MLP implementation

[ddveloper]

Multi-layer perception method

[ddveloper]

this can be separated into several tasks: 1) read inputs and labels 2) normalization inputs 3) split to train, validate and test set 4) training wrapper (for experiment different eta, momentum, hidden nodes) 5) MLP logic (take train and validate set as inputs) 6) test result (run on test set, build confusion matrix, tp, tf, np, nf) 7) plot result (if we used pylab)

[daltonbohning]

Tasks to be done with MLP:

• [x] Read inputs and labels
• [x] Normalize inputs
• [x] Split to train, validate and test set
• [x] Training wrapper (for experiment different eta, momentum, hidden nodes)
• [x] MLP logic (take train and validate set as inputs)
• [x] Test result (run on test set, build confusion matrix, tp, tf, np, nf)
• [ ] Use pylab to plot the results