Neurosymbolic AI
My research focuses on the mathematical foundations of neurosymbolic AI. I study the interaction between learning and reasoning using probabilistic circuits and Bayesian network representations, and investigate reasoning shortcuts in neurosymbolic systems.
I am also interested in uncertainty estimation in AI and the study of emergence in complex systems.
No published articles listed yet.
An investigation of conditional independence and deterministic bias in neurosymbolic AI, examining whether the tendency to favor a single valid solution can be attributed to independence assumptions.
Repository links will be added here.
A study of physics-informed neural networks, Fourier neural operators, and the Deep Ritz method for solving partial differential equations and variational problems.
Bayesian inference in linear normal models, with implementations of Markov chain Monte Carlo and approximate Bayesian computation in R.