SageMaker built-in algorithms now provides TensorFlow Image Classification algorithms
Amazon SageMaker provides a suite of built-in algorithms, pre-trained models, and pre-built solution templates to help data scientists and machine learning practitioners get started on training and deploying machine learning models quickly. These algorithms and models can be used for both supervised and unsupervised learning. They can process various types of input data including tabular, image, and text.
Amazon SageMaker provides a suite of built-in algorithms, pre-trained models, and pre-built solution templates to help data scientists and machine learning practitioners get started on training and deploying machine learning models quickly. These algorithms and models can be used for both supervised and unsupervised learning. They can process various types of input data including tabular, image, and text.
Starting today, Amazon SageMaker provides a new built-in algorithm for image classification: Image Classification - TensorFlow. It is a supervised learning algorithm that supports transfer learning for many pre-trained models available in TensorFlow Hub. It takes an image as input and outputs probability for each of the class labels. These pre-trained models can be fine-tuned using transfer learning even when a large number of training images are not available. It is available through the SageMaker Built-in algorithms as well as through SageMaker JumpStart UI inside SageMaker Studio.
Image classification TensorFlow in Amazon SageMaker provides transfer learning on many pre-trained models available in TensorFlow hub. In machine learning, the ability to utilize the training results from one model to produce another model is called transfer learning. According to the number of class labels in your training data, a classification layer is attached to the pre-trained TensorFlow hub model. The classification layer consists of a dropout layer and a dense layer, fully connected layer, with 2-norm regularizer, which is initialized with random weights. The model training has hyper-parameters for dropout rate of dropout layer, and L2 regularization factor for the dense layer. Then either the whole network, including the pre-trained model, or only the top classification layer can be fine-tuned on the new training data. The algorithm provides a wide-range of training hyper-parameters for fine-tuning on your custom dataset.
To learn how to use this algorithm, please see AWS documentation Image Classification - TensorFlow, and the example notebook Introduction to SageMaker TensorFlow - Image Classification. More detailed explanation for how to use these algorithms can be found in the following blogs on Transfer learning for TensorFlow image classification models in Amazon SageMaker blog.
All these algorithms can be used in all regions where Amazon SageMaker is available. To get started with these new models on SageMaker, refer to the documentation.