View Proposal

---

Proposer

Alessandro Suglia

Title

Learning to reason over long trajectories for Embodied AI tasks

Goal

Create models that are able to handle very long action trajectories in Embodied AI tasks

Description

In this project, we will explore efficient deep learning architectures for training artificial agents that can learn to execute actions in the environment. For instance, we will investigate whether it's possible to train models that can learn to play ATARI games.

Resources

Mamba: https://arxiv.org/abs/2312.00752 Flash attention: https://arxiv.org/abs/2205.14135 Decision Transformer: https://arxiv.org/abs/2106.01345

Background

The candidate student should have strong Python programming skills as well as Machine Learning knowledge with a focus on Deep Learning methods.

Url

Difficulty Level

High

Ethical Approval

None

Number Of Students

1

Supervisor

Alessandro Suglia

Keywords

deep learning, neural networks, embodied ai, robotics

Degrees

Master of Science in Artificial Intelligence
Master of Science in Data Science
Master of Science in Robotics