Opened 6 years ago

Closed 6 years ago

#1474 closed feature request (done)

Wrapper for neural network ensembles implemented in alglib

Reported by: gkronber Owned by: gkronber
Priority: high Milestone: HeuristicLab 3.3.6
Component: Algorithms.DataAnalysis Version: 3.3.6
Keywords: Cc:

Description


Change History (19)

comment:1 Changed 6 years ago by gkronber

  • Priority changed from medium to high

comment:2 Changed 6 years ago by gkronber

  • Milestone changed from HeuristicLab 3.3.x Backlog to HeuristicLab 3.3.6

comment:3 Changed 6 years ago by gkronber

  • Milestone changed from HeuristicLab 3.3.6 to HeuristicLab 3.3.x Backlog

comment:4 Changed 6 years ago by gkronber

  • Milestone changed from HeuristicLab 3.3.x Backlog to HeuristicLab 3.3.6
  • Status changed from new to accepted

comment:5 Changed 6 years ago by gkronber

  • Version changed from branch to 3.3.5

comment:6 Changed 6 years ago by gkronber

r6577: added first implementation of neural networks for regression wrapper for alglib.

comment:7 Changed 6 years ago by gkronber

r6578: added parameters for neural network regression algorithm.

comment:8 Changed 6 years ago by gkronber

r6579: implemented neural networks for classification.

comment:9 Changed 6 years ago by gkronber

r6580: added implementations for regression and classification with neural network ensembles (wrappers for alglib).

comment:10 Changed 6 years ago by gkronber

  • Owner changed from gkronber to mkommend
  • Status changed from accepted to reviewing

comment:11 Changed 6 years ago by mkommend

  • Owner changed from mkommend to gkronber
  • Status changed from reviewing to assigned

Reviewed the source of neural network data analysis algorithm. During testing it noticed that the default parameter values are quite different for neural networks and neural ensemble networks, and I wonder if this was on purpose. Additionally the ensemble network parameter values yield rather poor results (tested on poly-10 and default classification problem) compared to standard neural networks.

Last edited 6 years ago by mkommend (previous) (diff)

comment:12 Changed 6 years ago by gkronber

  • Status changed from assigned to accepted

comment:13 Changed 6 years ago by gkronber

r6718: fixed bug in alglib that resulting in random models when using the early stopping training method for multi layer perceptrons.

comment:14 Changed 6 years ago by gkronber

r6719: removed range limit for MLP output

comment:15 Changed 6 years ago by gkronber

After r6718:r6719 the performance of both algorithms, normal MLP and MLP ensemble, are comparable.

comment:16 Changed 6 years ago by gkronber

r6720: added (conditional) parameter hiding

comment:17 Changed 6 years ago by gkronber

  • Owner changed from gkronber to mkommend
  • Status changed from accepted to reviewing

comment:18 Changed 6 years ago by mkommend

  • Owner changed from mkommend to gkronber
  • Status changed from reviewing to readytorelease

comment:19 Changed 6 years ago by swagner

  • Resolution set to done
  • Status changed from readytorelease to closed
  • Version changed from 3.3.5 to 3.3.6
Note: See TracTickets for help on using tickets.