I do plenty of ML data-analytics work in areas like financial prediction and clinical trial data analytics, even some NLP stuff like medical question answering -- for commercial projects in those domains of course I am using the same quantitative accuracy measures as everyone else. This sort of thing certainly has its place, I just don't think it's the right sort of way to measure incremental progress toward AGI...
-- ben On Mon, Mar 8, 2021 at 4:57 PM <immortal.discover...@gmail.com> wrote: > > And it's sad you don't use evaluation, because when I try something new or > tweak the code parameters, my eval tells me right away if it improves > prediction. And while it is possible to find dead ends like in gradient > descent (ex. BWT predicts data but is the wrong, way, totally...) - at least > it [helps] tell you if you implemented something good or correctly. I'm dying > over here you don't use it, all pros use it. You on my naughty list, it's an > essential tool and is proven easily, bet you 5,000$. > Artificial General Intelligence List / AGI / see discussions + participants + > delivery options Permalink -- Ben Goertzel, PhD http://goertzel.org “He not busy being born is busy dying" -- Bob Dylan ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Ta5ed5d0d0e4de96d-M0ee0a10973b6cda6fef634a5 Delivery options: https://agi.topicbox.com/groups/agi/subscription