Build, Train and Deploy ML Models with Keras on Google Cloud

Module 1: A New Programming Paradigm

⏳ ~5 hrs

• Topics: Intro to ML/DL and TensorFlow’s programming paradigm. Includes discussion with Andrew Ng, neural network basics, and “Hello, World” neural nets.

• Hands-on: TensorFlow setup and simple classification model coding in Python.

Module 2: The Sequential Model API

⏳ ~6 hrs

• Topics: Build and train neural networks using the Keras Sequential API—cover layers, model compilation, fitting, evaluation, and prediction.

• Hands-on: Build CNNs in Colab for MNIST digit classification.

Module 3: Validation, Regularization & Callbacks

⏳ ~6 hrs

• Topics: Techniques to avoid overfitting, set up validation workflows, and use callbacks including EarlyStopping.

• Hands-on: Train models on Iris dataset, tune with regularization, and practice callback mechanisms.

Module 4: Model Persistence & Advanced Structures

⏳ ~6 hrs

• Topics: Save/load models, select weight-only vs full model saving, explore pretrained models. Also introduction to advanced architectures: CNNs, RNNs, transformers, and autoencoders.

• Hands-on: Use Keras for advanced model building and application.

Get certificate