Corey Clark Speaking at Big Data and AI Conference

Article Link: Constructivist Augmented Machine Learning (CAML): Using video games and human-in-the-loop techniques to directly transfer human knowledge into ML models

Constructivism is a broader learning theory which incorporates the addition of knowledge gained from past experiences as well as social interaction and collaboration with an expert. This allows for students to learn from an instructor’s past experience and knowledge as a supplement to the experimentation process. Adding CAML to an existing ML pipeline can improve model accuracy, compress model size, or allow model improvement in absence of large data sets. This talk will show examples of CAML being used to guide ML model when analyzing medical and satellite imagery as well as knowledge transfer directly into the Leela Zero deep learning model (Open Source version AlphaGo Zero).

Share