[
    {
        "type": "api-change",
        "category": "Comprehend",
        "description": "This release gives customers the ability to tokenize (find word boundaries) text and for each word provide a label for the part of speech, using the DetectSyntax operation. This API is useful to analyze text for specific conditions like for example finding nouns and the correlating adjectives to understand customer feedback. "
    },
    {
        "type": "api-change",
        "category": "SageMaker",
        "description": "Amazon SageMaker has added the capability for customers to run fully-managed, high-throughput batch transform machine learning models with a simple API call. Batch Transform is ideal for high-throughput workloads and predictions in non-real-time scenarios where data is accumulated over a period of time for offline processing."
    },
    {
        "type": "api-change",
        "category": "Snowball",
        "description": "AWS Snowball Edge announces the availability of Amazon EC2 compute instances that run on the device. AWS Snowball Edge is a 100-TB ruggedized device built to transfer data into and out of AWS with optional support for local Lambda-based compute functions. With this feature, developers and administrators can run their EC2-based applications on the device providing them with an end to end vertically integrated AWS experience. Designed for data pre-processing, compression, machine learning, and data collection applications, these new instances, called SBE1 instances, feature 1.8 GHz Intel Xeon D processors up to 16 vCPUs, and 32 GB of memory. The SBE1 instance type is available in four sizes and multiple instances can be run on the device at the same time. Customers can now run compute instances using the same Amazon Machine Images (AMIs) that are used in Amazon EC2."
    },
    {
        "type": "api-change",
        "category": "Polly",
        "description": "Amazon Polly adds new API for asynchronous synthesis to S3"
    }
]