Scholar
Jamie McGowan
Google Scholar ID: VUByCmgAAAAJ
MediaTek Research
Citations & Impact
Citations
161
H-index
3
i10-index
1
Publications
15
Co-authors
0
Publications
Resume
Academic Achievements
- 1. Exact, Tractable Gauss-Newton Optimization in Deep Reversible Architectures Reveal Poor Generalization (NeurIPS, 2024); 2. Efficient Model Compression Techniques with FishLeg (NeurIPS, Workshop on Machine Learning and Compression, 2024); 3. A Dataset for Learning Graph Representations to Predict Customer Returns in Fashion Retail (2023)
Research Experience
- Currently a Research Scientist at MediaTek Research, focusing on the training dynamics of neural networks and the interpretability of their predictions.
Education
- Background and training in Theoretical Physics, working on extending our understanding of the Standard Model in a data-driven environment.
Background
- Research Interests: Training dynamics of neural networks and interpretability of their predictions; Field: Theoretical Physics; Bio: A Research Scientist at MediaTek Research, focusing on understanding how deep neural networks learn and using this knowledge to design simpler algorithms and architectures, especially in multimodality.
Miscellany
- Personal interests include thinking about how neural networks can become plastic, how task-dependent behavior can be extracted or inserted in deep networks, and what comes after Adam.