About me

Jaromír Janisch finished his Master studies as a Software Engineer at University of West Bohemia in 2012. Until 2016 he worked in several software companies, mainly as a developer. He’s been always passionate about progress in artificial intelligence and decided to join the field as a researcher. In April 2017, he joined the AIC group at ČVUT in Prague, which led to his Ph.D. dissertation in 2024. In 2025, he’s taking a break from the AI science to develop a classic, old-school computer game at self-founded MOD42 studio.

In this blog, Jaromír strives to extract main points from sometimes complicated scientific research, reformulate them in his own words and make it more accessible for people outside the field.

Contact me at author@jaromiru.com.

Publications and Talks

ICML-2018 (July 2018, Stockholm, Sweden)

Presentation at the Adaptive Learning Agents (ALA) workshop.

AAAI-2019 (January 2019, Hawaii, USA)

Talk on the main conference:
J. Janisch, T. Pevný, V. Lisý - Classification with Costly Features using Deep Reinforcement Learning - paper / slides / poster / code / blog

Machine Learning Journal (March 2020)

J. Janisch, T. Pevný, V. Lisý - Classification with Costly Features as a Sequential Decision-Making Problem - paper / code

ICML-2021 (July 2021, online)

Two posters at the RL4RealLife workshop:
J. Janisch, T. Pevný, V. Lisý - Symbolic Relational Deep Reinforcement Learning based on Graph Neural Networks - paper / code / poster
J. Janisch, T. Pevný, V. Lisý - Hierarchical Multiple-Instance Data Classification with Costly Features - paper / code / poster

ESORICS-2023 (September 2023, Hague, Netherlands)

Presentation at the Security and Artificial Intelligence (SECAI) workshop:
J. Janisch, T. Pevný, V. Lisý - NASimEmu: Network Attack Simulator & Emulator for Training Agents Generalizing to Novel Scenarios - paper / code

Machine Learning Journal (May 2024)

J. Janisch, T. Pevný, V. Lisý - Classification with Costly Features in Hierarchical Deep Sets - paper / code

Ph.D. Dissertation (2024)

Applications of Deep Reinforcement Learning in Practical Sequential Information Acquisition Problems - pdf