AWS Continues AI Push for Cloud Services. Amazon Web Services Inc. (AWS), a cloud, was quick to use machine learning, deeplearning and other artificial intelligence technologies in order to improve its cloud services. It has continued its AI push with a new telecom service as the latest example. Yesterday, AWS introduced Machine Learning for Telecommunication. This framework is described as “an end-to-end machine-learning (ML) process that includes ad-hoc data exploration and data processing, feature engineering, model training and evaluation, and model testing and evaluation.” It targets the telecommunications sector with a customized dataset that illustrates the use of machine learning algorithms to test and train models for predictive analysis in telecommunication. The company’s fully managed ML services, Amazon SageMaker, and the open-source Web application Jupyter Notebook are used in the service. This allows developers to create and share live code, equations and narrative text. AWS announced that customers can use the notebooks to create their own ML models and modify the Jupyter notebooks to suit their needs. AWS also made these announcements in the last week to further demonstrate its AI push.
- AWS Deep Learning AMIs now support Chainer 5.0. AWS announced on Nov. 5: “The AWS Deep Learning AMIs (for Ubuntu and Amazon Linux) now include Chainer 5.0, which includes support Python 3.6 and iDeep 2.0.” Deep Learning AMIs are optimized builds of Chainer 5.0, which can be fine-tuned for high-performance deep learning on Amazon EC2 CPU or GPU instances.
- Amazon SageMaker support of Pipe Mode for datasets stored in CSV format. AWS announced on Nov. 5 that the built-in algorithms in Amazon SageMaker now support Pipe Mode to import CSV files. This improves the speed with which data can be streamed to SageMaker from Amazon Simple Storage Service, (S3) by up to 40% while also training machine learning (ML). This new enhancement extends the performance benefits of Pipe Mode to training datasets both in CSV and the protobuf recordIO formats that we released earlier in the year.
- Amazon Rekognition now allows for more accurate scene and object detection. AWS announced that its deep learning-based analysis tool, which is used to identify people, objects, scenes, and activities in images, can now detect objects and scenes more accurately (called “label identification”) and locate objects in images. AWS stated that label detection is used to identify objects and scenes within images. Amazon Rekognition was unable to locate the object within an image. Amazon Rekognition can now identify the location of common objects, such as dogs, people, and cars, in an image by returning object bounding box. It also has significantly improved accuracy for all existing scene and object labels across a variety use cases.
You can find more information about Amazon Rekognition, and other AWS AI services, such as Amazon Machine Learning, on the Artificial Intelligence website, the dedicated Machine Learning site on AWS, and the Machine Learning & Artificial Intelligence website, which details AWS Marketplace offerings.