I have a dataset of 4x24. When I am applying ANN to my dataset, I am getting different results with improper training. Although, I am doing normalization on my dataset but still I would like to know what could be going wrong in my case. Also, Where can I get a tutorial for using matlab with ANN.
Without specifics about the net or the problem, I try answer in general. You can do common pre-processing:
• Cleaning up data
• Perform outlier detection and removal
• Normalize in a smart way, to not throw away information. Be careful here, as there are many ways to normalise your data.
• Find and eliminate doublets which interfere with training
• Find appropriate dimensionality reduction
• Split data into train and test sets or do cross-validation
What exactly do you mean with a dataset of 4x24? Is that some fixed nomenclature to an existing dataset or are you considering 4x24 numbers? The latter sounds like a very small data set...
You can consider k-means or c-means clustering as pre-processing techniques.
You can also find the tutorial here :
http://www.image.ece.ntua.gr/courses_static/nn/matlab/nnet.pdf
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training