Student Name: H Molvi
About the student
A relentless innovator, I'm always eager to expand and enhance my projects' features—even before perfecting the current ones. Agile methodologies resonate with my approach. Such is my dedication that sleep often takes a backseat, and my chair can attest to the long hours with its distinct imprint. However, when the holidays roll around, I eagerly delve into the latest novels from my favorite authors, savoring every moment of respite.
About the Project
We did a side-by-side comparison to see what it takes in terms of computer resources to use traditional ways of solving mazes versus using a smart agent that learns as it goes. For this, we made a simulation in Java, using JavaFX to show it visually. Here's how it worked: Making the Maze: We used Prim's algorithm and the spiral backtracker algorithm to set up the maze. Solving the Maze: We had two methods to solve the maze: Traditional algorithms, like Trémaux (similar to Depth-First Search) and Dijkstra. A learning-based method using SARSA (a kind of reinforcement learning) that keeps track of the path it learns in a Q-table. Measuring Resources: For each attempt, we noted how much CPU was used, how much memory, and how long it took. Then, we compared the results.
