fda¶

Module: `missions.nodes.spatial_filtering.fda`¶

Fisher’s Discriminant Analysis and variants for spatial filtering

Inheritance diagram for pySPACE.missions.nodes.spatial_filtering.fda:

class pySPACE.missions.nodes.spatial_filtering.fda.FDAFilterNode(retained_channels=None, load_path=None, **kwargs)[source]¶

Reuse the implementation of Fisher’s Discriminant Analysis provided by mdp

This node implements the supervised fisher’s discriminant analysis algorithm for spatial filtering.

Parameters

retained_channels:

Determines how many of the FDA pseudo channels are retained. Default is None which means “all channels”.

(optional, default: None)

load_path:

An absolute path from which the FDA filter is loaded. If not specified, this matrix is learned from the training data.

(optional, default: None)

Exemplary Call

-
    node : FDAFilter
    parameters:
        retained_channels : 42

POSSIBLE NODE NAMES:
Author:	Jan Hendrik Metzen (jhm@informatik.uni-bremen.de)
Created:	2010/02/17
	FDAFilterNode FDAFilter FDA
POSSIBLE INPUT TYPES:
	TimeSeries

Class Components Summary

`_execute`(data)	Execute learned transformation on data.
`_stop_training`([debug])
`_train`(data, label)	Remember data and label for later learning of filters.
`input_types`
`is_supervised`()	Returns whether this node requires supervised training
`is_trainable`()	Returns whether this node is trainable.
`store_state`(result_dir[, index])	Stores the projection in the given directory result_dir

is_supervised()[source]¶: Returns whether this node requires supervised training

_train(data, label)[source]¶: Remember data and label for later learning of filters.

store_state(result_dir, index=None)[source]¶: Stores the projection in the given directory result_dir