AI Course Projects

Project 1: Genetic Algorithm for Fractional Knapsack Problem

• Description: We implement a genetic algorithm to solve the fractional knapsack problem. The goal is to maximize the value of items placed in a knapsack while respecting its weight capacity and value.
• Key Concepts: Genetic algorithms, optimization.

Project 2: Speaker and Digit Detection using Hidden Markov Models (HMM)

• Description: We develop an HMM-based system to detect both the speaker and the spoken digit in an audio dataset. We extract features using Mel-frequency cepstral coefficients (MFCCs) and implement HMMs from scratch as well as using the hmmlearn library.
• Key Concepts: Hidden Markov Models, feature extraction, audio processing.

Project 3: Flower Image Clustering

• Description: Our focus here is on clustering a dataset of flower images. We extract features using the pre-trained VGG16 model and apply K-means and DBSCAN clustering algorithms. We also use PCA for 2D visualization and optimization.
• Key Concepts: Convolutional neural networks (CNNs), clustering, dimensionality reduction.

Project 4: Boston Housing Price Prediction

• Description: We predict house prices in Boston using various machine learning models. After data preprocessing, we explore Linear Regression, KNN, Decision Trees, Random Forests, and XGBoost. Hyperparameter tuning is crucial for model performance.
• Key Concepts: Regression, ensemble methods, hyperparameter tuning.

Project 5: Suicide Intention Prediction from Twitter Data

• Description: We design a CNN to predict suicide intention based on suicidal tweets from Twitter. We experiment with different window-contexts, regularization methods, and loss functions (e.g., cross-entropy).
• Key Concepts: Natural language processing (NLP), deep learning.

Project 6: Competitive Snake Game with Q-Learning

• Description: We complete a competitive 2-snake game where snakes are programmed using Q-learning. The snakes autonomously learn to eat apples, avoid walls, and outperform opponents.
• Key Concepts: Reinforcement learning, Q-learning.