Research Intern – Machine Learning for Suspension Control
As a Research Intern at the Institute for Gravitational Research, I implemented machine learning approaches for active suspension control in gravitational-wave detector systems. My primary responsibilities included using reinforcement learning and convolutional neural networks for state estimation and control within these physics-based models. The work involved simulating models, preparing and validating data, and evaluating the effectiveness of machine learning solutions within real experimental setups. • Applied PPO and CNN architectures to experimental and simulated data. • Prepared, cleaned, and validated data sets for AI-based analysis and reinforcement learning. • Evaluated performance outcomes by comparing model predictions to measured physical results. • Collaborated with experts from Caltech, Google DeepMind, and INFN to verify results.