A practical, notebook-first deep learning learning path that moves from mathematical foundations and classical neural-network algorithms to computer vision, sequence models, NLP, and TensorFlow data pipelines. The repository brings together two complementary collections with Jupyter notebooks, datasets, visual explanations, and hands-on projects.
- Learning collections
- What you will learn
- Algorithm and technique map
- Repository structure
- Getting started
- Datasets and practical notes
- How to study the notebooks
- Contributing
- License
- Connect with me
The repository is organized into two independent collections. Each one has its own detailed README, learning roadmap, and setup guidance.
| Collection | Focus | Lessons |
|---|---|---|
| 100 Days of Deep Learning — CampusX | Neural-network intuition and practice: perceptrons, backpropagation, optimization, training improvements, ANN/CNN, and recurrent models. | 24 notebooks across the available day folders |
| Deep Learning with TensorFlow and Keras — Codebasics | Foundations through TensorFlow/Keras applications: classification, evaluation, image models, transfer learning, embeddings, Word2Vec, and input pipelines. | 22 notebooks across the numbered lessons |
The collections are useful independently. For a guided route, begin with the perceptron and matrix lessons, continue through gradient-based learning and training techniques, then explore CNN, RNN, and NLP applications.
- Matrix operations and neural-network building blocks
- Perceptron geometry, the perceptron trick, and perceptron loss
- Forward propagation, backpropagation, and gradient-based learning
- Batch, stochastic, and mini-batch gradient descent
- Sigmoid, tanh, ReLU, and leaky ReLU activation functions
- Loss functions, predictions, and decision boundaries
- Data preparation, feature scaling, and encoding
- Weight initialization, including Xavier/Glorot and He methods
- Early stopping, dropout, regularization, and batch normalization
- Confusion matrices, accuracy, precision, recall, and F1 score
- Class imbalance strategies, including class weighting, resampling, and SMOTE
- Training curves and practical ways to compare model behavior
- ANN classification for handwritten digits
- CNN fundamentals: padding, strides, convolution, and pooling
- CIFAR-10 image classification
- Cat-vs-dog image classification
- Image augmentation and transfer learning
- Keras embedding layers and Word2Vec with Gensim
- SimpleRNN, stacked RNN, LSTM, and GRU architectures
- IMDB sentiment analysis and next-word prediction
tf.data.Datasetpipelines and transformations- Dataset filtering, mapping, shuffling, and batching
- Caching and prefetching to improve input-pipeline throughput
- TensorBoard for inspecting model training
| Category | Algorithms and methods in the notebooks |
|---|---|
| Linear classifiers | Perceptron, perceptron trick, perceptron loss |
| Optimization | Gradient descent, batch gradient descent, stochastic gradient descent, mini-batch gradient descent, backpropagation |
| Activations | Sigmoid, tanh, ReLU, leaky ReLU |
| Loss and evaluation | Mean absolute error, binary cross-entropy, confusion matrix, accuracy, precision, recall, F1 score |
| Neural architectures | Artificial neural networks (ANN), convolutional neural networks (CNN), SimpleRNN, stacked RNN, LSTM, GRU |
| Generalization | Dropout, regularization, early stopping, data augmentation, batch normalization |
| Initialization | Zero, Xavier/Glorot, and He initialization |
| Imbalanced learning | Class weights, undersampling, oversampling, SMOTE, and ensemble-style undersampling |
| NLP | Trainable word embeddings and Word2Vec |
| Input pipeline | tf.data, caching, and prefetching |
.
├── 100-Days-of-Deep-Learning-CampusX/
│ ├── README.md
│ └── numbered day folders with notebooks and lesson data
├── Deep-Learning-with-Tensorflow-and-Keras-From-Codebasics/
│ ├── README.md
│ └── numbered lesson folders with notebooks and lesson data
├── Deep Learning.png
├── LICENSE
└── README.md
Open the README inside a collection for its complete lesson-by-lesson index. Notebooks use files in their own lesson folders, so preserve the directory layout when opening or running them.
- Python 3.x (use a Python version supported by the TensorFlow release available for your operating system)
- Jupyter Notebook or JupyterLab
- Common packages used across the collections: NumPy, Pandas, Matplotlib, Seaborn, scikit-learn, TensorFlow, Gensim,
imbalanced-learn, OpenCV, andmlxtend
Not every notebook requires every package. The collection-specific README files list additional dataset requirements and setup details.
Clone the repository and move into its directory:
git clone https://github.com/Md-Emon-Hasan/Complete-Deep-Learning-Algorithm-Beginner-to-Advanced.git
cd Complete-Deep-Learning-Algorithm-Beginner-to-AdvancedCreate and activate a virtual environment.
Windows PowerShell
python -m venv .venv
.venv\Scripts\Activate.ps1macOS/Linux
python -m venv .venv
source .venv/bin/activateInstall the libraries commonly used by the notebooks:
python -m pip install --upgrade pip
python -m pip install jupyter numpy pandas matplotlib seaborn scikit-learn tensorflow gensim imbalanced-learn opencv-python mlxtendStart Jupyter from the repository root:
jupyter notebookChoose a collection, open one of its lesson folders, and run a notebook from top to bottom. If a notebook needs a different package or a separately downloaded dataset, follow the notes in that collection's README and notebook.
- Many lessons include their CSV, JSON, image, or other supporting files in the lesson folder. Keep those files beside the notebook and preserve relative paths.
- MNIST, CIFAR-10, and IMDB examples use Keras dataset helpers; an internet connection may be required the first time the dataset is downloaded.
- The CampusX cat-vs-dog project expects an
archive.zipdataset that is not included. See that collection's README before running its data-preparation cells. - The Codebasics Word2Vec lessons use review data. Git LFS is configured for JSON files in that collection; install Git LFS and run
git lfs pullif a cloned file contains an LFS pointer instead of data. - One Codebasics lesson folder currently contains its dataset without the corresponding lesson notebook. The collection README identifies this gap.
- Most examples are suitable for CPU experimentation; image-model training may take longer without a GPU.
- Results can vary with package versions, random initialization, hardware, and dataset availability. These notebooks are educational examples rather than benchmark guarantees.
- Choose a topic from the roadmap in either collection's README.
- Read the notebook's explanations before running its cells.
- Run cells in order, from the correct lesson folder, so relative data paths resolve.
- Change one setting at a time—such as learning rate, activation, model size, or regularization—and compare the result.
- Use plots, metrics, and predictions to connect the algorithm to its behavior.
Corrections, clearer explanations, and focused lessons are welcome.
- Fork the repository and create a branch.
- Keep new material in the appropriate collection and follow its numbered folder style where practical.
- Include data instructions and package requirements for any new notebook.
- Avoid committing credentials, generated environments, caches, or large datasets without a clear need.
- Open a pull request describing the change and how to follow or review it.
This project is distributed under the Apache License 2.0. See LICENSE for details.
I'm Md Emon Hasan, a machine learning engineer interested in deep learning, applied machine learning, and practical AI projects. Reach out for questions, feedback, or collaboration.
- 🌐 Portfolio: emonlabs-ai.hitechparks.com
- 💻 GitHub: Md-Emon-Hasan
- 💼 LinkedIn: Md Emon Hasan
- 🤗 Hugging Face: emonHasan01
- 📧 Email: emon.mlengineer@gmail.com
- 📱 WhatsApp: +8801834363533
If these learning notes help you, please consider starring the repository. ⭐
