Alessandro Abate

A method for verifying safety in self-driving cars and drones using neural networks.

A new approach uses data to simplify the analysis of complex systems.

Using supermartingales to ensure the safety and efficiency of stochastic systems.

A new approach improves feedback collection for language models, saving time and costs.

This article introduces ValueWalk, a method for improving computer learning from expert behavior.

Examining the difficulties of creating effective reward functions in reinforcement learning.

A method for developing resilient policies for drones in uncertain environments.

A structured method to evaluate AI safety and effectiveness.

Setting rules for AI safety while avoiding sneaky tricks.