Community Profile

# Greg Heath

3,118 total contributions since 2011

Backgound in Electromagnetic Theory, Plasma Physics and Radar Target Identification using Neural Networks.
PhD Student, Research Assistant and Lecturer at Stanford;
AB, ScB, ScM Student; Research Assistant, Fellow and Professor at Brown;
27 yrs researching Ballistic and Theatre Missile Defense using Neural Networks at MIT Lincoln Laboratory. Retired 2003.

PLEASE DO NOT SEND QUESTIONS AND DATA TO MY EMAIL. HOWEVER, CAN SEND LINKS TO POSTS.
Professional Interests: Neural Netwoks, Spectral Analysis

View details...

Contributions in
View by

semanticseg producing marginally different values when inference is repeated
Use clear all, close all, clc, rng(0) on the 1st line Thank you for formally accepting my answer Greg

5 days ago | 0

combining two neural networks in one bigger network
Just 1. Save the outputs of net1 in a file 2. Use the file to train net2 Greg

5 days ago | 0

Problem with the TreeBagger Command
The sizes of the input function and output target must be [ I N ] = size(input) [ O N ] = size (target) Hope this helps...

16 days ago | 0

How do I create a neural network that will give multiple input and outputs?
ALWAYS arrange your data so that [ I N ] = size(input) [ O N ] = size(output) Hope this helps. Greg

16 days ago | 0

How to make a hybrid model (LSTM and Ensemble) in MATLAB
Replace your YES/NO data with either 1/0 or 1/-1. Hope this helps. Greg

17 days ago | 0

how to augment image data only for a specific class?
Separate class 0 and interpolate. If you have a good feel for the data you could extrapolate. However the latter might be tricky...

17 days ago | 0

NTSTOOL - How to get predicted values of "the future"?
The known time series is analyzed to yield a time-series model that uses past and present values to predict future values. The ...

4 months ago | 0

Timedelaynet output calculation principle
You did not include tHe 2 biases. Hope this helps. Greg THANK YOU FOR FORMALLY ACCEPTING MY ANSWER

4 months ago | 0

Understand number of weights of Neural Network
It is possible. In general, however, you don't have the slightest idea what choice would be significantly better than random. ...

5 months ago | 0

Using pca for features selections
PCA (Principal Coordinate Analysis) is a very useful method for regression (it ranks linear combinations of the original variabl...

5 months ago | 0

How do we decide the number of hiddenlayers in a PatternNet?
patternnet(10) indicates ONE HIDDEN LAYER WITH TEN NODES It is important to be mindful of the number of layers and nodes. The ...

5 months ago | 0

| accepted

Why sets Matlab automatically the activation functions for a neural network like this?
The simplest useful approximation is is a series of blocks with different heights and widths. The simplest useful DIFFERENTIAB...

6 months ago | 0

Artificial Neural Networks Hidden Layers
Number of input and output nodes is determined by the data. Number of hidden layers and nodes is determined by the program auth...

6 months ago | 0

| accepted

Neural Network Classification Results
The original class sizes are unequal. Hope this helps THANK YOU FOR FORMALLY ACEEPTING MY ANSWER Greg

6 months ago | 0

x5=data_inputs(5,1:17); x6=data_inputs(6,1:17); x7=data_inputs(6,1:17); x8=data_inputs(7,1:17); x9=data_inputs(9,1:17); Hop...

6 months ago | 0

Hyperparameter tuning of neural network
One hidden layer is always sufficient. However, sometimes 1. Knowledge of the physical or mathematical process may lead to a ...

6 months ago | 1

Different results in training a CNN with Matlab 2018a and Matlab 2019a
You are making the task difficult by going backwards. Start with a single hidden node and add nodes one at a time. Hope this...

6 months ago | 0

Question

NEURAL NETWORK DATA SET EXAMPLES
For demonstration of old AND new concepts and ideas, PLEASE use the sample NN data sets provided by MATLAB help nndatas...

6 months ago | 0 answers | 0

### 0

cross validation in neural network using K-fold
%i am using neural network for classification but i need to use instead of holdout option , K-fold. ==> FALSE!. You mean y...

6 months ago | 0

Can the number of Predictors be different for Train and Test data?
Of course not. The ultimate purpose of training is to create a model that works well on non-training data. Thank you for form...

6 months ago | 0

How to check the robustness of the Neural network model?
If you are going to test with white noise, include white noise in your design (i.e., training + validation) Then, given a fixed...

6 months ago | 0

| accepted

NARX with Complex Values Input
Decades ago I learned (the hard way) to forget about trying to use complex computations for NNs. However, if you insist, let us...

7 months ago | 1

Why sets Matlab automatically the activation functions for a neural network like this?
That is a standard configuation for a neural net. It's operation is explained in every elementary text. Thank you for formally...

7 months ago | 0

I get a "Performance function replaced with squared error performance" warning when trying to set 'crossentropy' as the performance function.
If you insist on using CROSSENTROPY, try PATTERNNET. Hope this helps. Thank you for formally accepting my answer Greg

7 months ago | 0

How to plot Network performance?
You have lost training information, So the only thing left is output vs input. Hope this helps. Thank you for formally ac...

7 months ago | 0

Elman Neural Network (ENN)
size(P_TRAIN) = [ 1296 1728] size(T_TRAIN) = [ 432 1728] Hope this helps. *Thank you for formally accepting my answer* ...

7 months ago | 0

NARX re-training in closed loop
Using 100 feedback delays makes no sense. Only use feedback delays that are within the correlation length of the function. See...

8 months ago | 1

Can CNN train separately instead of learning everything at one time?
I try to have the order of inputs as random and uncorrelated as possible. Otherwise the probability of extensive learning/unlear...

8 months ago | 0

In evaluating a neural net, should NMSE be based only on test subset of data?
For serious work I calulate FOUR values of NMSE: 1.70% Training 2.15% Validation 3.15% Test 4.100% All for 10 (typically...

8 months ago | 0

| accepted