Lecture Program

1.  Linear Temporal Logic (De Giacomo): LTLf, DFA

2.  Automata learning : (Paludo Licks) L*, merging states methods

3. Basics of Reinforcement Learning (Capobianco, Fanti): exploration/exploitation, Q-learning, Deep RL

4. Non Markovian RL (Patrizi): NMRDP, restraining bolts, reward machines

5. Deep Learning on Sequential Data (Umili): Deep learning for sequential data (models, applications), hybrids between RNN and DFA (L* extraction, DeepDFA), non- markov deep RL

6. Neurosymbolic (NeSy) AI (Umili): introduction and applications, grounding symbols, Logic Tensor Network (LTN), Tensorization of LTLf knowledge

7. NeSy RL (Umili): noisy symbols, Neural Reward Machines, transfer learning between LTL tasks

8. Reasoning Shortcuts (Marconato, Umili): Ungroundability and Reasoning shortcuts

9. Natural Language and LTL (Liu): towards using natural language in place of LTL for robotic applications

10. NeSy Sequence Generation (Van De Broeck): injecting logical knowledge in autoregressive models