2
$\begingroup$

I am looking for research and experience working with ML models to ingest data for tasks, like text analysis, and creates a system that copies (or in other words enciphers) the input data, to then reproduce it in the future without the original.

I'm interested in how ML models can be used in this way to obfuscate information without too much information loss by the model, e.g. overfitting on purpose to create a new representation of the input information.

$\endgroup$
2
$\begingroup$

It sounds like you are trying to compress data, and then recover the same data later.

The most common tool for this task is an autoencoder. This model accepts data as input, and then learns to compress it and decompress it to produce something as close as possible to the original data. By making the middle layer of an autoencoder narrower, you can make the compression more lossy. By making it wider, you can make it less lossy.

$\endgroup$
1
  • 1
    $\begingroup$ Hi, thanks, I'm aware that data storage is also a tendency for other models too, hence the idea of overfitting. For example an LSTM network with with layers of r larger enough size can learn to reproduce text from the input corpus. I just wondered if there had been other issues or research in this area. E.g. taking user data, training a model, deleting the user data but actually the data is still mostly present in the model to greater of lesser extent... im interested in this grey area $\endgroup$ – benbyford Oct 21 '19 at 17:54

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.