Why Did My Neural Network Fail with Single Input Parameter?

  • Thread starter Thread starter a b
  • Start date Start date
  • Tags Tags
    Network Neural
Click For Summary

Discussion Overview

The discussion revolves around the challenges faced by a beginner in neural network programming, specifically regarding the performance of a feedforward neural network designed to predict shooting angles and speeds based on input parameters of distance and time of flight. The focus is on understanding why a network trained with a single input parameter (distance) fails to achieve satisfactory accuracy.

Discussion Character

  • Exploratory, Technical explanation, Debate/contested

Main Points Raised

  • One participant describes their experience with a neural network that successfully predicts angle and speed using both distance and time of flight as inputs, but fails when only distance is used.
  • Another participant suggests that the failure is due to the theoretical limitations of neural networks in approximating differentiable functions, arguing that distance alone does not provide sufficient information to uniquely determine both angle and speed.
  • The same participant points out that multiple combinations of angle and speed can yield the same distance, which may lead to confusion for the network.
  • A participant shares code and executable files related to the neural network, expressing uncertainty about compatibility with all Windows systems.

    • Areas of Agreement / Disagreement

    Participants express differing views on the reasons for the network's failure, with one attributing it to the nature of the input parameters and another sharing code without addressing the underlying issue. No consensus is reached regarding the solution to the problem.

    Contextual Notes

    The discussion highlights potential limitations related to the input parameters and the nature of the functions being approximated by the neural network, but does not resolve these issues.

    Who May Find This Useful

    Individuals interested in neural network programming, particularly beginners exploring the challenges of function approximation and input parameter selection.

a b
Messages
14
Reaction score
0
Hello,
I'm interested in neural network programming but I am a beginner.
Recently I wrote an example in C++ that I found in a book. It was a feedforward 3 layers network that learns the examples with the backpropagation algorithm. The aim of the network was to learn to "throw a stone": I taught it with some hundred numerical examples of shooting angles and speeds in output (calculated with the physics formulas) and time of flight and distance reached in input, repeated for some tens of thousands of epochs. After that, I ran the network: by putting distance and time of flight in input I obtained relatively good values of angle and speed in output (when tested, they lead to values of distance and time very near to those in input).
But then I tried to make another network with only the distance parameter in input, and I taught the net again with other examples (randomly chosen ). This try failed: I couldn't go beyond a very poor level of accuracy.
I have a faint idea of what the problem can be:I think the main issue is that there are many possible combinations of angle and speed that lead to the same distance, and maybe the network gets "confused". Maybe I simply made some mistake. Any explanation? How can I solve the problem? Do I need another kind of neural network?
Thank you
 
Last edited:
Technology news on Phys.org
It's not working because neural networks are theoretically limited to approximating differentiable functions. The problem is that f(d) doesn't give enough information to define both the angle and speed in constant terms; it defines them in terms of each other. For example 30deg at 1mph would be the same as 60deg s, where s is some number I don't feel like solving for :-p. The time is required to give it enough information to define them in terms of constants.
 
Here is the code you asked for, there is also a readme.txt file with some instructions and the executable file for windows. I hope it will be useful.
I'm not 100% sure that the executable file can run on any Windows computer, anyway I think it has a good probability to run on any Windows XP system, if I correctly linked all the libraries needed.
 

Attachments

Last edited:
Thanks!
 

Similar threads

Why is my multilayered, feedforward neural network not working?
  • · Replies 50 ·
2
Replies
50
Views
4K
Python Why backpropagation dominates neural networks instead of interpolation
  • · Replies 4 ·
Replies
4
Views
3K
Normalization condition with a neural network
  • · Replies 3 ·
Replies
3
Views
2K
Debugging a Backprop Neural Network: Issues & Solutions
  • · Replies 3 ·
Replies
3
Views
2K
Can CNNs Be Tuned to Enhance Specific Texture Parameters in Image Simulation?
  • · Replies 1 ·
Replies
1
Views
2K
How can I implement a multi-task deep learning neural network?
  • · Replies 3 ·
Replies
3
Views
2K
Neural network problems with multi-input dependencies
  • · Replies 1 ·
Replies
1
Views
2K
Neural networks as mathematical models of intelligence
  • · Replies 4 ·
Replies
4
Views
2K
MHB Convexivity issues in a new type of network - NLP
  • · Replies 1 ·
Replies
1
Views
3K
Programs Diploma thesis - Neural Network Application in Physical Problems
  • · Replies 1 ·
Replies
1
Views
2K