3 ms·
The part of a system which monitors these changes (well, it mostly monitors single number, performance on a test set) isn't represented in the network itself.
by red75prime 9y ago
The part of a system which monitors these changes (well, it mostly monitors single number, performance on a test set) isn't represented in the network itself.
You can train dialogue NLP system as much as you wish, but you'll never get answers to a question "Have you learned this training set already?" which correlate with whether performance on the test set have plateaued.
I wouldn't call it awareness. We are certainly aware that one part of the system monitors another part, the system itself isn't.
- p1esk 9y agoI'm still not following you. What does the question "have you learned this training set already?" even mean? You measure a training set accuracy, and you compare it against the corresponding test set accuracy. It's trivial to add the code that monitors this correlation to your model. Then you can even have the model make adjustments to the learning process based on what it observes (e.g. deploy more regularization to reduce overfitting, etc). So, again, what are you talking about?
- red75prime 9y agoI was trying to show that humans can't be purely relying on certain kinds of ML algorithms because those algorithms can't provide properties which we observe in our thought processes. > It's trivial to add the code that monitors this correlation to your model. Yes, but while it's not there we can't say that the model is aware of the process of its own learning. The question serves as an experimental test of system's awareness of the process of its own learning. All in all, I tried to make poorly defined terms like "awareness" and "the way humans think" a little bit more technical.