Seminar will run in Czech or English (depending on students attending) and it will be organized as a series of presentations of research papers followed by Q&A OR solving challenges (with presentation) OR research workshops (with a written report).

For challenges, a group of 2-3 students solves some challenge problem (see web sites below) and has two presentations during the seminar (one introducing the problem and sketching the solving ideas; one presenting the results) :

• Kaggle
• AI crowd
• MicroRTS
• Legends of Code and Magic
• Flatland
• LORR

For paper presentations, the student will prepare 40-50 minutes presentation based on a research paper selected from the folowing list (students may also suggest another paper, but it must be approved by the teacher):

1. Mastering Chess and Shogi by Self-Play with a General Reinforcement Learning Algorithm [web]
2. Attention Is All You Need [web]
3. DeepSeek-V3 Technical Report [web] BOOKED
4. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning [web] BOOKED
5. A Path Towards Autonomous Machine Intelligence [web] BOOKED
6. Generative Adversarial Nets [web]
7. Adam: A Method for Stochastic Optimization [web]
8. Generative Agents: Interactive Simulacra of Human Behavior [web]
9. VALL-E 2: Neural Codec Language Models are Human Parity Zero-Shot Text to Speech Synthesizers [web]
10. Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model [web]
11. Adding Conditional Control to Text-to-Image Diffusion Models [web] + Improving Image Generation with Better Captions [web]

For research workshops, students will participate in research discussions regarding a specific topic/problem and then prepare a 2-page report on solution of the problem. The discussions will be on two topics: Multi-agent Path Finding and Hierarchical Planning, this is a list of sources:

1. R. Stern et a. Multi-Agent Pathfinding: Definitions, Variants, and Benchmarks [web]
2. J. Li, Z. Chen, Y. Zheng, S.-H. Chan, D. Harabor, P. Stuckey, H. Ma and S. Koenig. Scalable Rail Planning and Replanning: Winning the 2020 Flatland Challenge.Scalable Rail Planning and Replanning: Winning the 2020 Flatland Challenge [web]
3. H. Jiang, Y. Zhang, R. Veerapaneni, J. Li: Scaling Lifelong Multi-Agent Path Finding to More Realistic Settings: Research Challenges and Opportunities [web]
4. Keisuke Okumura: LaCAM: Search-Based Algorithm for Quick Multi-Agent Pathfinding. AAAI 2023: 11655-11662
5. Pascal Bercher, Ron Alford, Daniel Höller: A Survey on Hierarchical Planning - One Abstract Idea, Many Concrete Realizations. IJCAI 2019: 6267-6275
6. Chad Hogg, Hector Munoz-Avila, and Ugur Kuter: HTN-Maker: Learning HTNs with Minimal Additional Knowledge Engineering Required. In Proceedings of the Twenty-Third AAAI Conference on Artificial Intelligence (AAAI-08). AAAI Press.
7. Pat Langley: Learning Hierarchical Problem Networks for Knowledge-Based Planning. ILP 2022: 69-83
8. Songtuan Lin, Daniel Höller, Pascal Bercher: Modeling Assistance for Hierarchical Planning: An Approach for Correcting Hierarchical Domains with Missing Actions. SOCS 2024: 55-63
9. Songtuan Lin, Alban Grastien, Pascal Bercher: Towards Automated Modeling Assistance: An Efficient Approach for Repairing Flawed Planning Domains. AAAI 2023: 12022-12031
10. D. Nau, T.-C. Au, O. Ilghami, U. Kuter, W. Murdock, D. Wu, and F.Yaman: SHOP2: An HTN Planning System. JAIR, volume 20, pp. 379-404, 2003.
11. Pascal Bercher, Shawn Keen, Susanne Biundo: Hybrid Planning Heuristics Based on Task Decomposition Graphs. SOCS 2014: 35-43
12. Robert P. Goldman, Ugur Kuter, and Richard G. Freedman. Stable plan repair for state-space HTN planning. HPlan Workshop 2020

This is a preliminary program of the seminar.