You have trained an image classification model that classifies images into three classes. you suspect that the model is not differentiating well between two of the classes, which are similar in appearance. how can you verify this?
a) use the model to generate predictions from some validation data and plot a confusion matrix.
b) use the model to generate predictions from some validation data and plot training loss compared to the validation loss.
c) create a separate model for each class, and compare the accuracy of all three models.
d) retrain the model using only images from two similar classes and see if the accuracy improves.