automl | automatic machine learning toolkit , including hyper | Machine Learning library

My API_ENDPOINT is set to europe-west1-aiplatform.googleapis.com.

I define a pipeline:

I have been following this video: https://www.youtube.com/watch?v=1ykDWsnL2LE&t=310s

Code located at: https://codelabs.developers.google.com/vertex-pipelines-intro#5 (I have done the last two steps as per the video which isn't an issue for google_cloud_pipeline_components version: 0.1.1)

I have created a pipeline in vertex ai which ran and used the following code to create the pipeline (from video not code extract in link above):

I have created a custom processor using google AutoML entity extractor and trained few pdfs. The Pdf's actually contains Photo identity card. I was able to test it in their UI and it was able to extract the entity properly. Now Im using their Java client library to do it using code given below. Here is the sample

https://github.com/googleapis/java-automl/blob/b4c760c01efbd2174d93af85c5fbab3c09eee9f2/samples/snippets/src/main/java/com/example/automl/LanguageEntityExtractionPredict.java

Here I see that they pass the text content into the library instead I want to send the PDF content. I don't want to use the google cloud storage bucket instead I want to load file locally and sent it to the entity extractor. I tried using the Document class as below

Document.parseDelimitedFrom(FileInputStream("test.pdf")) but it gives me an error.

Any help is highly appriciated.

I am experiencing a persistent error while trying to use H2O's h2o.automl function. I am trying to repeatedly run this model. It seems to completely fail after 5 or 10 runs.

I am able to use ML.NET to manually train a model, save it, load it to create a PredictionEngine and make predictions. But when I try to use the AutoML feature, I run into problems loading the model due to schema binding issue below.

Does anyone know what the issue is? I can get my model to load fine and make predictions if I train and save it without using AutoML so this really puzzles me.

Using the h2o package for R, I created a set of base models using AutoML with StackedEnsemble's disabled. Thus, the set of models only contains the base models that AutoML generates by default (GLM, GBM, XGBoost, DeepLearning, and DRF). Using these base models I was able to successfully train a default stacked ensemble manually using the h2o.stackedEnsemble function (i.e., a GLM with default params). I exported the model as a MOJO, shutdown the H2O cluster, restarted R, initialized a new H2O cluster, imported the stacked ensemble MOJO, and successfully generated predictions on a new validation set.

So far so good.

Next, I did the exact same thing following the exact same process, but this time I made one change: I trained the stacked ensemble with all pairwise interactions between the base models. The interactions were created automatically by feeding a list of the base model Ids to the interaction metalearner_parameter. The model appeared to train without issue and (as I described above) was able to export it as a MOJO, restart the h2o cluster, restart R, and import the MOJO. However, when I attempt to generate predictions on the same validation set I used above I get the following error:

I am hoping to have some low code model using Azure AutoML, which is really just going to the AutoML tab, running a classification experiment with my dataset, after it's done, I deploy the best selected model.

The model kinda works (meaning, I publish the endpoint and then I do some manual validation, seems accurate), however, I am not confident enough, because when I am looking at the explanation, I can see something like this:

4 top features are not really closely important. The most "important" one is really not the one I prefer it to use. I am hoping it will use the Title feature more.

Is there such a thing I can adjust the importance of individual features, like ranking all features before it starts the experiment?

I would love to do more reading, but I only found this:

Increase feature importance

The only answer seems to be about how to measure if a feature is important.

Hence, does it mean, if I want to customize the experiment, such as selecting which features to "focus", I should learn how to use the "designer" part in Azure ML? Or is it something I can't do, even with the designer. I guess my confusion is, with ML being such a big topic, I am looking for a direction of learning, in this case of what I am having, so I can improve my current model.

I have trained the AutoMl classification model on Vertex AI, unfortunately model does not work with batch predictions, whenever I try to score training dataset (same which was used for the successful model training) with batch predictions on Vertex AI I get a following error:

"Due to one or more errors, this training job was canceled on Nov 11, 2021 at 09:42AM".

There is an option to get a details from this error and those say the following thing:

"Batch prediction job customer_value_label_cv_automl_gui encountered the following errors: INTERNAL"

Does anyone know what might be the reason for getting this kind of error? I am very surprised that the model cannot score the dataset that it was trained on. My dataset consists of 570 columns and about 300k of records.

I’m using the following script to execute an AutoML run, also passing the test dataset