From sensor data to an automated expert – AI rule mining for personalized training feedback
Project description
We are researching to predict equine workout feedback, using human language feedback written by an equine expert. Using this feedback, we convert it into labels suitable to train a Machine Learning model to automate this process.
Context
Equine sports, including Olympic teams.
Results
We successfully converted human language into training output labels. Using this, we can automate the feedback process in a way that is transparent and explainable.
For example, we can successfully predict the workout session observations with an accuracy up to 80% for multi-label outputs, just on human language training data.