Alex Wozniakowski
Currently, I am a software engineer at Apple; and I have interests in applied and computational mathematics, machine learning, and statisics research and applications.
Previously, I obtained my Ph.D. in mathematics from Nanyang Technological University (NTU) in 2022. In my dissertation, I developed a hybrid of additive and varying coefficient models that enables a fitting algorithm to leverage prior knowledge encoded in predictions from a base model to build a complex model.
- Notably, this approach resulted in a generalization of gradient boosting, known as base boosting.
- That is, gradient boosting initializes its fit of an additive model at a constant model then iteratively improves upon it.
- Base boosting, on the other hand, initializes its fit of the hybrid additive and varying coefficient model at the base model, which may be a non-constant model, then iteratively improves upon it.
- Empirically, base boosting outperformed popular boosting libraries and Google’s quantum physics model on quantum device and Taylor series data sets.
Additionally, I am an original member of the Mathematical Picture Language Project at Harvard University, where I established the link between the mathematics of planar algebras and the science of quantum information and computation.
- This led to a collaboration in which multiple charged string picture languages were developed.