question

NateTripp-8044 avatar image
0 Votes"
NateTripp-8044 asked NateTripp-8044 commented

Access company's fileshare from Azure ML Compute

I have been doing some searching around and can't seem to find anything particularly related to this question. Is it possible to connect to a local fileshare system from an Azure ML instance? Basically, I have a large amount of data in the form of images stored on a local drive and would like to use ml.azure to train and make predictions on this data.

azure-machine-learning
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

1 Answer

YutongTie-MSFT avatar image
0 Votes"
YutongTie-MSFT answered NateTripp-8044 commented

Hello,

I am not pretty sure what's "company fileshare" you are mentioning. But Azure Machine Learning Studio does support upload data from local/ datastore/ Web URL/ public dataset. You can create your dataset for training.

Reference:https://docs.microsoft.com/en-us/azure/machine-learning/how-to-connect-data-ui#create-datasets

Besides Studio, you can also connect to datastore and storage.

There are two dataset types, based on how users consume them in training; FileDatasets and TabularDatasets. Both types can be used in Azure Machine Learning training workflows involving, estimators, AutoML, hyperDrive and pipelines.

A FileDataset references single or multiple files in your datastores or public URLs. If your data is already cleansed, and ready to use in training experiments, you can download or mount the files to your compute as a FileDataset object.

A TabularDataset represents data in a tabular format by parsing the provided file or list of files. This provides you with the ability to materialize the data into a pandas or Spark DataFrame so you can work with familiar data preparation and training libraries without having to leave your notebook. You can create a TabularDataset object from .csv, .tsv, .parquet, .jsonl files, and from SQL query results.

Reference: https://docs.microsoft.com/en-us/azure/machine-learning/how-to-create-register-datasets#dataset-types

138755-image.png

Hope this will help. Please let us know if any further queries.



  • Please don't forget to click on 130616-image.png or upvote 130671-image.png button whenever the information provided helps you. Original posters help the community find answers faster by identifying the correct answer. Here is how

  • Want a reminder to come back and check responses? Here is how to subscribe to a notification

  • If you are interested in joining the VM program and help shape the future of Q&A: Here is how you can be part of Q&A Volunteer Moderators








image.png (168.5 KiB)
· 3
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

@NateTripp-8044

Please let us know if you still need help.

Regards,
Yutong

0 Votes 0 ·

I guess I should explain a little bit better. The issue I am having is that my company wants to make predictions on 1000's of images a time. All these images are stored on a local drive so my initial thought was to be able to connect to my company's network and access files directly inside of their internal file system. That way when I attempt to make predictions, I can pass in file paths instead of having to provide full images to the endpoint.

0 Votes 0 ·

@YutongTie-MSFT Please see my previous comment. Thank you.

0 Votes 0 ·