Learning non-linearly separable boolean functions with linear threshold unit trees and madaline-style networks Export

@inproceedings{citeulike:3773482, abstract = {This paper investigates an algorithm for the construction of decisions trees comprised of linear threshold units and also presents a novel algorithm for the learning of nonlinearly separable boolean functions using Madalinestyle networks which are isomorphic to decision trees. The construction of such networks is discussed, and their performance in learning is compared with standard BackPropagation on a sample problem in which many irrelevant attributes are introduced. Littlestone\&\#039;s Winnow algorithm is also explored within this architecture as a means of learning in the presence of many irrelevant attributes. The learning ability of this Madaline-style architecture on non-optimal (larger than necessary) networks is also explored.}, author = {Sahami, Mehran}, booktitle = {In AAAI-93 [2}, citeulike-article-id = {3773482}, citeulike-linkout-0 = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.48.4238}, citeulike-linkout-1 = {http://robotics.stanford.edu/users/sahami/papers-dir/AAAI-93.pdf}, keywords = {decisiontree, linear-threshold-units, winnow-hybrid}, pages = {335--341}, posted-at = {2008-12-11 10:50:16}, priority = {0}, title = {Learning non-linearly separable boolean functions with linear threshold unit trees and madaline-style networks}, url = {http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.48.4238}, year = {1993} }

TY - CONF ID - citeulike:3773482 L3 - citeulike-article-id:3773482 N2 - This paper investigates an algorithm for the construction of decisions trees comprised of linear threshold units and also presents a novel algorithm for the learning of nonlinearly separable boolean functions using Madalinestyle networks which are isomorphic to decision trees. The construction of such networks is discussed, and their performance in learning is compared with standard BackPropagation on a sample problem in which many irrelevant attributes are introduced. Littlestone's Winnow algorithm is also explored within this architecture as a means of learning in the presence of many irrelevant attributes. The learning ability of this Madaline-style architecture on non-optimal (larger than necessary) networks is also explored. EP - 341 TI - Learning non-linearly separable boolean functions with linear threshold unit trees and madaline-style networks T2 - In AAAI-93 [2 SP - 335 KW - decisiontree KW - linear-threshold-units KW - winnow-hybrid AU - Sahami, Mehran PY - 1993/// UR - http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.48.4238 ER -