Watson Studio (Deploy)

Showing 37 of 13335

Cheating Sheet for WML (Algorithm Selection)

WML provides the automated model build experience though, it would be nice to have the cheating sheets for the algorithm selection as to the user will be able to understand how it should work.Example: Microsoft Azure has; https://docs.microsoft.co...

almost 5 years ago in Watson Studio (Deploy) 0 Not under consideration

2 VOTE

WML model selection reasoning like SPSS Modeler

Like DataRobot, often users required to check how and why the model selection is made by WML. It would be nice to have more detail of the model selection (not only ROC etc ...) - like SPSS Modeler has. i.e., which variable was most important etc ...

almost 5 years ago in Watson Studio (Deploy) 0 Functionality already exists

1 VOTE

Keep an auditable log for webservice requests

In IBM SPSS C&DS, you can enable a real-time scoring webservice, and you can keep a configurable audit log where you can store all request data and all response data. In WML, you can deploy a model and automatically generate a webservice, ...

over 3 years ago in Watson Studio (Deploy) 0 Planned for future release

1 VOTE

FTP to transfer dataset (as webbased takes FOREVER

No description provided

over 3 years ago in Watson Studio (Deploy) 0 Not under consideration

1 VOTE

Support human-readable reports on machine learning model change to provide traceability

Today Watson Studio supports lineage of both model and data via graphical UI. However it doesn't seem to support generating a human readable report to describe the changes (e.g. adding / removing features, model parameters). In order to better sup...

about 4 years ago in Watson Studio (Deploy) 0 Planned for future release

1 VOTE

Add custom loss function support in WML

Today when you train a deep learning model using a custom loss function, you cannot deploy it in WML. Custom loss functions are very important when you are dealing with unusual specific cases.

about 4 years ago in Watson Studio (Deploy) 1 Not under consideration

1 VOTE

Running the WML scoring on Docker at the edge.

WML service is a great method to score on-Cloud application though, it is sometime necessary to score the logic at the edge in which may or may not have the connection to the internet, or has to be inside the firewall. It would be highly beneficia...

over 4 years ago in Watson Studio (Deploy) 0 Planned for future release

1 VOTE

Why can you just link me here?

//WHY CANT YOU JUST LINK ME INSTEAD OF TELLING ME THIS?// NOTE: you must manually construct wml_credentials hash map below using information retrieved // from your IBM Cloud Watson Machine Learning Service instance

almost 5 years ago in Watson Studio (Deploy) 2 Not under consideration

1 VOTE

update deployed model made by spss or flow

Deployed model made by machine learning can update but deployed model made by spss or flow cannot update in machine learning. update means overwite deployed model (same name,same address) We want to update deployed model made by spss or flow in ma...

almost 5 years ago in Watson Studio (Deploy) 0

1 VOTE

More major algorithms for WML model builder

MS Azure ML and other ML vender solution has more algorithms than WML Model builder provides. i.e., Azure ML has more algorithms; https://docs.microsoft.com/en-us/azure/machine-learning/studio/algorithm-choice WML only has supervised learning - ...

almost 5 years ago in Watson Studio (Deploy) 0 Not under consideration

Please enter your email address

FILTER BY CATEGORY