Analyzing Attribute Dependencies

By Aleks Jakulin, Ivan Bratko. 2003

163 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Science Applications

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Science Applications

Citations in the corpus

Interpreting Classifiers through Attribute Interactions in Datasets (2017, neutral): “Tatti (2011) ordered attributes according to their dependencies while Jakulin & Bratko (2003) quantiﬁed the degree of attribute interaction and Jakulin & Bratko (2004) factorised the joint data distribution and presented a method for signiﬁcance testing of attribute interactions.”
Testing the significance of attribute interactions (2004, background): “Jakulin and Bratko (2003) suggested that a considerably high or low interaction information among attributes is a heuristic indication that the attributes interact and should not be factorized.”
Nomograms for visualizing support vector machines (2005, background): “This problem is referred to as a negative interaction or as attribute redundancy [13].”
Searching for interacting features in subset selection (2009, unknown): “They further provide in [26] a justiﬁcation of the interaction information, replace the notion of ‘high’ and ‘low’ in [28] with statistical signiﬁcance, and illustrate the signiﬁcant interactions in the form of interaction graph.”
Applying Discrete PCA in Data Analysis (2004, methodology): “An inves- 2See http://www.ai.mit.edu/∼jrennie/20Newsgroups/. tigation into the interactions here using methods of (Jakulin & Bratko, 2003) revealed that in the hierarchical model much of the interaction is up and down the hierarchy.”
Gene selection with guided regularized random forest (2012, background): “For example, the measures can not capture the exclusive OR relationship Y = XOR ( X 1 , X 2 ), in which neither X 1 nor X 2 is predictive individually, but X 1 and X 2 together can correctly determine Y [14].”
Finding Statistically Significant Attribute Interactions (2016, methodology): “[ 10] proposed a method for quantifying the degree of interaction and [11] consider factorising the joint data distribution and presented a method for testing the significance of the found attribute interactions (experiments limited to two and three-way interactions).”
Searching for Interacting Features (2007, background): “They further provide in [Jakulin and Bratko, 2004] a justification of the interaction information, and replace the notion of ‘high’ and ‘low’ in [Jakulin and Bratko, 2003] with statistical significance and illustrate the significant interactions in the form of interaction graph.”