Name: Gianluigi Lopardo
Type: User
Company: Inria & Université Côte d'Azur
Bio: PhD candidate in applied mathematics, working on interpretability of machine learning models.
Twitter: gigilopardo
Location: Nice, France
Blog: gianluigilopardo.science
Gianluigi Lopardo's Projects
Prediction of absenteeism at work, with machine learning models for classification.
A Sea of Words: An In-Depth Analysis of Anchors for Text Data
A Comparison of Feature Importance and Rule Extraction for Interpretability on Text Data
Code for the paper "Attention Meets Post-hoc Interpretability: A Mathematical Perspective"
The purpose of this paper is to analyze how and how much a film's attributes affect its rating, using several regression techniques.
Stima giornaliera del valore R(t) del COVID-19 nelle regioni italiane
(Part of) a project for the Business intelligence class. The goal is to analyze a dataset of financial articles and apply machine learning technique to extract useful information.
This project is about the analysis of credit risks of German Credit Data. Different classification models and preprocessing methods are used and compared.
This notebook is ispired by the AIX360 HELOC Credit Approval Tutorial, which shows different explainability methods for a credit approval process. Here XGBoost is used for classification, achieving better accuracy than most of the models used in that notebook. Then, feature importance methods are shown, to be compared with the Data Scientist explanations methods provided in the above notebook. The first ones come directly with XGBoost and the other is based on SHAP.
The project's goal is to get familiar with cutting-edge models capable of acting in an open world, incremental learning approaches in image classification and open set strategies
Scientific programming and computing project carried out in MATLAB.
The purpose of this work is to predict which customers are about to leave the bank. To do this, the main classification algorithms will be used to predict whether a customer from 1999 will still be a customer in 2000 or not.
My solution of a simple Reinforcement learning problem
A New Method for the Interpretability of Composite Decision Systems.
A series of statistical models applied to different case studies
The aim of this project is to apply and explore Transfer Learning. The dataset used is Caltech101, the neural network used for the first part of the project is AlexNet.