Yes - this can be done, but the approach will differ based on your specific dataset. I've included some examples of different approaches below:

Weighted Loss Function: In a neural network, assign higher weights to the minority class samples to ensure that they have a larger impact on the loss function. This helps the model pay more attention to the minority class during training, and improves its performance on that class.

Ensemble Methods: Ensemble methods can be effective in dealing with class overlap. You can train multiple neural networks on different subsamples of the data and then combine their predictions to make the final decision. This can help to reduce the impact of class imbalance by averaging out the errors.

One-Class Classification: One-class classification allows you to train a neural network to identify one specific class, and then use that network to identify the other class. This can be useful when one class is much more prevalent than the other, as it can help to avoid false positives.

Transfer Learning: Transfer learning can be useful when you have a pre-trained neural network that has been trained on a similar task. You can use the pre-trained network as a starting point for your own network, which can help to improve its performance on the minority class.

Hope this helps.