Practical Machine and Deep Learning

Welcome to the Practical Machine and Deep Learning repository! This repository contains a collection of Jupyter notebooks covering a wide range of machine learning and deep learning experiments. The goal is to provide practical, hands-on implementations of key ML/DL techniques, along with best practices and tools for model training and evaluation.

📌 Repository Overview

Each notebook in this repository explores a different ML/DL task, including computer vision, natural language processing (NLP), multimodal learning, and time-series forecasting. The experiments leverage popular frameworks such as TensorFlow, PyTorch, and HuggingFace Transformers.

🏆 Experiments & Notebooks

1️⃣ DNN Training with Tracking Tools

• Task: Train a deep neural network (DNN) to detect blondes in images.
• Key Topics: Best practices for CNN training, visualization using TensorBoard or ClearML.
• Dataset: A subset of CelebFaces dataset.
• Notebook: [Link]

2️⃣ Classify Product Reviews

• Task: Classify Amazon product reviews based on title, text, and additional parameters.
• Key Topics: Text classification, NLP, feature engineering.
• Notebook: [Link]

3️⃣ Part-of-Speech Tagging

• Task: Implement part-of-speech (POS) tagging using a many-to-many NLP approach.
• Key Topics: Universal POS tagging categories, sequence modeling.
• Dataset: Universal POS dataset.
• Notebook: [Link]

4️⃣ Fine-tuning of Vision Transformers (ViT)

• Task: Train a Vision Transformer (ViT) for image classification.
• Key Topics: HuggingFace Transformers, Vision Transformers (ViT), Low-Rank Adaptation (LoRA) for efficient fine-tuning.
• Notebook: [Link]

5️⃣ Question Answering with BERT

• Task: Fine-tune a BERT-based model for question-answering.
• Key Topics: Transformer-based NLP, transfer learning, context-based question answering.
• Notebook: [Link]

6️⃣ Multimodal Model Training

• Task: Train a multimodal model that matches images with their corresponding text descriptions.
• Key Topics: Multimodal learning, metric learning, Mean Reciprocal Rank (MRR) evaluation.
• Notebook: [Link]

7️⃣ Time Series Forecasting

• Task: Predict stock prices using historical time series data.
• Key Topics: Time series analysis, feature engineering, forecasting models, Mean Squared Error (MSE) evaluation.
• Notebook: [Link]

---

🔧 Tools & Frameworks Used

• Python 🐍
• TensorFlow/Keras 🏗️
• PyTorch 🔥
• HuggingFace Transformers 🤗
• Scikit-learn 📊
• Pandas & NumPy 📉
• Matplotlib & Seaborn 🎨
• TensorBoard & ClearML 📈 (for tracking and visualization)

🤝 Contributing

Contributions are welcome! Feel free to open issues, suggest improvements, or add new notebooks covering different ML/DL tasks.

---

🚀 Happy coding and exploring ML/DL!