epistemic perspectives

Informing engineering from the cognitive sciences

About Larry and his work

My name is Larry Muhlstein

I grew up in Pennington New Jersey, a quaint suburban town surrounded by farms, wilderness, universities, and history. Slowly migrating westward with a four-year pitstop in Cleveland Ohio, I eventually found myself in sunny La Jolla.

I am currently a Cognitive Science PhD student at UCSD. I am advised by a computer scientist, Julian McAuley

I work on:

  • Omnidirectional neural networks
  • Recurrent neural networks for modeling heterogeneous data (multiple sequences with multiple contextual attributes)
  • Atemporal (feedforwardish) neural networks for modeling multimodal data with noise
  • Advances in learning theory that make it both more theoretically general and more relevant to practitioners of machine learning
  • What might be called meta cognitive-science which blends the philosophy of science with the domain-centric questions in cognitive science with real suggestions about how real researchers can really learn lots of real things by performing science really well
  • Understanding how the assumptions and biases necessarily present in theories and models for them to be able to predict anything affect the quality of their resultant predictions
  • Building computational models of how we understand what people mean by what they say
  • Understanding how it is possible for humans to communicate and to learn to communicate given the extreme poverty of knowledge (and a relative wealth of weak evidence)
  • Deep natural language processing