Machine Learning

Welcome! The Azure Machine Learning team invites you to share and vote for features to help you build, manage, and deploy custom machine learning models.

Have a technical question, or want to learn more? Please visit our documentation, MSDN forum or StackOverflow.


  1. Support Synapse Built-in SQL pool as a datastore

    I know that Syapse's Built-in SQL pool (see attached image) cannot currently be registered as a data store. Please let me know if there is a plan to support it. If it has provided as a preview, let me know how to apply.

    29 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  2. Data validation component in Azure Machine Learning designer

    A dataset validation component that is aimed at the Citizen Data Engineer as typically the data is pretty dynamic.

    37 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  3. Using AML Studio features under Vnet without private endpoint

    Currently, in order to use convenient features on AML Studio (such as Designer & AutoML), it is necessary to configure private endpoint when the datastore (not default storage account) is under Vnet. If private endpoint is not available, the access to the datastore will be denied. For instance, if using dataset validation, AutoML or Designer to access dataset, the error below (attached image files) will occurred.

    It will be a huge advantage if the configuration of private endpoint can be omitted when setting the network of datastore.

    Please consider such a implements in the future.

    21 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  4. Need samples using AzureMLDataset in the designer

    It is very useful to have samples for Designer that uses AzureMLDataset (exported dataset of labeled data). There is much demand, so please consider it.

    40 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  5. 17 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  6. Add Object Detection Module in Azure ML Designer

    Though It is easy to create an image classification model in Azure ML Designer by using DenseNet or ResNet module, it's more convenient if some object detection modules can be provided in designer.

    13 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  7. Enable Forecasting models when lagging features are enabled

    Currently, Forecasting models are disabled when user uses lagging features such as target lags and/or rolling windows. They should be added to the roadmap and made available even when lagging features are enabled.

    https://docs.microsoft.com/en-us/python/api/azureml-automl-core/azureml.automl.core.shared.constants.supportedmodels.forecasting?view=azure-ml-py
    -> 'AutoArima','Prophet','Average','Naive','SeasonalAverage','SeasonalNaive'

    9 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  8. Enable Datastores Service in AML to access the default Storage Account using "Resource Instances" in the integrated Firewall of the SA

    Resource instances by adding Microsoft.MachineLearningServices in the integrated Firewall of the Storage Account when it's behind a Vnet enable us to authorize the workspace to write experiments output, models and logs.

    Activating the integrated firewall however blocking the Datastores of the workspace to access the storage account (such as Designer or Explore DS) in scenarios where the private endpoint is not applicable to be implemented.

    It will be a huge advantage if the Datastore uses the same Resource Instance as (or some other feature) in order to access the default storage account of the workspace privatly without the need of…

    7 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  9. Ask for more unsupervised Models as Train Anomaly Detection Model in the Designer

    AML Studio (classic) supports One-Class SVM, but Designer does not. I would like Designer to support more unsupervised Models, such as One-Class SVM and Local Outlier Factor.

    36 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  10. allow specifying the resource group of a virtual network when creating an container instance (endpoint) using the sdk (and portal).

    when running through the mnist tutorial with our setup i run into the following problem.

    aciconfig = AciWebservice.deployconfiguration(cpucores=1,

                                               memory_gb=1, 
    
    vnet_name="vnetname",
    subnet_name="subnetwithdelegation",
    tags={"data": "MNIST", "method" : "sklearn"},
    description='Predict MNIST with sklearn')

    there is no way to define the resource group of the virtual network in the deploy_configuration method and it defaults to the resource group of the machine learning services workspace. it would be nice to define the resource group name here so i can connect to a different vnet

    38 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  11. 4 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  12. Automatic Shutdown for Compute VMs

    It would be great if we could have the ability to automatically shut down Compute VM's at a specified time - similar to what you can do with VM's provisioned outside of the AML Portal.

    32 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  13. Make AutoML a module in the Designer

    It would be very useful if AutoML could be used from within the Designer.

    3 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  14. Avoiding LGPL license restriction in azureml-core phyton library

    Hello all,

    We are using azureml-core phyton library to interact with Azure Machine Learning Workspace. The library is hosted on https://pypi.org/project/azureml-core It is authored by Microsoft and maintained by pypi community. One of its dependencies of azureml-core is "chardet" phyton library and it has LGPL license restriction. https://pypi.org/project/chardet/

    Our compliance team would like us to avoid the use of LGPL licensed open source libraries to avoid future license implications. I couldn't find an alternative to azureml-core phyton library.

    I noticed that there are two GitHub initiatives to replace chardet . Please refer to below links:

    https://github.com/psf/requests/issues/4848

    https://github.com/encode/httpx/issues/1018

    I would like…

    18 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  15. Need to be able to delete unfinished runs in an experiment

    I need to terminate still-running runs.
    when I view details of a run, I see that there is is a button named
    "Mark as cancelled "
    But ... When I select that, it comes to with the warning;- " Cancel run You have requested to cancel run: 8 Note that this operation will only mark the run as canceled in the UI since it is executed on local compute. This will not terminate the run on the compute if it is still running. Ensure that you terminate the run and shut down any compute to avoid any unexpected billing charges.…

    1 vote
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  16. Don't have invisible meta data from scoring, that cannot be replaced

    It turns out that the Score activity in the Designer, adds some invisible meta data, called Score, that if you do any Arithmetic work on, will remove that metadata on the column, and make the Evaluator stop working. Just spent about 14 tech support calls tracking this down.

    Here is what he said:
    This is because the Evaluate Model module is expecting a ‘Scored Dataset’, which means the Score columns (predicted column, Scored Labels, Scored Probabilities) all have metadata assigned to them so that their ‘Feature type’ is Numeric/String Score instead of Numeric/String Feature. Passing the scored dataset into ‘Apply…

    8 votes
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
  17. Problem with dependency resolution of Azure ML SDK

    Open-Source AzureML-SDK (already posted somewhere else) and deliver the requirements. pip install azureml-sdk takes an eternity now. Maybe also build a conda installation version, as the new backtracking algorithm really slows down installing that package.

    1 vote
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  18. tensorboard profiler is absent from azureml-tensorboard

    It's not installed when you start a cluster with azureml. To fix, pip install -U tensorboard-plugin-profile where tensorboard is run. Without this, ppl using AzureML can't easily profile memory use or IO of their models as models are training or doing inference. I can't find any guides on how to do this with Azure ML.

    1 vote
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  19. Random seed for AutoML

    Hi, can you please include an argument in AutoMLConfig() to set the random seed for re-producibility? I was working on a classification task and ran AutoML using SDK 5 times and got 5 different ensemble model pipelines each time. It's very hard to reproduce the results. Can you please include a arandom_state argemnt like scikit-learn. Thanks.

    1 vote
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    0 comments  ·  Flag idea as inappropriate…  ·  Admin →
  20. Create AML Datasets from within the Designer

    If the Designer could create fully functional AML Dataset (Tabular and/or File), it would be possble to have citizen data scientists work on data transformation, preparation and feature engineering and then use AutoML or the SDK to do actual Machine Learning with these Datasets.

    At the moment there is a disconnect between Designer and other authoring methods.

    1 vote
    Vote

    We're glad you're here

    Please sign in to leave feedback

    Signed in as (Sign out)
    You have left! (?) (thinking…)
    1 comment  ·  Flag idea as inappropriate…  ·  Admin →
← Previous 1 3 4 5 11 12
  • Don't see your idea?

Machine Learning

Categories

Feedback and Knowledge Base