Maze Tank Game

Maze Tank is a game created on Unreal Engine 5, in which the goal is to defeat all the enemy towers to release the exit from the maze and, in the end, find the boss tower that has a higher amount of shots per second. The game was developed in pairs for a college course, using the C++ language and blueprints.

The player controls a tank that needs to navigate through the maze and destroy the enemy towers, avoiding their attacks. The towers have different levels of difficulty and require different strategies to defeat. At the end of the maze, the player faces the boss tower, which is the most powerful and challenging.

The game features high-quality graphics and immersive gameplay, as well as an exciting soundtrack. The combination of action and strategy elements makes Maze Tank a fun and challenging experience for players.

Machine Learning to classify helicopters and airplanes 

It was an assignment for the machine learning course in college that consisted of using convolutional neural networks to classify helicopters and airplanes. Four CNN models were used to achieve the desired classification accuracy.

The first CNN model used was the sequential model, which achieved 77% accuracy. Next, the data augmentation model was used, which increases the amount of training data, but achieved an accuracy of only 68%.

The third model used was the feature extraction model, which uses a pre-trained neural net to extract important features from the input data. This model achieved an accuracy of 87%, showing a significant improvement over the previous models.

Finally, the fine-tuning model was used, which consists of adjusting the weights of the pre-trained neural network for the specific data of the problem at hand. This model achieved an impressive 95% accuracy.

In summary, the work demonstrated the importance of using different CNN techniques to achieve a desired accuracy in data classification. The fine-tuning model was the most effective, but all the models used made significant contributions to the project.


AI for predicting stock values 

The project consists of using artificial intelligence to forecast values in the Brazilian stock market.

(In Progress)