Differences

This shows you the differences between two versions of the page.

--- courses:ae4m33bia:guidelines_ann [2011/03/02 16:31]
kubalik
+++ courses:ae4m33bia:guidelines_ann [2015/03/08 21:00]
kubalik
@@ Line 1: / Line 1: @@
-[[courses:ae4m33bia:start|Back to the startpage]]
+[[courses:a4m33bia:start|Back to the startpage]]
-\\
-**Guidelines for ANN Assignment**
-  - Choose an appropriate dataset.
-  - Search for an optimal ANN architecture (a number of hidden layers and their neurons).
-  - Experimentally find appropriate number of epochs.
-  - Experiment with:
-    * Backpropagation,
-    * Batch Backpropagation,
-    * Backpropagation with momentum,
-    * Quickpropagation,
-    * Resilient propagation (RPRop).
-  - For backpropagation test at least 3 settings of a learning rate/step width and 3 settings of momentum constant for a version of BP with Momentum.
-  - Focus on a learning speed and ANN total error.
-  - Write a report documenting all your experiments. It should include:
-    * Description of the dataset - number and types of attributes, number of instances. For classification task also the number of classes and the class distribution.
-    * Description of a chosen ANN topology and the process of finding the optimal topology.
-    * Description of experiments.
-    * Typical learning behavior plots for chosen learning algorithms and parameter settings including the discussion.
-    * Analysis of experiments and a Conclusion.
-  - Your work does not have to exactly match the above guidelines. These should rather give you an overview of the amount of work required.
-\\
-[[courses:ae4m33bia:start|Back to the startpage]]