BigARTM for semi-supervised learning

[juliaserebr]

Hi,

A part of my data is labeled and I am trying to use BigARTM for classification. If I understand correctly, I should use LabelRegularizationPhiRegularizer, but there is no documentation or examples for it. Do you have any tutorials or documentation for python interface?

Thank you,
Julia

[Oleksandr Frei]

Hi,

BigARTM classification is fairly simple feature, but unfortunately there is no tutorial.

Basically, you need to set predict_class_id argument in artm.transform:

http://docs.bigartm.org/en/stable/api_references/python_interface/artm_model.html#artm.ARTM.transform

• predict_class_id (str) – class_id of a target modality to predict. When this option is enabled the resulting columns of theta matrix will correspond to unique labels of a target modality. The values will represent p(c|d), which give the probability of class label c for document d.

More details about classification are here:

http://www.machinelearning.ru/wiki/images/d/da/Voron17fast.pdf (setionc IX.B - Multimodal ARTM / The class modality)

First you need to infer topics for a given document. Later, you use

to find distribution on classification labels. Then for hard classification you take argmax - e.i class with highest p(c|d').

Kind regards,

Alex

--
You received this message because you are subscribed to the Google Groups "bigartm-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bigartm-users+unsubscribe@googlegroups.com.
To post to this group, send email to bigart...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/bigartm-users/aeb0cc12-b84d-4efb-9217-11a4bf803a17%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Мурат Апишев]

Hi!

 

See this tutorial in docs: http://docs.bigartm.org/en/stable/tutorials/python_userguide/m_artm.html

 ---

С уважением, Мурат Апишев.

Моя страница на machinelearning.ru

 

Regards, Murat Apishev.

04.07.2018, 18:13, "juliaserebr" <julia...@gmail.com>:

--

You received this message because you are subscribed to the Google Groups "bigartm-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to bigartm-user...@googlegroups.com.

[juliaserebr]

Alex, Murat,

Thank you so much!

Kind regards,

Julia

среда, 4 июля 2018 г., 18:13:33 UTC+3 пользователь juliaserebr написал: