CRN examines several innovative cloud AI and machine learning offerings from leading cloud computing vendors, startups, and other providers.
Last week, Google Cloud’s machine learning-powered Document AI platform, along with Lending DocAI and Procurement DocAI, became publicly accessible. It has already been used to process tens of billions of pages of documents for government agencies and the lending and insurance industries, among others.
The serverless Document AI platform is a centralised document processing console that lets users use a single API to access Google Cloud’s form, chart, and invoice parsers, software, and offers, such as Procurement DocAI and Lending DocAI. It uses AI/ML to identify, extract, and enrich data from scanned and digital documents on a large scale, including structured data from unstructured documents, making it easier to understand and analyse.
Google technologies such as computer vision, optical character recognition, and natural language processing are used in Doc AI solutions to create pre-trained models for high-value and high-volume documents, as well as Google Knowledge Graph to validate and improve fields in documents.
Gartner forecasts that by 2025, AI will be the most important factor influencing IT infrastructure decisions, resulting in a tenfold increase in compute requirements. According to Gartner, half of all companies will have AI orchestration systems to operationalize AI by 2025, up from less than 10% in 2020.
Industry-specific Google Cloud services By automating mortgage document preparation, Lending DocAI aims to reduce the time – from weeks to days – and cost of closing loans for the mortgage industry. It uses advanced machine learning models to process borrowers’ income and asset records, and it now has more specialised parsers for documents like paystubs and bank statements.
Companies may use Procurement DocAI to automate procurement data collection at scale, converting unstructured records like invoices and receipts into structured data. A utility parser for electric, water, and other bills was recently introduced to Google Cloud.
The new Lending and Procurement DocAI advanced parsers are compatible with Google Cloud’s AutoML Text & Document Classification and AutoML Document Extraction services.
Google Cloud is also introducing Human-in-the-Loop AI, a new DocAI feature that allows human verification and corrections before data is used in critical business applications, to help customers improve document processing accuracy.
Here are nine more cloud AI services and software to consider.
EdgeOps Platform by Adapdix
Adapdix EdgeOps is a platform-as-a-service suite for edge-optimized machine learning and artificial intelligence deployments. Advanced AI/ML analytics was combined with a distributed, edge-based architecture.
According to Adapdix, a Pleasanton, Calif.-based company, the predictive analytics solution, which is based on industrial-grade data mesh technology, enables ultra-low latency and adaptive maintenance to minimise unplanned downtime while providing automatic control for self-correcting and self-optimizing behaviour on equipment without human interference.
The six-year-old startup’s platform customers are initially manufacturing companies in the semiconductor, communications, and automotive industries.
SageMaker by Amazon
Amazon SageMaker is Amazon Web Services (AWS) flagship, fully managed machine learning service, which allows data scientists and developers to rapidly design, train, and deploy machine learning models into production-ready hosted environments.
According to the business, SageMaker, which was launched in November 2017, is one of the fastest growing services in AWS history, with tens of thousands of active, external customers using it each month.
SageMaker provides tools for marking, data planning, feature engineering, statistical bias identification, auto ML, training, running, hosting, explainability, tracking, and workflows, as well as for each phase of the ML creation lifecycle. It includes an integrated Jupyter authoring notebook instance for easy access to data sources for exploration and analysis, as well as native support for bring-your-own algorithms and frameworks for versatile distributed training options. TensorFlow, PyTorch, MXNet, and Hugging Face are all assisted by SageMaker.
In the last year, AWS has added more than 50 new SageMaker features. AWS announced SageMaker Feature Store, a fully managed repository for storing and sharing machine learning features, which are the attributes or properties models use during training and inference to make predictions; SageMaker Clarify, which helps developers detect bias in ML models and understand model prediction; and SageMaker Pipelines, which is defined as the first purpose-built CI/CD pipelines, at its re: Invent conference in December. It also announced Amazon SageMaker Data Wrangler, a SageMaker Studio feature that allows users to import, plan, turn, feature engineer, and analyze data from start to finish; and SageMaker Edge Manager, which allows users to optimize, protect, track, and maintain machine learning models on fleets of edge devices such as smart cameras, robots, personal computers, and mobile devices.
Aporia ML Monitoring Platform
Based in Tel Aviv, Israel Aporia’s customizable monitoring framework for machine learning models, which supports both private and public clouds, came out of stealth mode this month.
According to Aporia, which was established in 2019, its platform enables data scientists to quickly and easily build their own advanced output monitors to track the performance of their machine learning models, maintain data integrity, and provide responsible AI. It takes only a few lines of code to set up and monitors asynchronously, allowing it to handle billions of daily predictions without affecting latency, according to the company.
The Aporia dashboard displays ML output status in a single pane of glass and compares actual performance to projected results in real time. Kind mismatch, data and prediction drift, model staleness, and custom metric deterioration are among the behaviours that Aporia can proactively track, and a warning engine sends out updates when issues arise.
