Scholar

Maciej Wołczyk

Google Scholar ID: f6Xi7aoAAAAJ

Google

continual learningdeep learningreinforcement learning

Citations & Impact

All-time

Citations

797

H-index

i10-index

Publications

Co-authors

list available

Contact

Publications

2 items

2026

Cited

2025

Cited

Resume (English only)

Academic Achievements

- Paper 'Fine-tuning Reinforcement Learning Models is Secretly a Forgetting Mitigation Problem' accepted as a spotlight at ICML 2024
- Paper 'AdaGlimpse: Active Visual Exploration with Arbitrary Glimpse Position and Scale' accepted at ECCV 2024
- Received FNP Start scholarship in May 2024
- Workshop paper on mixing and retrieving states in State-Space Models accepted at the Next Generation Sequence Models workshop at ICML 2024 in June 2024
- Paper on the issue of forgetting in RL fine-tuning accepted as a spotlight at ICML 2024 in May 2024
- Paper on continual learning with weight interval regularization accepted to ICML 2022 as a short presentation in May 2022
- Two papers, 'Zero Time Waste' and 'Continual World', accepted to NeurIPS 2021 as poster presentations in September 2021

Research Experience

- Joined Google, Paradigms of Intelligence, and moved to Zürich in January 2025
- Talked about work on forgetting in RL fine-tuning at UCL Dark seminar in July 2024
- Co-organized the Next Generation Sequence Models workshop at ICML 2024 in June 2024
- Defended PhD thesis with distinction in February 2024
- Started work as a postdoc at IDEAS NCBR in October 2023
- Conducted a research internship with João Sacramento at ETH Zurich from March 2022 to September 2022
- Worked on imitation learning for planning in self-driving cars at Woven Planet Level-5 (previously Lyft Level-5) from April 2021 to September 2021

Education

Background

Currently a Research Scientist at the Google Paradigms of Intelligence team. Main research interests include: Sequential Decision Making, Multi-agent systems, Fine-tuning RL models, and Adaptation in foundation models.

Miscellany