“The component of prediction tasks that can be easily automated is the one that does not involve any expert knowledge. Prediction tasks require expert knowledge to specify the scientific question (what input and what outputs) and to identify/generate relevant data sources. (The extent of expert knowledge varies across different prediction tasks.18) However, no expert knowledge is required for prediction after the inputs and outputs are specified and measured in a particular dataset. At this point, a machine learning algorithm can take over the data analysis to deliver a mapping and quantify its performance. The resulting mapping may be opaque, as in many deep learning applications, but its ability to map the inputs to the outputs with a known accuracy is not in question.” Miguel A. Hernán, John Hsu, Brian Healy ( July 12, 2018 )
Like this:
Like Loading…
Related