Artificial intelligence isn’t science fiction anymore. For some enterprises, the technology already provides many benefits. With machine learning algorithms, for example, applications can “learn” from and predict possible outcomes from ever-growing data sets. Top cloud vendors, including Google, now offer various services that bring AI and machine learning to the enterprise.
The Google cloud machine learning suite offers various APIs that can analyze and translate text, perform image and speech recognition and search videos with metadata. Enterprises now have access to technology that Google uses for its own applications, such as the image search capabilities used for Google Photos.
Learn about six key features that combine together to form Google Cloud machine learning services:
Google Cloud Machine Learning Engine is a managed service that enables users to build and train machine learning models at scale, using TensorFlow training applications. It also hosts those models in the cloud to make online and batch predictions based on new data.
The Google Cloud Machine Learning Engine service is made up of four major components: REST API, gcloud command-line tool, Google Cloud Platform Console and Google Cloud Datalab. It integrates with other Google services, such as Google Cloud Dataflow and Google Cloud Storage, to process and store data. The pricing structure is based on training cluster tiers — basic, standard, premium, basic GPU and custom cluster configuration — which are billed by the hour. Google charges separately for prediction requests.
Speech API is a service that converts audio to text in real time. It uses neural network algorithms to complete the conversion and has three core methods for speech recognition: synchronous, asynchronous and streaming. The service recognizes over 80 languages and various dialects. Google Cloud Speech API works with devices that can send gRPC or REST requests, such as phones and internet of things devices, and integrates with other Google products, such as Google Cloud Storage. The pricing structure is based on usage and is free for the first 60 minutes, then priced per 15 seconds afterward.
Translation API is a responsive interface that uses Google Neural Machine Translation to identify and translate the text into more than 100 languages. Users access it through the Google REST API, which receives updates regularly to add new languages and language pairs. The pricing model is based on usage and follows a per-character model.
Vision API analyzes and categorizes images. It supports various image types, including JPEG, GIF, and PNG8, and is accessed via a REST API. It has nine image detection features: label, text, safe search, facial, landmark, logo, image properties, the web and document detection. Users can apply multiple features to a single image. Google charges for these services on a monthly basis, based on the numbers of features, or units, users apply.
Video Intelligence API, currently in beta, is a REST API that enables users to search videos through metadata extraction. The API also annotates videos in Google Cloud Storage. Its four main features are label detection, shot change detection, video segmentation and regionalization. These features enable users to detect objects and scene changes, segment videos and choose a region to process the video, based on compliance regulations. The first 1,000 minutes are free, and then the price ranges from $.05 to $.10 per minute, depending on which feature you use. Contact Musato Technologies for more details on cloud services.