Forthcoming Presentations:

Poised-For Learning: A New Kind of Machine Learning
Selmer Bringsjord & Konstantine Arkoudas
Air Force Research Labs, Rome

We present a new form of machine learning: poised-for learning. The driving idea behind p-f learning is that, at least in theory, you could ascertain if a human had learned a domain by direct inspection of the brain, not by giving a test of performance after learning was supposed to have taken place. The trick is to make this idea precise, indeed precise enough to be implemented. P-f learning is particularly well-suited to engineering a computational system capable of learning by doing something that no such system has hitherto been able to do: namely, by reading. Accordingly, we explain p-f learning in the context of our DARPA-sponsored project to build a computational system capable of reading.