Agent: Case Closed

Inspiration

We were very inspired to learn about and create a reinforced learning algorithm that is able to outsmart and outperform other models.

What it does/how it works

The way we trained it was by putting a reinforced learning ai model to train for, and we trained it against many different heuristic pathfinding algorithms in order to best prepare it for competition against other models.

Challenges we ran into

We made a localized version of the judge_engine.py file in order to visualize the models as they were training. The challenges that we ran into was trying to get the global variables and the local variables to sync up.

Accomplishments that we're proud of

The accomplishments that we are most proud of would be successfully implementing a reinforced learning algorithm from scratch.

What we learned

We learned not only just reinforced learning, but we had to learn about all sorts of different algorithms in order to successfully implement the model training.

Built With

• python
• pytorch