Study core neural-network ideas through seven Jupyter-based labs spanning TensorFlow, Keras, and PyTorch. After a conceptual introduction to deep learning, you will work with tensors and computation, train regression models, assemble fully connected networks, and build a text-classification model.
The course combines framework tours with partially completed implementations. Its examples use California housing data, handwritten digits, IMDB movie reviews, and synthetic regression data, giving you several concrete contexts for layers, losses, optimizers, training loops, and predictions.
What You Will Learn
- Explain key milestones, concepts, and basic structures in deep learning
- Work with TensorFlow tensors, sessions, computational graphs, variables, placeholders, and matrix operations
- Implement TensorFlow linear regression with least squares, manual gradients, optimizers, and minibatches
- Build and train a multilayer fully connected TensorFlow network on handwritten-digit data
- Explore Keras layers, Sequential models, the Functional API, training, evaluation, and model visualization
- Prepare IMDB review sequences and construct a Keras Sequential sentiment classifier
- Use PyTorch tensors, autograd, neural-network modules, and optimizers
- Complete a PyTorch linear-regression model and training loop on synthetic data
Who This Course Is For
This course is for learners with Python and basic machine-learning familiarity who want to compare how three major frameworks express tensors, layers, models, and training. Although the labs are marked beginner, the compact notebooks assume you can follow NumPy-style arrays, matrix operations, model losses, and optimization code. It is a framework foundation, not a modern production deployment or advanced deep-learning course.
Prerequisites: Basic Python, NumPy-style array operations, and introductory understanding of linear regression and neural networks are recommended. Familiarity with Jupyter Notebook is helpful.
Learning environment: All seven labs run in a provided Ubuntu 22.04 Jupyter environment. Prepared notebooks mix explanations, runnable examples, and incomplete code cells for you to finish.
Frequently Asked Questions
Do I build models with all three frameworks?
Yes. TensorFlow is used for linear regression and a multilayer digit model, Keras for a Sequential IMDB sentiment model, and PyTorch for a linear-regression implementation. The framework guide notebooks also cover their core abstractions.
Is this suitable for someone completely new to Python?
No. The notebooks move quickly through tensors, matrix multiplication, losses, gradients, and training loops. A basic Python and introductory machine-learning foundation will make the exercises much more useful.
Does the course use current framework APIs?
Not throughout. The TensorFlow notebooks use 1.x patterns such as Session and placeholder; the Keras material uses the standalone keras package; and the PyTorch guide includes older Variable and CUDA 8/9 material. The concepts remain useful, but current documentation is necessary when writing new projects.
Which data sets and tasks are included?
The notebooks use California housing for TensorFlow regression, scikit-learn digits for a fully connected classifier, IMDB reviews for sentiment classification, and generated data for PyTorch regression.
