Physics
Research focused on understanding various physical systems
-
Modeling Charge Carrier Dynamics
Mie Scattering Models
FDTD
FEM
Transfer Matrix
-
Exploring Monte Carlo and analytical models for
Decay Systems
Diffusion Systems
Many Body Problems
Hopfield Networks
Ising Models
Navier Stokes
FEM
Grid based Wave Systems
Node based Wave Systems
-
Classical N-body Simulations
Computational
Researched aimed at exploring computational error analysis, exploring bias in machine learning, and finding distributed methods and environments for large scale physics simulations and models.
-
Charge Carrier Dynamics
Decay
Scattering
Population Dynamics
Network Dynamics
RNG Quality
-
Synchronization of simulations
Methods to parallelize simulations
Independent Decoupled Neural Modeling
Parallel computation on GPU and TPU
Integration of cloud, VM or other distributed systems
-
Optimization
Shallow, Narrow Networks
Minimal mechanical systems required to solve a given learning problem
Education
Research aimed at exploring methods, and teaching tools to help tomorrow STEM students utilize machine learning, and physics efficiently and understand that machine learning is just as predictable as other areas of science, given the foundations of which the models where built.
-
Use machine learning and physics models to explain communication theory
-
Exploring how to best teach that literature shows classical and quantum information theory shares much of the same underlying theory used in machine learning, namely Shannon Entropy. If many undergrads can write a simple auto encoder, Hinton et al. describes how this is related to deep concepts in information theory, in his paper AE, Helmholtz Machines.
-
Find and develop courses to help students better understand the limitations of compute, learning and other types of models, even analytical modes generate error, for example radio active decay fails to accurately predict the large deviations in very small populations.