I am a telecommunication and data scientist, specialized in parallel and distributed programming, data analytics and deep learning techniques.
My recent work includes the analysis of the orders from a food aggregator company.
The project demonstrates effective techniques for building robust CNNs. Key insights include using convolutional layers to extract features, pooling layers for dimensionality reduction, and fully connected layers for predictions. To prevent overfitting, I have employed data augmentation, dropout, early stopping, and L2 regularization. Then I have used Batch normalization to stabilize and accelerate the training process. Optimizers like Adam and learning rate schedules enhance training efficiency. Performance is measured using accuracy and confusion matrices. These strategies collectively improve model generalization and efficiency, offering a comprehensive approach to training effective neural networks while mitigating common pitfalls like overfitting.
Strongly convinced in the importance of continuous learning and in the value of the scientific research based on a multidisciplinary approach.