How to classify multitasking and multi-label in MATLAB using deeplearning?

In the past alexnet, googlenet, etc. are multi-category single-label classification network.Recently, I just want to train a single network can be a picture of many aspects of the description, I do not know whether the current MATLAB can achieve such a problem?

[Tarily]

Dear, do you know this answer? I have the same question now 🥺

[xingxingcui]

Perhaps this example will give you some inspiration and you can change it to suit your needs.

https://ww2.mathworks.cn/help/deeplearning/ug/multilabel-image-classification-using-deep-learning.html

[xingxingcui]

You should be looking for answers on your own, based on misinformation, rather than asking for solutions without thinking.

Answer 1

[xingxingcui]

Multi-label classification can be done by changing the loss function to "binary cross-entropy loss", as far as I can remember, using a statement like the following

loss = crossentropy(Y,T. TargetCategories="independent")