Scribd Logo
Upload

Welcome to Scribd!

  • 428 Training Text Classifiers in Create ML

    Uploaded by

    Hesam Saeidi
    18 views31 pages
    The document discusses text classification using Create ML. It explains that transfer learning allows using a pre-trained model on a new dataset by taking the pre-trained model's embeddings and training a new classifier on top. The document provides examples of sentiment analysis, spam detection, and topic classification as potential text classification problems. It demonstrates how to create a text classifier using transfer learning in Create ML, including training, evaluating, and saving the model. Some considerations for text classification like choosing an appropriate algorithm, providing balanced classes, and ensuring data consistency are also mentioned.

    Original Description:

    Original Title

    428_training_text_classifiers_in_create_ml

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    The document discusses text classification using Create ML. It explains that transfer learning allows using a pre-trained model on a new dataset by taking the pre-trained model's embeddings and training a new classifier on top. The document provides examples of sentiment analysis, spam detection, and topic classification as potential text classification problems. It demonstrates how to create a text classifier using transfer learning in Create ML, including training, evaluating, and saving the model. Some considerations for text classification like choosing an appropriate algorithm, providing balanced classes, and ensuring data consistency are also mentioned.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    18 views31 pages

    428 Training Text Classifiers in Create ML

    Uploaded by

    Hesam Saeidi
    The document discusses text classification using Create ML. It explains that transfer learning allows using a pre-trained model on a new dataset by taking the pre-trained model's embeddings and training a new classifier on top. The document provides examples of sentiment analysis, spam detection, and topic classification as potential text classification problems. It demonstrates how to create a text classifier using transfer learning in Create ML, including training, evaluating, and saving the model. Some considerations for text classification like choosing an appropriate algorithm, providing balanced classes, and ensuring data consistency are also mentioned.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    Download as pdf or txt
    of 31

    •

    Text Classifiers in Create ML

    Tao Jia, Core ML


    Text Classification
    Text Classification

    Lorem ipsum dolor sit amet, consectetur


    adipiscing elit, sed do eiusmod tempor
    incididunt ut labore et dolore magna aliqua.
    Label 1

    Ut enim ad minim veniam, quis nostrud


    exercitation ullamco laboris nisi ut aliquip ex
    ea commodo consequat.
    Label 2

    Excepteur sint occaecat cupidatat non


    proident, sunt in culpa qui officia deserunt
    mollit anim id est laborum
    Label 3
    Text Classification

    Lorem ipsum dolor sit amet, consectetur


    adipiscing elit, sed do eiusmod tempor
    incididunt ut labore et dolore magna aliqua.
    Label 1 Sentiment Analysis 👍 👎

    Ut enim ad minim veniam, quis nostrud


    exercitation ullamco laboris nisi ut aliquip ex
    ea commodo consequat.
    Label 2

    Excepteur sint occaecat cupidatat non


    proident, sunt in culpa qui officia deserunt
    mollit anim id est laborum
    Label 3
    Text Classification

    Lorem ipsum dolor sit amet, consectetur


    adipiscing elit, sed do eiusmod tempor
    incididunt ut labore et dolore magna aliqua.
    Label 1 Sentiment Analysis 👍 👎

    Ut enim ad minim veniam, quis nostrud


    exercitation ullamco laboris nisi ut aliquip ex
    ea commodo consequat.
    Label 2 Spam/Not Spam

    Excepteur sint occaecat cupidatat non


    proident, sunt in culpa qui officia deserunt
    mollit anim id est laborum
    Label 3
    Text Classification

    Lorem ipsum dolor sit amet, consectetur


    adipiscing elit, sed do eiusmod tempor
    incididunt ut labore et dolore magna aliqua.
    Label 1 Sentiment Analysis 👍 👎

    Ut enim ad minim veniam, quis nostrud


    exercitation ullamco laboris nisi ut aliquip ex
    ea commodo consequat.
    Label 2 Spam/Not Spam

    Excepteur sint occaecat cupidatat non


    proident, sunt in culpa qui officia deserunt
    mollit anim id est laborum
    Label 3 Topic Classification
    Data
    Sports Entertainment Nature

    article_01.txt article_02.txt article_15.txt

    article_04.txt article_03.txt article_10.txt

    article_05.txt article_06.txt article_21.txt

    article_08.txt article_07.txt article_22.txt


    •

    Demo
    •

    Text Classifier using Transfer Learning


    •

    What is Transfer Learning?


    Pre-trained Model
    Your Data
    Pre-trained Model
    Your Data
    Pre-trained Model

    Your Model
    I was able to park my car near the park entrance.
    I was able to park my car near the park entrance.
    Embeddings
    Your Text
    Your Text
    Your Text

    Embeddings
    Your Text

    Embeddings
    // Create Model using Transfer Learning
    let parameters = 

    MLTextClassifier.ModelParameters(algorithm: .transferLearning(.dynamicEmbedding))

    let classifier = try MLTextClassifier(trainingData: .labeledDirectories(at: trainDirectory),


    parameters: parameters)

    // Evaluate Model
    let evaluation = classifier.evaluation(on: .labeledDirectories(at: testDirectory))

    // Save Model
    try classifier.write(to: writeToPath)
    // Create Model using Transfer Learning
    let parameters = 

    MLTextClassifier.ModelParameters(algorithm: .transferLearning(.dynamicEmbedding))

    let classifier = try MLTextClassifier(trainingData: .labeledDirectories(at: trainDirectory),


    parameters: parameters)

    // Evaluate Model
    let evaluation = classifier.evaluation(on: .labeledDirectories(at: testDirectory))

    // Save Model
    try classifier.write(to: writeToPath)
    Considerations
    Considerations

    Choose an algorithm that fits your use case best


    Considerations

    Choose an algorithm that fits your use case best

    Advances in Natural Language Framework Thursday, 3:00


    Considerations

    Choose an algorithm that fits your use case best

    Provide balanced classes

    Advances in Natural Language Framework Thursday, 3:00


    Considerations

    Choose an algorithm that fits your use case best

    Provide balanced classes

    Ensure data consistency

    Advances in Natural Language Framework Thursday, 3:00


    More Information
    developer.apple.com/wwdc19/428

    Machine Learning Lab Friday, 2:00

    You might also like