Deep Learning – Complete Lecture-wise Notebooks

This repository contains my complete Deep Learning learning journey, organized lecture-wise and implemented using Jupyter Notebooks.
Each folder groups a range of lectures, covering concepts from basic neural networks to CNNs, RNNs, LSTMs, Attention, and Transformers.

The notebooks focus on both theoretical understanding and hands-on implementation using Python and TensorFlow/Keras.

---

📂 Folder Structure

Deep-Learning/

├── lec1-5/

├── lec6-10/

├── lec11-20/

├── lec21-35/

├── lec36-50/

├── lec51-73/

└── README.md

---

📘 Lecture-wise Concepts Covered

🔹 lec1–5 : Foundations of Neural Networks

• Introduction to Deep Learning
• Types of Neural Networks
• Perceptron Model
• Perceptron Learning Rule
• Perceptron Trick

---

🔹 lec6–10 : ANN Basics & Forward Propagation

• Loss Functions in Perceptron
• ANN vs CNN Comparison
• Multi-Layer Perceptron (MLP)
• Forward Propagation
• Understanding Network Flow

---

🔹 lec11–20 : Training Neural Networks

• ANN with Real-world Dataset (Churn Prediction)
• Classification using ANN
• Regression using ANN
• Loss Functions (Deep Dive)
• Backpropagation Algorithm
• Why Backpropagation Works
• Gradient Descent (Batch, Stochastic, Mini-batch)
• Vanishing Gradient Problem in ANN
• MLP Memorization

---

🔹 lec21–35 : Improving Neural Network Performance

• Techniques to Improve NN Performance
• Early Stopping
• Data Scaling & Feature Scaling
• Dropout Layer
• Regularization Techniques
• Activation Functions
• ReLU Activation
• Weight Initialization Techniques
• Xavier & He Initialization
• Batch Normalization
• Optimizers in Deep Learning
• Exponentially Weighted Averages
• SGD with Momentum
• Nesterov Accelerated Gradient (NAG)

---

🔹 lec36–50 : Convolutional Neural Networks (CNN)

• AdaGrad Optimizer
• RMSProp & Adam Optimizer
• Hyperparameter Tuning (Keras Tuner)
• Introduction to CNN
• CNN vs Cortex & Convolution Operation
• Padding & Strides
• Pooling Layer
• CNN Architectures (LeNet)
• Backpropagation in CNN
• How CNN Backpropagation Works
• Data Augmentation

---

🔹 lec51–73 : Advanced Deep Learning & NLP Foundations

• Pretrained CNN Models
• CNN Filters & Feature Maps
• Transfer Learning
• Keras Functional API
• Introduction to RNNs
• Recurrent Neural Networks
• RNN-based Sentiment Analysis
• Types of RNNs
• Backpropagation Through Time (BPTT)
• Problems with RNNs
• LSTM Networks
• LSTM Architecture
• LSTM Word Prediction
• Gated Recurrent Unit (GRU)
• Deep & Stacked RNNs
• Bidirectional RNN
• Introduction to LLMs
• Encoder–Decoder Architecture
• Bahdanau Attention Mechanism
• Transformers
• Self-Attention
• Scaled Dot-Product Attention

---

🛠️ Technologies Used

• Python
• NumPy
• Matplotlib
• TensorFlow
• Keras
• Jupyter Notebook

---

▶️ How to Run the Notebooks

1. Clone the repository:

   git clone https://github.com/Aqsaabbasi2690/Deep-Learning.git
   

⚠️ Notes

Notebook outputs are cleared to keep file sizes manageable.

Datasets and trained models are not included.

This repository is for learning, practice, and revision purposes.

Learning Goal

This repository represents my structured journey toward mastering Deep Learning, with a long-term goal of advancing into NLP, LLMs, and Computational Linguistics research.

👩‍💻 Author

Aqsa Abbasi