0
$\begingroup$

i have an image dataset and there are 6300 images with 5 classes . The features extracted and dataset reduced to 256 features. This dataset gives good results(%99) when tested ANN with Backpropagation(tensorflow). I'm working on ANN weights optimization and this dataset gives bad results( under %30) when tested on weight optimized ANN. But different datasets(iris,wine,..) gives good results when tested on weight optimized ANN. Lots of test applied for different population sizes or epochs and results didn't change. Where is the problem?

$\endgroup$

1 Answer 1

0
$\begingroup$

It seems that the problem might be related to the weight optimization method you are using for the specific image dataset. Since the weight optimized ANN works well with other datasets (like iris, wine), it indicates that the optimization method might not be suitable for the image dataset with 256 features.

Here are a few suggestions to investigate and potentially resolve the issue:

  1. Data preprocessing: Make sure that the image dataset is preprocessed correctly, including normalization, scaling, and any other necessary transformations. This can have a significant impact on the performance of the weight optimization method.

  2. Initialization: Check the initialization of the weights in the optimized ANN. Different initialization methods can have a significant impact on the convergence and performance of the optimization algorithm.

  3. Optimization algorithm: You might want to try different optimization algorithms or adjust the hyperparameters of the current optimization method. Some optimization algorithms might work better for specific types of datasets or problems.

  4. Network architecture: Experiment with different ANN architectures, such as varying the number of layers, neurons per layer, or activation functions. The architecture of the ANN can significantly affect the performance of the weight optimization method.

  5. Regularization: Consider adding regularization techniques, such as L1 or L2 regularization, to prevent overfitting and improve the generalization of the optimized ANN.

  6. Cross-validation: Perform cross-validation to ensure that the results are consistent across different subsets of the dataset. This can help identify any potential issues with the dataset or the optimization method.

By investigating these aspects, you should be able to identify the cause of the poor performance of the weight optimized ANN on the image dataset and make the necessary adjustments to improve the results.

$\endgroup$
1
  • $\begingroup$ Thanks for contributions, i will consider each of them. $\endgroup$ Jul 8, 2023 at 9:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.