Answered
How to calculate prediction interval of ANN?
Assuming you mean a NARXNET, there is no analytic formula. It depends on the significant cross-correlations of the input and ...

9 years ago | 1

| accepted

Answered
how to plot MSE while using newpnn
NEWPNN is a classifier which does not involve training. See help newpnn doc newpnn type newpnn Hope this helps. ...

9 years ago | 0

| accepted

Answered
Can spread value in the newpnn code be a vector?
The spread must be scalar.

9 years ago | 0

| accepted

Answered
Neural Network Regeneration Based on Result Weight and Bias
You didn't take into account the default normalization of inputs and targets followed by the denormalization of the output. H...

9 years ago | 1

| accepted

Answered
Multistep ahead prediction using NARX/NAR?
The ph_dataset is not a good example. Choose one of the others from help nndatasets doc nndatasets Hope this helps. ...

9 years ago | 0

Answered
How can I use neural network function generated by Matlab?( I use R2015a)
As stated above, the timeseries is not stationary. See http://www.mathworks.com/matlabcentral/answers/302908-narxnet-with-mu...

9 years ago | 0

Answered
ALWAYS SAME RESULT WITH NEURAL NETWORK SCRIPT
The only way for this to happen with 2014a is for the RNG to be initialized to the same state at the beginning. Hope this hel...

9 years ago | 1

Answered
GRNN/PNN vs RBF
NEWRB can be used for both REGRESSION and CLASSIFICATION (I have done both zillions of times). I cannot think of any case whe...

9 years ago | 0

Answered
Error: Undefined operator '-' for input arguments of type 'cell'.
Cut and paste into the command line a = 10, b=5 A = num2cell(a) B = num2cell(b) c = b-a C = gsubtract(B,A) C = B-A ...

9 years ago | 1

Answered
How can I determine if I need a time series or a fitting neural network?
For a timeseries function points at neighboring times are highly correlated so that the autocorrelation function is nonrandom an...

9 years ago | 0

| accepted

Answered
Neural Network for predictions. How to improve it
You seem to be wasting a lot of time and space writing code that is unnecessary because you do not rely on defaults. My experien...

9 years ago | 1

| accepted

Answered
Small data train ANN
Use k-fold Cross validation for a range of k values and multiple random initial weights for each value of k.. *Thank you fo...

9 years ago | 0

| accepted

Answered
About neural net fitting (at neural network toolkit)
IW = net.IW B = net.B LW = net.LW *Thank you for formally accepting my answer* Greg

9 years ago | 0

| accepted

Answered
Calculate velocity and acceleration in each time step in neural network
You are implicitly assuming that the summary statistics of the second half are the same as those of the first half. Why not p...

9 years ago | 0

Answered
whether the number of nodes in the pattern layer (which is equal to the the number of input samples) in the General Regression Neural network architecture causes over fitting of the model?
OVERFITTING is when the number of unknown weights, Nw, exceeds the number of training equations, Ntrneq. For N pairs of I-dimens...

9 years ago | 0

Answered
Feed forward NN is taking much time with my current dataset and output is not matching with target
With "I"nput and "O"utput target matrices with sizes [ I N ] = size(input) [ O N ] = size(target) Remove rows from BOTH...

9 years ago | 0

| accepted

Answered
Getting NaN values in neural network weight matrices
Remove all input and corresponding output vectors if EITHER OR BOTH contain NaNs. Hope this helps *Thank you for formally ...

9 years ago | 1

| accepted

Answered
GRNN/PNN vs RBF
I do not recall any significant difference between NEWPNN, NEWGRNN & NEWRBE They can ALL be replaced by the more flexible ...

9 years ago | 0

Answered
Generalization in the ANN
> Can we conclude that the larger network learnt too much from the examples given during the training, thus loosing the capabi...

9 years ago | 0

Answered
Feed forward NN is taking much time with my current dataset and output is not matching with target
The next time think twice before posting, WITHOUT WARNING, a MEGABYTE data set with NaNs and zero variance rows.

9 years ago | 0

Answered
Accuracy of neural network in k fold cross validation
Typically, the test subset error rate is the only one that is considered to be UNBIASED because it is not directly involved in d...

9 years ago | 0

| accepted

Answered
Neural Networks Test set
0.7/0.15/0.15 train/val/test random datadivision is a design default which can be overwritten via one of the alternates. See...

9 years ago | 0

Answered
How to improve the performance (error) of my neural network?
Typically, initial weights and trn/val/tst datadivision are random and it is not known, apriori, how many hidden nodes are optim...

9 years ago | 0

Answered
Fitting a Function with trainlm to find function parameters?
Think of a neural net as a mapping between I-dimensional "I"nputs and O-dimensional "O"utputs achieved by using N pairs of I-dim...

9 years ago | 0

Answered
Best traing performance value?
You are putting the emphasis on the wrong thing. The performance estimate of the training subset is highly biased because yo...

9 years ago | 0

| accepted

Answered
MLP Neural network and k-fold cross validation
No. Your code is not correct. You have chosen a HIGH-DIMENSIONAL-CLASSIFICATION DATASET for which that version of my code is inn...

9 years ago | 0

| accepted

Answered
One-Class classifier using Neural Network
Use an RBF or EBF net. Generate points in between and outside of the clusters obtained from the original data. Train the net...

9 years ago | 0

| accepted

Answered
How to save the result of the training weigths in a matlab neural network (e.g. NARX)
When training in a loop the weights at the beginning of non-initial loops are AUTOMATICALLY the same as the weights at the end o...

9 years ago | 0

| accepted

Answered
How can I use the Genetic Algorithm (GA) to train a Neural Network with inputs matrix [12 x 100] and targets [1 x 100]?
None of the GA NN posts yields a good answer. In addition, they are orders of magnitude slower than backprop. Search both NEWGRO...

9 years ago | 0

| accepted

Answered
Neural Networks - Break down of training, testing and validation groups
The bottom line is that TRAINBR has at least one bug w.r.t. datadivision. trainbr bug http://www.mathworks.com/matlabcen...

9 years ago | 1