NYU-HSRN-Network-Data-Science-Group / AutoZeekWatch

An online, deployable machine learning network intrusion detection system for Zeek.
MIT License
3 stars 0 forks source link

train_batch #12

Closed olive-jy-song closed 9 months ago

olive-jy-song commented 9 months ago

.fit() instead of .fit_partial() is used, considering cases with input shape (n_samples, n_features). from source code, fit_partial() only receives input shape (n_features, ), while .fit() is just iterating through the input and apply fit_partial() to each sample. in other words, the batch size is always 1 in kitNet, which i think we can definitely try optimizing in the future. :D