I'm working with my model 2 .Now is ok. I use experiment Manager for optimization Buffers and now is time with NeuralNetworks in technomatix plant simulation.
In experiment manager : For the multi-level experimental design is :
I put my Experiment Manager in NeuralNet
Can someone explain me , what doesn mean in the diagrams (digram1,diagram2,diagram3 and diagram4). I used Examples in Technomatix Plant simulation and also Help in Technomatix Plant Simulation, step by step researched every model, but I still don't understand some informations about this diagrams.
I really appreciate your answers
Solved! Go to Solution.
there is a relatively extensive documentation of this tool in chapter Reference Help. Basic knowledge about Neural Networks is required. The shown diagrams can be reproduced by a corresponding example of the Example Collection: Category: Tools and Optimization, Topic: Neural networks, Example: Production system (In this way all readers can have benefit.).
The quality of the training is analyzed in the lower group of the second tab and explained in chapter Training and Checking the Neural Network of the Reference Help. You must know that the data from the Experiment Manager are divided into training data and validation data (as it is mentioned on this page of the documentation). Your first and second diagram shows large relative errors during the training progress (The x-axis is the number of learning steps).
I recommend trying other settings in the Configuration dialog.
Maybe you can increase the number of hidden layers on the first tab. Please note, that if you want to make a change you must reset the training.
It is also possible to improve the training by the adequate treatment of the noise of the data. Try the setting Noise by percentage. The variance is frequently too large.
Such large errors often happen after a small number of learning steps. You selected 100. Increase the number of training steps to 1000.
Also the fourth diagram shows large differences with respect to the values of an input value on the x-axis. It seems that the training was not successful.
But your third diagram is surprisingly good: The curve of the dependence of a pair of input value and output value has the expected shape.
Thank you Peter,
It is very interesting for me this Manufacturing process in Neural Networks.
What type of Neural Networks is in Technomatix Plant Simulation . Is it
I trying to understand this process and MLP general ---
I have knowledge about Neural networks general-theoretical, there are more chapters in web site-scribd,academia,google about Artificial Neural Networks ,but I didn't find any Chapters which has explanation about numbers -
Numbers -What does mean in production process.
the Structure of the Neural Network (NN) is described in a corresponding chapter of the Reference Help. The notation perceptron is not used in this documentation.
There are many text books about NN. Two books are mentioned in the Reference Help on the first page of the chapter Artificial Neural Network. A short overview of technical details is in the Technical Report Back Propagation Family Album (1996).
I want to explain the basic ideas. A trained NN is described by two (or three) matrices, the so-called weights. The backpropagation learning algorithms minimizes the error between the output of the NN (calculated by the weights and the activation function) and the training data. The learning algorithm uses the gradient idea.
The Magnitude of activation Beta is evaluated by the activation function.
If the relative error is too small then the training algorithm cannot detect the direction of improvement. A reinitialization of the weights is necessary.
For each learning step we get matrices of weights. An update of the weights uses the weights of the last learning step and the step before. Both methods are applied in each step of the training. The first method uses the dynamically adapted learning rate Eta. The second method is called momentum method and uses the parameter Alpha.
At the beginning of the training the weight are choosen at random. Its size is determined by Magnitude of weights.
The dialog contains recommendation for these parameters. The success of the training depends from these parameters. Please try only small changes.
I would like to know, what does noise do in neural network?
I was not able to find the exact requirement of noise in Help and other sources of the internet, can you please explain it.
Reference Help suggests that the only way of finding the optimum solution is by increasing the iteration rate which results to reduce error, is that correct ??
this setting takes noise in your data into account during the training of your neural network. This means that errors within a given threshold ( in this example 5 percent) are accepted, resulting in an error of 0 percent. You can also have the variance of the output values define this threshold. Then the value of your threshold depends on the range in which your output values deviate. If this deviation is large on average, then your threshold value will also be accordingly large. If you use neither of both options, then your error will be calculated directly as absolute deviation from your output values.
Furthermore, the optimal solution is generally not just found by increasing the iteration rate. This may of course lead to a reduced training error, but on the other hand you may run into the problem of overfitting. Here your network starts to also model the noise in your data and thus the validation error goes up again. This is in particular a problem if you don`t define an error threshold for the training as described above.
Can you please explain about weights, I have three inputs but here I got four input data
you can see the image when I select the first cell(1,1) the second cell(2,1) is not highlighted but the rest cells (3,1) &(4,1) out highlighted
the extra column you are seeing here is reserved for the bias term that is added to the respective neuron in each layer.
The table shows the input values, which approximate the desired output value as closely as possible and which cause the least amount of costs. Delta describes the discrepancy between the desired output value and the approximated output value using the shown set of input values. You can define the costs for a certain combination of input values in the “Settings” under “Forecast”, next to other parameters for the optimization algorithm for the forecast. The “Increments of the input values“ option determines the accuracy with which the input values are calculated. The optimization algorithm used will execute the “Number of iteration steps” that you define here at most. For a proportional cost calculation, you can enter the cost factors for each increment of an input value under “Costs table”. You can also run more complex cost calculations with a method using “Individual costs function”. The parameters for this method are provided by a table with the input values. The output values of this method are then the resulting costs.