Access Millions of academic & study documents

Problems That May Be Encountered When Fitting A Neural Network Mode

Content type
User Generated
Subject
Writing
School
Grand Canyon University
Type
Homework
Showing Page:
1/1
1
Problems that may be encountered when fitting a Neural Network Model
The neural network model has been in existence since 1950, and it works by mimicking
any continuous function. According to Naidu and Govinda (2018), the neural network model
includes interconnected computing nodes that work with neurons in the human brain. However,
this model often fails to perform up to the mark, or even it takes a lot of time to produce decent
results.
Some problems encountered when fitting a neural network model include data overfitting
and getting the same output for every input when predicting. The disadvantage of data overfitting
in this model is that the quality of the results is compromised when one tries to learn too much
from the sample. On the other hand, the neural network can fail to converge due to low
dimensionality. Time complexity may also become too high, and the gradient zero.
These problems are not specific to the neural network model; other models experience
these as well. Some of the solutions that can be implemented to overcome the insufficient
minority class problems include changing activation function frequently and increasing the
number of neurons in the model. In addition, one can normalize data, increase data, and perform
weight initialization.
Reference
Naidu, G. P., & Govinda, K. (2018, January). Bankruptcy prediction using neural networks. In
2018 2nd International Conference on Inventive Systems and Control (ICISC) (pp. 248-
251). IEEE.

Sign up to view the full document!

lock_open Sign Up
Unformatted Attachment Preview
1 Problems that may be encountered when fitting a Neural Network Model The neural network model has been in existence since 1950, and it works by mimicking any continuous function. According to Naidu and Govinda (2018), the neural network model includes interconnected computing nodes that work with ...
Purchase document to see full attachment
User generated content is uploaded by users for the purposes of learning and should be used following Studypool's honor code & terms of service.
Studypool
4.7
Indeed
4.5
Sitejabber
4.4

Similar Documents