Title :
Generalization and generalizability measures
Author :
Wah, Benjamin W.
Author_Institution :
Dept. of Electr. & Comput. Eng., Illinois Univ., Urbana, IL, USA
Abstract :
Defines the generalization problem, summarizes various approaches in generalization, identifies the credit assignment problem, and presents the problem and some solutions in measuring generalizability. We discuss anomalies in the ordering of hypotheses in a subdomain when performance is normalized and averaged, and show conditions under which anomalies can be eliminated. To generalize performance across subdomains, we present a measure called “probability of win” which measures the probability that one hypothesis is better than another. Finally, we discuss some limitations in using probabilities of win, and we illustrate their application in finding new parameter values for TimberWolf, a package for VLSI cell placement and routing
Keywords :
VLSI; circuit layout CAD; generalisation (artificial intelligence); intelligent design assistants; network routing; TimberWolf; VLSI cell placement; VLSI routing; anomaly elimination; credit assignment problem; generalizability measures; generalization; hypothesis ordering; machine learning; normalized averaged performance; parameter value finding; probability of win; subdomains; Artificial intelligence; Learning systems; Machine learning; Packaging machines; Production; Psychology; Routing; Search problems; Signal processing; Very large scale integration;
Journal_Title :
Knowledge and Data Engineering, IEEE Transactions on
