Local Novelty Scorer

This node calculates a local KNFST-model based on the nearest neighbors in the kernel space for every single data point that is to be tested. The corresponding source code is a Java adaption of the Matlab code provided by Bodesheim et al. for their paper "Local Novelty Detection in Multi-class Recognition Problems".

Options

Training Columns

The columns which should be used for training (right now only numeric columns can be used for training)

Class Column

The column containing the class of the training samples

Kernel

Choose the Kernel that should be used

• HIK - Histogram intersection kernel
• EXPHIK - Exponential variant of the HIK
• RBF - Radial basis function kernel (with parameter sigma)
• Polynomial - Polynomial kernel (with parameters gamma, bias and power)

Number of Neighbors

Select the number of neighbors that should be used for training the local models.

Normalize Novelty Scores

Check to normalize the Novelty Scores by the minimum distance between target points.

Sort Training Table

Check to sort training table by class prior to calculation of the local models. Only use if training table is not already sorted by class.

Input Ports

KNFST model

Test Data

Output Ports

Novelty Scores

Installation

To use this node in KNIME, install KNIME Active Learning from the following update site:

A zipped version of the software site can be downloaded here.

You don't know what to do with this link? Read our NodePit Product and Node Installation Guide that explains you in detail how to install nodes to your KNIME Analytics Platform.

Developers

You want to see the source code for this node? Click the following button and we’ll use our super-powers to find it for you.