Human-centric machine learning.

Опубликовано 22 июня 2016, 1:42

How can we enable a human being to carry out machine learning that is of value to themselves and to others? Not, I would argue, by reducing that human to an operative that simply selects some data and fires off a giant automated algorithm! Rather, we need to enable that human to interact creatively with every step of the machine learning process, and with other humans, in order to extract whatever it is that they want and need: understanding, new protocols for gathering data, predictive models, whatever. Illustrated with examples based on many years of experience of using machine learning within a particular domain – ecological and environmental science – I’ll outline 6 key barriers that I believe need to be broken in order to realize the true societal potential of machine learning.