All projects
07 / 08 · Research
Neural Networks & Permutations
GRU architectures for cycle structure prediction
- Year
- 2025
- Published
- NHSJS
- Tools
- Python, GRU models
- Outcome
- 72% accuracy, top 2% nationally
As co-author of "Optimal Neural Network Architecture for Predicting Cycle Structure in Swap Permutations" I designed GRU models and found that a two-layer architecture reached 72% accuracy.
The work earned a Distinction and placed in the top 2% at the University of Melbourne Mathematics & Statistics Research Competition.