Description
The dataset used for this problem (high-resolution images) has specifically being separated as I am using a sliding window approach (sliding window at multiple scales) to further subdivide each image hence separating them as D1 and D2. This is to avoid exposing image windows of the same image during both training and testing.
Scenarios
No Train validation Testing
1. D1 D1 D2
2. D1 D2 D2
Which of the above approach is correct? In my case, I assume its No 1.
What if you obtain good performance in testing and validation for the case of scenario No 1 but poor testing performance.? Can the reason be the problem is not well represented by the D1 used during the training process?
The strategy you using will over-fit your model.
Better to do as :
D 1 D1 D1
D2 D2 D2
D3 D3 D4
...
Dn Dn Dn
Good Luck.
Dear Friend,
The following text may help:
Training Set is the data set is used to adjust the weights on the neural network.
Validation Set is the data set is used to minimize overfitting. If the accuracy over the training data set increases, but the accuracy over the validation data set stays the same or decreases, then you are overfitting your neural network and you should stop training. A validation dataset is a sample of data held back from training your model that is used to give an estimate of model skill while tuning the model’s hyperparameters.
For example, it is used to choose the number of hidden units in a neural network.
Testing Set is the data set is used only for testing the final solution in order to confirm the actual predictive power of the network.
https://stackoverflow.com/questions/2976452/whats-is-the-difference-between-train-validation-and-test-set-in-neural-netwo
Dear Muhammad Ali ,I have edited my question to make it more clear
Dear Sumit Das Thank you
- Badges
- Science topic