DynaDojo ML Benchmarking Platform
Research

DynaDojo ML Benchmarking Platform

Expanding benchmarking capabilities for dynamical systems

Role

Research Intern - Data integration, visualization development, OOD detection

Expanded DynaDojo benchmarking library from 20 to 130+ systems by integrating Gilpin's database of chaotic dynamical systems with precomputed time series data. Developed 3D visualization capabilities to enable algorithm evaluation on real-world systems with unknown equations. Built Out-of-Distribution detection framework using PCA and Haar wavelet transforms, improving detection performance by 60%.

Tech Stack

PythonNumPyPCAHaar Wavelets3D Visualization

Gallery