Alberto Marinelli's repositories
QRI-ELM-and-ELM-with-Standard-Momentum
In this project it is used a Machine Learning model based on a method called Extreme Learning, with the employment of L2-regularization. In particular, a comparison was carried out between: (A1) which is a variant of incremental extreme learning machine that is QRIELM and (A2) which is a standard momentum descent approach, applied to the ELM.
Data-Mining-UniPi
Data Mining project carried out on two datasets extracted from the Twitter platform, one on Users and one on Tweets
Helm-AI-SA22
The orchestrator repository for the Helm project, an AI-based research assistant for innovation managers and academics
ISPR-midterms
Assignments for the course of Intelligent Systems for Pattern Recognition (university of Pisa)
Neural-Network-from-scratch
Neural Network trained through a classical Back-Propagation (BP) approach and employing both the momentum gradient-based optimization technique and L2-regularization implemented from scratch using the MATLAB programming language.
Ontology-for-italian-research-organisation
Creation of an ontology for an Italian research organisation
Parallel-genetic-tsp
Parallelization of a genetic algorithm to solve the Travelling Salesman Problem (TSP). A sequential version is developed, followed by an analysis to identify components suitable for parallelization. Two parallel implementations are created using standard threads and FastFlow.
Siamese-Transformer-Networks-for-Key-Point-Analysis
The aim of this project is to implement and explore architectures based on Siamese Transformer networks to solve Natural Language Processing tasks, in particular the one proposed by IBM’s shared task 2021