Potential and Pitfalls of Biometric Recognition Methods Based On Photoplethysmography

This project evaluates PPG-based user recognition methods using wearable data across controlled and real-world conditions.

Robust PPG Authentication

Investigating how three real-world sources of variability — temporal variability (elapsed time between enrollment and authentication), user-specific variability, and state-dependent variability (e.g., stress, sleep, mood) — affect the reliability of user identification based on photoplethysmography (PPG) signals.

Transformer

Transformer encoder-decoder model to solve mathematical problems.

Convolutional neural networ

Convolutional neural network for image classification (CIFAR-10).

LSTM

LSTM language model for predicting fables using PyTorch.

Spreading Tweet News Project

Spreading tweet news simulation, case study of Giuseppe Conte followers network.

Iterative Methods

Solve symmetric and positive definite matrices.

Compress and decompress Images

Implementation of the algorithm DCT2 and IDCT2 in R.