Scholar
Mihaela Rosca
Google Scholar ID: MxkDwD0AAAAJ
Deepmind
Citations & Impact
Citations
5,338
H-index
18
i10-index
18
Publications
20
Co-authors
0
Publications
Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
- Published multiple papers such as 'On a continuous time model of gradient descent dynamics and instability in deep learning', 'Why neural networks find simple solutions: the many regularizers of geometric complexity'; contributed to a book chapter on GANs in Kevin Murphy's Probabilistic Machine Learning; delivered talks at various academic conferences including Mediterranean Machine Learning Summer School, UCL, etc.
Research Experience
- Before joining DeepMind as a Research Engineer in 2016, worked as a Software Engineer at Google Zurich focusing on NLP using neural networks.
Education
- PhD from UCL, supervised by Prof. Marc Deisenroth, passed viva without corrections in May 2023; 4-year Computing MEng degree from Imperial College London, graduated with first-class honors and an excellence award for outstanding overall performance in 2014.
Background
- Senior Staff Research Scientist at DeepMind, currently working on training LLMs as part of Gemini. Professional interests include generative models, reinforcement learning, natural language processing, and scalable and safe machine learning. Currently focused on understanding optimization in deep learning.
Miscellany
- Contributed to open-source projects like Optax (JAX optimization library), Monte Carlo Gradient estimation in machine learning source code, Deep Compressed Sensing source code, and more.
Co-authors