Ex Machina C3 AI
C3 AI Ex Machina is a cloud-native predictive analytics framework that enables users to quickly integrate data as well as create, scale, and deliver AI-based insights without having to write code.
The new end-to-end, no-code AI solution was released into general availability in January by C3 AI, a Redwood City, Calif.-based company that specialises in enterprise AI applications.
C3 AI Ex Machina enables users to access and prepare petabytes of disparate data using prebuilt connectors to data sources like Snowflake, Amazon S3, Azure Data Lake Storage, and Databricks; create and manage AI models using an intuitive drag-and-drop interface powered by AutoML; and publish predictive insights to enterprise systems or custom business applications.
Customer churn avoidance, supplier delay reduction, asset reliability estimation, and fraud detection are all examples of customer use cases.
H2O Self-Driving AI
H2O Driverless AI is artificial intelligence (AI) enterprise platform for automated machine learning (ML) that is designed for the rapid creation and deployment of predictive analytics models.
According to Mountain View, Calif.-based H20 AI, which specialises in open source AI and ML, it automates some of the most challenging data science and ML workflows, such as feature engineering (using a library of algorithms and feature transformations to automatically engineer new, high-value features for a given dataset), model validation, model tuning, model selection, and model deployment. Automatic visualisations and machine learning interpretability are also available on the platform.
More than 130 open-source recipes can be used to customise and expand H2O Driverless AI. It runs on commodity hardware and was designed to take advantage of graphical processing units (GPUs), such as multi-GPU workstations and servers including IBM’s Power9-GPU AC922 server and the NVIDIA DGX-1, to speed up training.
Watson is IBM’s virtual assistant.
Users may create their own branded live chatbot or virtual “assistant” with IBM’s Watson Assistant and integrate it into devices, applications, or channels, such as webchats, voice calls, and messaging platforms, to provide customers with quick, reliable, and accurate responses.
Watson Assistant uses AI to understand natural language questions from customers and learns from customer conversations and machine learning models created from the company’s data to enhance its ability to address customer problems in real time on the first try. Companies may use machine learning to find popular topics in customer chat logs, for example, in order to train the assistant.
Watson Assistant is available on the IBM, Amazon Web Services, Google Cloud Platform, and Microsoft Azure clouds, as well as on-premises with IBM Cloud Pak for Data.
Azure Cognitive Search is a service provided by Microsoft.
Microsoft’s Azure Cognitive Search is an AI-powered cloud search service designed for mobile and web app creation.
The platform-as-a-service offering includes built-in vision, language, and speech cognitive skills, as well as the ability for developers to add their own custom machine learning models to extract insights from information. Optical character recognition, key phrase extraction, and named object recognition are among its AI capabilities.
Deep learning models are used to understand user intent and contextually rank specific search results based on that intent rather than just keywords in Azure Cognitive Search’s new semantic search functionality, which is incorporated into the query architecture and available as of March.
Azure Cosmos DB, Azure SQL Database, Azure Blob Storage, and Microsoft SQL Server hosted in an Azure virtual machine all have built-in indexers. Microsoft Word, PowerPoint, and Excel, as well as Adobe PDF, PNG, RTF, JSON, HTML, and XML, are all supported by Azure Cognitive Search.
Database of Pinecone Vectors
Pinecone Systems, an ML cloud computing startup based in Sunnyvale, Calif., came out of stealth mode in January, releasing a serverless vector database for machine learning.
According to the company, the cloud-native database allows large-scale, real-time inference as simple as querying a database. CEO Edo Liberty, the former director of research at AWS and head of Amazon AI Labs, also oversaw the AWS community developing algorithms used by Amazon SageMaker customers, founded the company.
According to the Sunnyvale, Calif.-based business, the Pinecone database answers complex queries over billions of high-dimensional vectors in milliseconds. Personalization, semantic text search, image retrieval, data fusion, deduplication, suggestion, and anomaly detection are just a few of the real-time applications it supports.
TensorIoT’s Safety Visor
TensorIoT’s Safety Visor is located in Irvine, California. The SafetyVisor from TensorIoT is an easy-to-use, machine-learning-powered IoT solution for automatically monitoring compliance with social distancing policies.
When social distancing violations occur, SafetyVisor uses computer vision to track the space between individuals, compensating for scene geometry to measure distance, and sending real-time warnings, including photos. When staff and visitors are not wearing face masks, it can detect this and send warnings. It also has a space utilisation feature that generates heatmaps from people’s movement over time, allowing for data-driven cleaning and disinfecting in high-traffic areas.
The cloud-based platform, which is powered by AWS IoT and ML, is designed to work with customers’ existing camera systems and wireless sensors without requiring any additional training.