Raymond Wynne

Object Detection, Large Language Models, Vision Language Models

- Machine learning research for fundamental particle physics with the CMS detector - Integrating graph neural network particle tracking pipeline into CMS software environment - Developing novel anomaly detection technique based on geometric goodness-of-fit statistical tests with systematic uncertainties

- Developed and bench-marked a novel graph-based algorithm for gene set enrichment analyses - Extended novel MultiModal-VAE for unsupervised clustering of multi-modal datasets (patent pending) - Applied causal machine learning to multi-modal genomics data, benchmarking results with novel metrics - Developed flexible statistical analysis pipeline for large-scale brain-behavior ordinal confounder studies

- Developed quark/gluon tagging algorithm via graph neural networks - Developed model agnostic anomaly detection algorithm for new physics searches at the LHC - Anomaly detection algorithm, entirely unsupervised, boosts event-excess significance by factor of 3

- Designed and synthesized custom FPGA hardware for Lattice Quantum Chromodynamics algorithms - Implemented advanced numerical linear algebra algorithm to solve Dirac equation - Accelerated expensive matrix-matrix and matrix-vector multiplications with hardware kernel

- Used machine learning regression techniques to predict redshift values from high dimensional feature space - Extensive use of machine learning libraries in R for advanced regression modeling and ensemble learning - Parallelized 100-fold cross validation scheme: reduced runtime by 3 hours

-Developed robust model to determine unknown parameters of gravitationally lensed quasars through image positions to deduce if a system is a random quartet or a gravitationally lensed system -Discovered the images lie at the intersection of a hyperbola and an ellipse, giving an analytic solution to a formerly computationally intractable problem