Michele Alessi Website

Michele Alessi

I'm Data Scientist

About

Michele Alessi

CV: PDF
Email: michele.alessiareasciencepark.it
City: Trieste, Italy

I am a first-year PhD student at the LADE Data Engineering Lab, focusing on applying density-based unsupervised learning techniques to generative models. My theoretical research concerns studying new prior distributions for variational autoencoders. I also work on energy-based models applied to Physics-Informed Neural Networks (PINNs). Since the beginning of my PhD, I have been working on variational inference models applied to scRNA-seq. I hold a Bachelor's in Mathematics and a Master's in Data Science and Scientific Computing from the University of Trieste.

RESUME

Education

Ph.D. in Artificial Intelligence & Applied Data Science

2024 - present

Area Science Park, Trieste, Italy

Supervisor: Prof. Alejandro Rodriguez Garcia , Prof. Alessio Ansuini

Honor Student

2019 - 2024

Collegio di Merito Luciano Fonda, Trieste, Italy

M.Sc. in Data Science & Scientific Computing

2022 - 2024

University of Trieste, Trieste, Italy

Supervisor: Prof. Alejandro Rodriguez Garcia

B.Sc. in Mathematics

2019 - 2022

University of Trieste, Trieste, Italy

Supervisor: Prof. Luca Manzoni

Work Experience

Machine Learning Engineer

Jun 2024 - Nov 2024

Asimov AI, Rome, Italy

Research Intern

May 2024 - present

SISSA mathLab, Trieste, Italy

Data Analyst

Nov 2023 - May 2024

Social Fox S.r.l., Udine, Italy

PROJECTS

As an MSc student in Artificial Intelligence, I have worked on various projects that leverage advanced deep learning techniques. Here are some of the projects I have completed or am working on.

Evolutionary GAN

Evolutionary GAN

Theoretical project about Evolutionary Algorithms applied to Generative Adversarial Networks. During this project, we studied and implemented a new crossover operator acting on a population of generators.

Latent Semantic Analysis

Latent Semantic Analysis

This project aims to implement different models for Latent Semantic Analysis (LSA) using deep neural networks and SVD techniques and compare their performances. The goal of the assessment is to evaluate the quality of the learned latent space.

High Performance Computing

High Performance Computing

Final project for HPC course, implementing the Game of Life using OpenMP and MPI.

Fashion MNIST

This project explores FashionMNIST dataset with unsupervised and supervised learning techniques to evaluate their effectiveness in classification.

Hidden Markov Model for Text Decryption

Hidden Markov Model for Text Decryption

This is the final project for Probabilistic Machine Learning course. It applies a MCMC and HMM methods for text decryption. The method is applied for decrypting messages which have been encoded using substitution cipher, homophonic cipher an double cipher.

Flappy Bird Reinforcement Learning

Flappy Bird Reinforcement Learning

Final project for the Reinforcement Learning course: the goal is to teach an agent how to play the flappy bird game. The agent has no knowledge of the environment but perfect observability of the states. As such it is phrased as a model-free reinforcement learning problem. Standard Reinforcement Learning techniques for policy controls was attempted and compared.

Topological Analysis of Neural Networks

The aim of this study is to conduct topological analysis on a FCNN trained using the MNIST dataset. Various clustering techniques were applied to identify the network's critical neurons, and the results were analyzed by examining their corresponding weights.

Clustering Methods

Clustering Methods

A small repository implementing classical clustering algorithm from scratch.

Magnitude Homology

Study of asymptotic properties of the magnitude for a graph and how magnitude homology groups behave on graphs.

Lotka-Volterra Model

Il modello di Lotka-Volterra si propone di studiare la dinamica dell’evoluzione di due specie che convivono in un certo ambiente, sotto determinate ipotesi.

CONTACT

Address

Area Science Park 99, Padriciano, Trisete, Italy

Email

michele.alessi@areasciencepark.it