At AWS Innovate Online Conference – AI & Machine Learning Edition, Amazon reinforced its leadership in artificial intelligence and ML stack. Amazon Web Services (AWS) has positioned itself as one of the key players for companies from different verticals looking to digitisation and deployment of AI or ML services. It claims to have more than 1.5 million AWS customers using database, analytics, or machine learning services.
At AWS Innovate – AI & Machine Learning Edition, AWS explains how the data explosion shows the reach of ML growing. According to Statista, the worldwide data volume will hit 181 ZB by 2025, from 64 ZB in 2020. This increased data volume will lead to an increase in global spending as well.
IDC estimates the global spending on AI to grow from $50.1B in 2020 to more than $110 billion in 2024. While the number may not look gigantic, it is important to note that cloud computing took nearly 12 years to reach such a level of spending. This spending also means that ML is rapidly evolving from the pilot phase to the operational phase. Gartner says that 75 per cent of enterprises will shift from piloting to operationalising AI by the end of 2024.
Broadest and Deepest AI and ML services
Amazon got a significant lead in the world of cloud computing with its ingenuity to offer extra computing as a service. However, it has new competition in the form of Microsoft’s Azure, Google Cloud, and the e-commerce giant is looking beyond cloud computing with its AI and ML stack that rely on Amazon Web Services.
The company offers the broadest and most complete set of machine learning capabilities. It offers AI services that make it easy for everyday developers to incorporate ML into their applications. There is also a suite of pre-trained ML models that can be accessed via an API and requires no ML experience.
It also has a dedicated ML Services called Amazon SageMaker, an end-to-end ML platform that streamlines the process of building, training, and deploying ML. This ML platform is used by data scientists looking to focus on data science and not worry about managing the underlying infrastructure.
Lastly, there is a suite of ML Frameworks & Infrastructure offering full control over the infrastructure and supports all major frameworks such as TensorFlow, PyTorch and mxnet. This suite is for expert ML practitioners, who are comfortable building, tuning, training, deploying, and managing ML models.
Philippe Battel, Head of Data, Analytics and AI/ML, EMEA, AWS says this complete stack is designed to support business priorities. The ML stack allows optimisation of businesses with new efficiencies, making smarter and faster decisions, adding new capabilities to existing products, and inventing net-new products.
Digital transformation across asset-heavy industries
One of the overarching themes of AWS Innovate Online Conference – AI & Machine Learning Edition was digitisation happening across industries. From asset-heavy industries, financial services to pharmaceutical and tech, digitisation has gained momentum due to the rapid development of technology and the need to digitise due to the pandemic.
Lionel Billon, Head of Analytics & ML – EMEA South & Emerging Markets, says digital transformation is increasing across asset-heavy industries like discrete and process manufacturing, transportation and logistics, agriculture, power and utilities, renewables, energy, and mining.
Billon says AWS and ML Analytics portfolio comprising data visualisation insights, analytics, data lake infrastructure and management, and data movement as the key tools being used in the asset-heavy industries. However, industries are not without challenges and chief among them is data access, data management, security, real-time decision making, and scale.
One of the highlights in AWS suite is Amazon Lookout for Vision, a tool that allows companies to spot defects and anomalies using computer vision. The Lookout for Vision tool does not require ML experience and businesses can start building a model with as few as 30 images. AWS customer ENGIE is using the tech to ensure that its pipelines are failure proof.
Bringing agility to financial services and pharmaceutical industries
In Europe, AWS has become a force to reckon with in the financial services industry. From the likes of Experian, BBVA, HSBC, Barclays to fintech startups such as Klarna and Wise, AWS has become the de facto choice. At AWS Innovate Online Conference – AI & Machine Learning Edition, Amazon is showing its platform is making the financial services industry agile.
From the use of SageMaker to generate credit scores to the use of facial recognition to authenticate users, AWS is powering a number of backend financial operations. AWS does not miss this opportunity to highlight how the data transacted on its platform is secured with the masking of sensitive user information such as identity numbers.
Amazon Web Services is also touting major growth in the field of pharma and life sciences industry. Luis Campos, EMEA Data, Analytics, & AI/ML modernisation lead, says AWS ML stack allows the pharmaceutical industry to not only become agile but also compliant with global standards.
He says the use of advanced ML leads has led to cost reduction, elimination of unnecessary inventory and drop in machine downtime. Campos and Battel both cite Moderna as an example for use of AWS in the pharma industry. Moderna is one of the earliest champions in developing an effective COVID-19 vaccine.
ML is changing how products are built at tech companies
Oren Steinberg, Head of AI & ML in North and South EMEA, says that AI and ML are changing how products are built. Steinberg explains that traditionally programmers take the data, design the program and get the right output. However, in the field of machine learning, it is actually the other way around. ML engineers start with the data and the desired output while the machine itself builds the program. Steinberg further explains that “we get the output as a program from the model itself.”
This change in programming has led to every company becoming a data company and thus an AI-fuelled company. Some of the most effective use cases in tech companies are being built using artificial intelligence and machine learning right now. He further explains that AI and ML has led to a major change in product development. These days, software and internet companies rely on use case finder discovery workshops.
This is explained as, “If we knew insights, we could do actions in order to get KPI.” AWS says the common use cases of ML are in the field of vision, language, personalisation, ranking, forecasting, recommendation, classification, regression, clustering, and anomaly detection.
While machine learning itself can be divided into subcategories – supervised learning, unsupervised learning, and reinforcement learning – AWS is seeing increased adoption of reinforcement learning among tech companies to build new products and services. Steinberg points out the recommendation engines powering services such as Spotify, Netflix and even Amazon, which are built using reinforced learning.
Amazon Web Services started as a cloud storage service with S3 but it now has instances like EC2 powered by its own custom silicon called Inferentia. At reinvent last year, Amazon introduced Trainium as its second machine learning chip powering EC2 Trn1 instances for high-performance deep learning training. As its peers accelerate their own AI and ML tools on cloud, AWS is building and expanding the capabilities of its own stack to stay competitive.
Photo: Tony Webster / Flickr.