Perceptual Machine Learning
Cognitive Audio, Probilistic Programming, and ML on the Edge
David worked extensively with Ishwarya Ananthabhotla developing 'Cognitive Audio' models-- pushing our perceptual models towards phenomenological, stochastic representations (and away from low-level models based on simple statistics like loudness and pitch). They have published extensively on the topic, which has been covered by NPR in 2019.
David has also spent time working on audio perception deep learning at the edge, publishing his work as a researcher with Google AI in Zurich.
David's machine learning work has drifted more and more towards PPLs, with guidance from Jan-Willem van de Meent, one of the early pioneers of PPLs including Anglican and Pyro. Probilistic Programming Languages elevate stochastic primates and generalize numerical approaches to inference.
David has also spent time working on audio perception deep learning at the edge, publishing his work as a researcher with Google AI in Zurich.
David's machine learning work has drifted more and more towards PPLs, with guidance from Jan-Willem van de Meent, one of the early pioneers of PPLs including Anglican and Pyro. Probilistic Programming Languages elevate stochastic primates and generalize numerical approaches to inference.
