Noel Selastin

I'm passionate about machine learning and AI. I build machine learning based projects on leisure time to hone my skills. I regularly read AI news and is fascinated by the progress of AI. I have a habit of being dedicated to and obsessed with the job I'm assigned to. I'm confident that my passion and dedication will boost the growth of the company. I'm very analytic, I like to go in depth and reach the bottom of a problem. My analytic skills will outshine when it comes to practicality, which is what is essential for a company. I'm very competitive and I always have a natural tendency to push myself and be a level above my peers. I tend to constantly better whatever skill I'm doing, and master it.

1. Chess game using Java:

• A 2D 8x8 matrix to store positions of pieces.
• Algorithms for movement of each type of pieces.
• Used Java swing and Java AWT libraries for display and animation of board and pieces.
• Inherited MouseAdapter class of java.awt for listening to mouse events.

2. Reinforcement learning project:

• Involves an algorithm that finds the shortest path from initial to final square in a 5x5 grid without falling into the "holes".
• Uses a reinforcement learning technique called Q-learning, uses the Q-learning formula.
• There will be reward for reaching the goal and punishment for reaching the hole.
• Finds the best path in less than 20 seconds of run time, for any arrangement of goal, origin and holes.
• The Q-learning formula predicts the Q-value by considering rewards in the future state, the formula involves back propagation.
• It doesn't involve neural networks, as neural networks were tested to be slower for this particular problem.

3. Supervised learning image classifier:

• Multi classifier that identifies about 50 different buildings.
• Trained on about 1000 labeled images.
• Achieved about 95% accuracy.
• Used Convolutional Neural Network.
• Integrated with virtual tour guide app, for college mini-project.