Machine-Learning-Python

The repository contains Machine Learning related assignments or projects. The sources are either come from a class at Florida Atlantic University or side project from the web (sources in the jupyter notebook)

Current Topics:

1. Digit Classification Using Machine Learning

• Load & Visualize the MNIST dataset
• Split the dataset into Train/Test set (hold-out method)
• Classification using SGD Classifier, RandomForest Classifier
• Plotting Confusion Matrix
• K-Fold Cross Validation on the training set and trained model
• Data Normalization / Standardization
• Data Augmentation
• Hyperparameters Optimization

2. Digit Classification Using Neural Networks

• Load & Visualize the MNIST dataset
• Split the dataset into Train/Test set (hold-out method)
• Set up 3-layers (input, hidden, output) model & 2-layers (input, output) model and train the network using SGD
• Classification using MLP Classifier from scikit-learn
• Plotting Confusion Matrix, Learnning Curve
• K-Fold Cross Validation on the training set and trained model
• Data Normalization / Standardization
• Hyperparameters Optimization

3. Digit Classification Using Convolution Neural Network

• Load & Visualize the MNIST dataset
• Split the dataset into Train/Test set (hold-out method)
• Build & Train the Deep MLP in Kera
• Build & Train the Convolution Neural Network in Keras
• Plotting Confusion Matrix, Learnning Curve
• Hyperparameters Optimization

4. Image Classification Using Transfer Learning

• Classify an image using a pretrained(on ImageNet) model(ResNet50)
• Load and Split the CIFAR10 dataset
• Visualize a sample of the CIFAR-10 dataset
• Classify images using a CNN built from scratch
• Perform Data Augmentation on CNN Model Built from scratch
• Pretrained VGG19 Model as Feature Extractor to Train a Conventional Machine Learning Classifier
• Pretrained VGG19 Model as the Base Model with Additional Layers (freezing base layers and add appropriate head/output layer)
• Plot accuracy and loss

5. Supervised Image Classification Using TensorFlow

• Classify an Image using a Pretrained Inception V3 (Imagenet)
• Load & Visualize the flower_photos dataset
• Data Augmentation using ImageDataGenerator
• Run the Pretrained Inception V3 (use the classifier as it is) on a Batch of Images
• Pretrained Inception V3 Model as the Base Model with Additional Layers
• Plot Learning Curves

6. Classification with XGBoost

• Classification task on the Iris dataset using XGBoost

7. Reinforcement Q-Learning with OpenAI Gym

• Using Brute-Force Approach for the Self-driving Tax Problem
• Using Reinforcement Learning (Q-Learning Algorithm) for the Self-driving Tax Problem

8. K-Means Clustering and Support Vector Machines

• Principle Component Analysis (PCA)
• K-Means Clustering
• Support Vector Machine (SVM) / Support Vector Classification (SVC)

9. Movie Recommender System

• Memory-based approach: Item Based Collaborative Filtering
• Model-based approach: K Nearest Neighbor (KNN)