MDP Modular toolkit for data processing
Download

MDP Ranking & Summary

MDP Tags

data analyze process data analysis process data data processor

MDP Description

Modular toolkit for data processing MDP is a Python data processing framework. Implemented algorithms include: Independent Component Analysis, Slow Feature Analysis, Principal Component Analysis, Independent Slow Feature Analysis, and many more.NOTE: MDP is licensed and distributed under the terms of the GNU Library or Lesser General Public License (LGPL). Requirements: · Python What's New in This Release: · Added online detection of numerical backend, parallel python support, symeig backend and numerical backend to the output of unit tests. Should help in debugging. · Integration of the cutoff and histogram nodes. · Fixed bug in parallel flow (exception handling). · Fixed bug in LLENode when output_dim is a float. Thanks to Konrad Hinsen. · Fixed bugs in parallel flow for multiple schedulers. · Fixed a bug in layer inverse, thanks to Alberto Escalante. · Added a LinearRegressionNode. · PCANode does not complain anymore when covariance matrix has negative eigenvalues iff svd==True or reduce==True. If output_dim has been specified has a desired variance, negative eigenvalues are ignored. Improved error message for SFANode in case of negative eigenvalues, we now suggest to prepend the node with a PCANode(svd=True) or PCANode(reduce=True). · Migrated from old thread package to the new threading one. Added flag to disable caching in process scheduler. There are some breaking changes for custom schedulers (parallel flow training or execution is not affected). · Added svn revision tracking support. · Removed the copy_callable flag for scheduler, this is now completely replaced by forking the TaskCallable. This has no effect for the convenient ParallelFlow interface, but custom schedulers get broken. · Implemented caching in the ProcessScheduler. · make_parallel now works completely in-place to save memory. · Added container methods to FlowNode. · Added CrossCovarianceMatrix with tests. · Added IdentityNode. · Added a helper function in hinet to directly display a flow HTML representation. · Allow output_dim in Layer to be set lazily. · Added total_variance to the nipals node. · Always set explained_variance and total_variance after training in PCANode. · Modified symrand to really return symmetric matrices (and not only positive definite). Adapted GaussianClassifierNode to account for that. Adapted symrand to return also complex hermitian matrices. · Fixed one problem in PCANode (when output_dim was set to input_dim the total variance was treated as unknown). Fixed var_part parameter in ParallelPCANode. · Added var_part feature to PCANode (filter according to variance relative to absoute variance). · Fixed missing axis arg in amax call in tutorial. Thanks to Samuel John! · Fixed the empty data iterator handling in ParallelFlow. Also added empty iterator checks in the normal Flow (raise an exception if the iterator is empty). · Modified pca and sfa nodes to check for negaive eigenvalues in the cov matrices · symeig integrated in scipy, mdp can use it from there now. · Added ParallelFDANode. · Updated the train callable for ParallelFlow to support additional arguments. · Rewrite of the make parallel code, now supports hinet structures. · Rewrite of the hinet HTML repesentation creator. Unfortunately this also breaks the public interface, but the changes are pretty simple. · Shut off warnings coming from remote processes in ProcessScheduler · Fixed problem with overwriting kwargs in the init method of ParallelFlow. · Fixed pretrained nodes bug in hinet.FlowNode. · Fixed critical import bug in parallel package when pp (parallel python library) is installed.

MDP Related Software