1
$\begingroup$

When I train a model with my dataset on my PC using Tensorflow, Keras, scikit-learn, Pandas or NumPy, do the developers of these libraries can gain access to my dataset?

(May be when I train the model, dataset is uploaded to some servers for processing, for improving the library or for something else. Am I right?)

$\endgroup$
-1
$\begingroup$

Modules you import in python for machine learning, like tensorflow or sklearn or pytorch are open source. They are not hosted on any particular environment. Instead, as you import module, you get raw code, which you may modify as per your requirement. Thus, as you can see, there is no way for them to steal your data while you are using these modules on your own computer

$\endgroup$
5
  • $\begingroup$ Can you be a bit more specific than "I'm pretty sure"? And how exactly is the situation different in those other cases? Any documentation that suggests that? $\endgroup$ Jun 30 at 13:07
  • $\begingroup$ Modules you import in python for machine learning, like tensorflow or sklearn or pytorch are open source. They are not hosted on any particular environment. Instead, as you import module, you get raw code, which you may modify as per your requirement. Thus, as you can see, there is no way for them to steal your data while you are using these modules on your own computer $\endgroup$ Jun 30 at 17:31
  • $\begingroup$ However, in the case you decide to use their free platforms like Google Collab or Kaggle kernals, You do need to note that your data is required to be stored in the RAM of their virtual PC. And yes it is still safe even if you use those kernals $\endgroup$ Jun 30 at 17:34
  • $\begingroup$ Thanks! Do you want to add that to your answer? $\endgroup$ Jul 1 at 7:52
  • 1
    $\begingroup$ Even if a piece of software is open-source, it doesn't mean that there isn't some code that could maybe make your computer system vulnerable or insecure. So, yes, generally, open-source software, as the name suggests, can be seen and explored by all people, but this does not directly mean that it doesn't contain malicious code. So, I think that this answer can be misleading. So, I would suggest that you reformulate your answer so that you point this out to the OP. $\endgroup$
    – nbro
    Jul 1 at 20:32

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.