I have a 10gb file of a time series 1D signal. I want to find some patterns within this signal, I know CNN's are great for this but the problem is I don't have any training data.

Now I could of course spend an entire week slowly making 100 versions of a certain pattern to train the CNN with. But maybe there is some other way?

Maybe there is a way for the neural network to work out patterns on its own and simply categorize them? Like this is pattern A, this is pattern B.

My ultimate goal is to look at any size data and find the occurrences of patterns within the data.

Does anyone have any idea how this problem could be solved? I am just starting with machine learning so I am slowly learning of what's possible in this field.



You should look into unsupervised learning, which is machine learning without a training set. CNN's are cool but they need a training set.

| improve this answer | |
  • $\begingroup$ "which is machine learning without a training set". That's not true. If you consider auto-encoders unsupervised learning models, then they need a dataset to be trained. The difference is that this dataset is not labelled, but unlabelled. It's also not necessarily true that CNNs need a labelled dataset, for the same reason. In fact, you can have convolutional auto-encoders (i.e. auto-encoders with convolutions). So, I suggest that you edit this answer and fix the issues. $\endgroup$ – nbro Nov 15 at 13:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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