What is "machine learning"?
-Machine learning describes the way that machines learn from repeated examples. If they see enough examples of a certain object (such as the letter "A"), it will pick up small patterns and learn to recognize things that it has never even seen before.
How did the IBM team use this? How was this advantageous?
-The IBM team dumped thousands of old, used Jeopardy questions, along with their answers, into Watson's database. They counted on the computer to use machine learning to find patterns between these questions and their correct answers, and help it answer future questions correctly. It could do this by combining the patterns that it had picked up from the questions with the huge wealth of encyclopedic knowledge that was programmed into its system. This was advantageous because the team did not have to try to input all of the common sense knowledge, and knowledge of puns and other strange wordings.
What is "Empirical Skepticism"? How does it relate to machine learning?
-Empirical skepticism is basically the reason why the human brain is still superior to the robot. Our brains can actually think, and don't run off of programmed data. Empirical skepticism is the idea that certain things have different meanings, and each particular meaning can only be deciphered by calling to mind a past experience or prior knowledge, and analyzing context.
This relates to machine learning because it is the way that IBM is trying to create their robot, by programming tons of prior knowledge, and artificial past experiences into it.
No comments:
Post a Comment