Packet 3: Bonus 19

This framework titles a 2013 general audience book on “Nature’s Algorithms for Learning and Prospering in a Complex World” by its coiner, Leslie Valiant. For 10 points each:
[10h] Name this framework in the theory of machine learning, often known by a three-letter acronym, which aims to provide guarantees that a learner is highly accurate most of the time.
ANSWER: probably approximately correct learning [or PAC learning]
[10e] PAC results often depend on the VC dimension, which is the largest cardinality of points that can be learned via this task. This supervised learning task aims to predict which of a discrete number of categories a point lies in.
ANSWER: classification [or word forms like classifying; accept binary classification]
[10m] PAC guarantees hinge on whether the sample type of this quantity is polynomial in the PAC terms epsilon and delta. Kolmogorov names a type of this quantity for a string that gives the shortest program that can produce it.
ANSWER: complexity [accept sample complexity or Kolmogorov complexity]
<TM, Other Science (Computer Science)> | NAFTA-Packet-3

HeardPPBE %M %H %
4014.7568%68%13%

Back to bonuses