy Level 2 based on ICAS Computer Skills Assessment Framework
Minimum 3 GCE 'O' Levels Passes including English or WPL Level 5 (Average of Reading, Listening, Speaking & Writing Scores)

Attitude