• Publications
  • Influence
On Graph Kernels: Hardness Results and Efficient Alternatives
TLDR
We propose a generalized family of graph kernels based on walks which includes the kernels proposed in [4] as special cases while still being polynomially computable. Expand
  • 809
  • 89
  • PDF
Multi-Instance Kernels
TLDR
We propose a kernel on multi-instance data that can be shown to separate positive and negative sets under natural assumptions. Expand
  • 509
  • 75
  • PDF
Subgroup Discovery with CN2-SD
TLDR
This paper presents a subgroup discovery algorithm, CN2-SD, developed by modifying parts of the classification rule learner: its covering algorithm, search heuristic, probabilistic classification of instances, and evaluation measures. Expand
  • 382
  • 62
  • PDF
The Geometry of ROC Space: Understanding Machine Learning Metrics through ROC Isometrics
TLDR
This paper provides a derivation of ROC space from first principles through 3D ROCspace and the skew ratio, and redefines metrics in these dimensions. Expand
  • 268
  • 38
  • PDF
Rule Evaluation Measures: A Unifying View
TLDR
This paper provides a systematic analysis of rule evaluation measures used in machine learning and knowledge discovery. Expand
  • 419
  • 35
  • PDF
Learning Decision Trees Using the Area Under the ROC Curve
TLDR
We show how a single decision tree can represent a set of classifiers by choosing different labellings of its leaves, or equivalently, an ordering on the leaves. Expand
  • 329
  • 21
  • PDF
Database Dependency Discovery: A Machine Learning Approach
TLDR
In this paper we address this problem of dependency discovery, understood as characterising the set of dependencies that are satisfied by a given collection of data. Expand
  • 164
  • 21
  • PDF
ROC ‘n’ Rule Learning—Towards a Better Understanding of Covering Algorithms
TLDR
This paper provides an analysis of the behavior of separate-and-conquer or covering rule learning algorithms by visualizing their evaluation metrics and their dynamics in coverage space, a variant of ROC space. Expand
  • 242
  • 17
  • PDF
Naive Bayesian Classification of Structured Data
TLDR
In this paper we present 1BC and 1BC2, two systems that perform naive Bayesian classification of structured individuals. Expand
  • 139
  • 14
  • PDF
...
1
2
3
4
5
...