About 30 international professors and young researchers discussed the latest developments in Deep Reinforcement Learning for dynamic decision problems in logistics and healthcare in a 3-day workshop at our TUM Campus Heilbronn.
The workshop was organized by professors Gudrun Kiesmüller, Jingui Xie and Stefan Minner and included presentations and interactive sessions, such as a research pitch and a coding challenge.
“As a PhD student, I presented my research project and received a lot of valuable feedback. I also discussed my working paper with two experienced researchers, which was a great exchange of ideas and I could learn from their detailed suggestions,” said Yihua Wang.
The workshop focused on applying Deep Reinforcement Learning (DRL) to decision-making problems in inventory management, transportation, production, auction games, and healthcare. Due to the 𝘂𝗻𝗶𝗾𝘂𝗲 𝗰𝗵𝗮𝗿𝗮𝗰𝘁𝗲𝗿𝗶𝘀𝘁𝗶𝗰𝘀 𝗼𝗳 𝘁𝗵𝗲𝘀𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺𝘀, it is challenging to adapt reinforcement learning to efficiently solve real-world problems. Typically, the problems are high-dimensional and have large uncertainties, which complicates the learning process.
“I massively enjoyed the workshop, and the presentations by senior faculty and Ph.D. students were truly thought-provoking. For example, I counted four ingenious and generalizable ideas for dealing with large action spaces: Ideas that will contribute to further adoption of DRL in practical contexts. It was great catching up with collaborators on the subject, but it was perhaps even more valuable to connect with so many new people who bring exciting new ideas to the area.“ says Prof. Willem van Jaarsveld of Eindhoven University of Technology.
Key learnings include:
Some real-world problems can be solved better with smart other heuristics instead of applying DRL
New promising ideas have been developed to solve problems with DRL and large action spaces
Good domain knowledge is needed to solve real-world problems with DRL
Thank you to all participants for your great contributions!
About 30 international professors and young researchers discussed the latest developments in Deep Reinforcement Learning for dynamic decision problems in logistics and healthcare in a 3-day workshop at our TUM Campus Heilbronn.
The workshop was organized by professors Gudrun Kiesmüller, Jingui Xie and Stefan Minner and included presentations and interactive sessions, such as a research pitch and a coding challenge.
“As a PhD student, I presented my research project and received a lot of valuable feedback. I also discussed my working paper with two experienced researchers, which was a great exchange of ideas and I could learn from their detailed suggestions,” said Yihua Wang.
The workshop focused on applying Deep Reinforcement Learning (DRL) to decision-making problems in inventory management, transportation, production, auction games, and healthcare. Due to the 𝘂𝗻𝗶𝗾𝘂𝗲 𝗰𝗵𝗮𝗿𝗮𝗰𝘁𝗲𝗿𝗶𝘀𝘁𝗶𝗰𝘀 𝗼𝗳 𝘁𝗵𝗲𝘀𝗲 𝗽𝗿𝗼𝗯𝗹𝗲𝗺𝘀, it is challenging to adapt reinforcement learning to efficiently solve real-world problems. Typically, the problems are high-dimensional and have large uncertainties, which complicates the learning process.
“I massively enjoyed the workshop, and the presentations by senior faculty and Ph.D. students were truly thought-provoking. For example, I counted four ingenious and generalizable ideas for dealing with large action spaces: Ideas that will contribute to further adoption of DRL in practical contexts. It was great catching up with collaborators on the subject, but it was perhaps even more valuable to connect with so many new people who bring exciting new ideas to the area.“ says Prof. Willem van Jaarsveld of Eindhoven University of Technology.
Key learnings include:
Some real-world problems can be solved better with smart other heuristics instead of applying DRL
New promising ideas have been developed to solve problems with DRL and large action spaces
Good domain knowledge is needed to solve real-world problems with DRL
Thank you to all participants for your great contributions!
