Hfrxdjt

What is the reasoning behind standardization (dividing by standard deviation)?

Error in master's thesis, I do not know what to do

Why do I have a large white artefact on the rendered image?

Why doesn't the fusion process of the sun speed up?

Norwegian Refugee travel document

Is this Pascal's Matrix?

What (if any) is the reason to buy in small local stores?

What is the tangent at a sharp point on a curve?

Is "inadequate referencing" a euphemism for plagiarism?

How to test the sharpness of a knife?

Single word to change groups

Recursively updating the MLE as new observations stream in

Do native speakers use "ultima" and "proxima" frequently in spoken English?

Unfrosted light bulb

Should a narrator ever describe things based on a characters view instead of fact?

How do researchers send unsolicited emails asking for feedback on their works?

The English Debate

pipe commands inside find -exec?

Hackerrank All Women's Codesprint 2019: Name the Product

How do you justify more code being written by following clean code practices?

How can I create URL shortcuts/redirects for task/diff IDs in Phabricator?

10 year ban after applying for a UK student visa

Is VPN a layer 3 concept?

Weird lines in Microsoft Word

Learning with missing features (MNAR)

Supervised Learning with Necessarily Missing DataIs there any difference between feature extraction and feature learning?Missing Categorical Features - no imputationDeploying the prediction model under missing values for test dataDoes encoding missing values help?Can we look just at the other features when we have a missing vaue?Imputation missing values other than using Mean, Median in pythonDealing with NaN (missing) values for Logistic Regression- Best practices?Balance data using different criteriaHow to deal with missing data for Bernoulli Naive Bayes?

0

$begingroup$

I want to learn from features that may have some missing informations. The value of the variable that's missing is related to the reason it's missing (MNAR)

To better understand my case, here is an example :

I want to learn a model for voice recognition. When using the model I'll know who will be speaking.

I have training data for those speakers, but I also have other data with speakers that will not use the model.

I don't want to add features for those speakers, only for those that will use the model.

How can I process the inputs before training the neural network without risking to damage the performance of the end users ?

For now I intend to use this method :

Each end user, for wich I know the identity, will have a dedicated feature. When training on speaker without dedicated feature :

• A feature that represent "have dedicated identity feature" will be set to -1.


• All identity features will be put to 0.

For example :

End user 1 → [1 -1 -1 ... 1]

End user 2 → [-1 1 -1 ... 1]

End user 2 → [-1 -1 1 ... 1]

...

speakers only for training → [0 0 0 ... -1]

Is this the right thing to do ? Is there a better way ?

P.S. - I can only modify the inputs, the neural network's architecture cannot be modified, so I can't use things like dropout.

machine-learning neural-network missing-data

share|improve this question

asked 1 hour ago

Adrian B.Adrian B.

11

$endgroup$

add a comment | 

0

$begingroup$

I want to learn from features that may have some missing informations. The value of the variable that's missing is related to the reason it's missing (MNAR)

To better understand my case, here is an example :

I want to learn a model for voice recognition. When using the model I'll know who will be speaking.

I have training data for those speakers, but I also have other data with speakers that will not use the model.

I don't want to add features for those speakers, only for those that will use the model.

How can I process the inputs before training the neural network without risking to damage the performance of the end users ?

For now I intend to use this method :

Each end user, for wich I know the identity, will have a dedicated feature. When training on speaker without dedicated feature :

• A feature that represent "have dedicated identity feature" will be set to -1.


• All identity features will be put to 0.

For example :

End user 1 → [1 -1 -1 ... 1]

End user 2 → [-1 1 -1 ... 1]

End user 2 → [-1 -1 1 ... 1]

...

speakers only for training → [0 0 0 ... -1]

Is this the right thing to do ? Is there a better way ?

P.S. - I can only modify the inputs, the neural network's architecture cannot be modified, so I can't use things like dropout.

machine-learning neural-network missing-data

share|improve this question

asked 1 hour ago

Adrian B.Adrian B.

11

$endgroup$

add a comment | 

$begingroup$

I want to learn from features that may have some missing informations. The value of the variable that's missing is related to the reason it's missing (MNAR)

To better understand my case, here is an example :

I want to learn a model for voice recognition. When using the model I'll know who will be speaking.

I have training data for those speakers, but I also have other data with speakers that will not use the model.

I don't want to add features for those speakers, only for those that will use the model.

How can I process the inputs before training the neural network without risking to damage the performance of the end users ?

For now I intend to use this method :

Each end user, for wich I know the identity, will have a dedicated feature. When training on speaker without dedicated feature :

• A feature that represent "have dedicated identity feature" will be set to -1.


• All identity features will be put to 0.

For example :

End user 1 → [1 -1 -1 ... 1]

End user 2 → [-1 1 -1 ... 1]

End user 2 → [-1 -1 1 ... 1]

...

speakers only for training → [0 0 0 ... -1]

Is this the right thing to do ? Is there a better way ?

P.S. - I can only modify the inputs, the neural network's architecture cannot be modified, so I can't use things like dropout.

machine-learning neural-network missing-data

share|improve this question

asked 1 hour ago

Adrian B.Adrian B.

11

$endgroup$

share|improve this question

asked 1 hour ago

Adrian B.Adrian B.

11

share|improve this question

asked 1 hour ago

Adrian B.Adrian B.

11

asked 1 hour ago

Adrian B.Adrian B.

11

asked 1 hour ago

Adrian B.Adrian B.

11