Luca Corbucci

We all rely on AI tools every day, but would we take responsibility for the decisions they make? From classic ML to LLMs and agents, this talk explores the risks of AI through real-world failures. We will see how and why AI systems break, and how we can avoid repeating those mistakes.

Discover how Federated Learning enables collaborative training of Machine Learning models without directly sharing your private data. Discover the fundamentals, explore real-world applications and learn how to simulate decentralized training using tools like Flower.

Join me in uncovering privacy vulnerabilities of Machine Learning, and discussing attacks like Membership Inference. Learn how techniques like Differential Privacy and tools such as Opacus can mitigate the privacy risk of your models.

Streamlit è un progetto open source che permette di creare in pochi minuti delle data-web-app (e non solo) scrivendo solamente codice Python senza necessità di avere esperienza di front-end. In questa sessione scopriremo il suo funzionamento, come deployare un’app e vedremo un possibile caso d’uso.