Interview with Swami, Vice President of Amazon AWS: Machine learning is experiencing a renaissance on the cloud

In this day and age, AI is already within reach and is becoming a core driver of strategic technologies leading the future and driving industrial change.

During the 2018 World Artificial Intelligence Conference, 4 global artificial intelligence innovation projects such as "Amazon AWS Cloud AI Innovation" won the highest honor of the competition, sail award, and on September 17, Swami Sivasubramanian, vice president of Amazon AWS, was interviewed by the Daily Economic News reporter (hereinafter referred to as NBD) on topics such as machine learning and basic research on the cloud.

NBD: What kind of business value can machine learning technology and cloud services create?

swami: Machine learning is not a completely new concept, and deep learning techniques may have been available as early as 30 years ago. But why have deep learning and machine learning not been adopted or landed on a large scale before? There may be several reasons: first, machine learning requires a lot of data; second, it requires huge computing power; and third, it requires highly skilled operators. So today machine learning is experiencing a renaissance on the cloud.

Now with cloud services, you have good access to massive amounts of data at a very low price, just a few cents per g, and on-demand computing power. Any organization, any type of organization with such a service, through cloud computing services can be well obtained AI technology.

In fact, more and more enterprises are starting to use machine learning for the cloud, and just from aws's customer base, the growth rate of enterprises using machine learning services on the cloud is increasing day by day. We now have tens of thousands of active developers training machine learning on aws, just last year up 250% year-on-year, and more than 80% of our customers are running on aws cloud computing.

NBD: How can enterprises in different fields improve their AI technology by mobilizing cloud platforms?

SWAMI: I'll give a few examples, such as medicheedy, which can help companies mobilize machine learning techniques to better gain insight into detection imaging, analyze X-rays, and identify major symptoms and problems faster and better, even before they become serious diseases.

Another example is the field of sports, which is now the season of American football, using AI technology, which can obtain data from sensors on players or some videos to study running routes, predict which road athletes may run in the future, and the laws behind it.

The same is true in financial services, pharmaceuticals, autonomous driving, and education, where companies can extract, analyze, and predict data and information on the platform. In the future, we are also considering how to apply deep learning technology to the structure of the database, apply it to natural language processing, computer vision technology and other fields, and how to make it more accurate and more scalable.

NBD: For service providers that provide ARTIFICIAL intelligence technology, how does machine learning optimize the internal structure of the company?

swami: First of all, for customers, what a cloud platform with AI technology does is to open up the technology of machine learning to all developers, and there are a wide range of tools on the cloud platform that can support all the frameworks and provide specialized AI services.

For Amazon, machine learning has penetrated completely into every aspect of our business. The head of each of Amazon's business units does a 6-page annual plan every year, and in each year's documents, we have to answer the question: How can we better use machine learning to improve business development? So, it can be said that innovation and machine learning are already the DNA of Amazon, fully infiltrating our daily work every day.

NBD: What do you think of the process of promoting AI technology?

swami: I think it's because of the cloud that ai has a wider range of applications. In the past, only large Internet companies and well-capitalized enterprises could run such a large system on their own databases and data centers, and small enterprises rarely could afford it independently. But it is because of the emergence of the cloud that this state has changed. Now, even a start-up can use AI on demand from scratch, and it can be very low cost, and the data can be very secure.

In addition, some companies often encounter the challenge that they may not have enough professional talents and professionals to build machine learning models. So for us, all we can offer is that we can give our customers the ability to use our solutions very easily, to help them build and train machine learning models.

NBD: What does it mean that more and more tech companies are starting to build research institutes around the world?

swami: In fact, there is still too much work to be done in terms of research, too many things, including how to build and better utilize ai systems, and how to know that it has begun to play a role, is starting, including related data centers, database technologies, scalable technologies, machine learning utilization and deployment. So we feel that this is only a very early stage of future development, and there is still a lot of room for exploration of the future prospects.

If you look at the adoption and use of artificial intelligence technology, I think ai is an innovative technology, and the potential for cockroaches to save time and costs and increase productivity is very large.

Now there are many CEOs of companies who ask the question: What is their company's AI strategy? So now is not a question of whether to use it or not, but when to start adopting ai's solutions.

Now the platform service can be built with a few mouse clicks to build their own machine learning model, but I think the best news is that the platform service provider has been constantly simplifying their solutions and processes, hoping that the customer enterprise in the operation is more convenient, such as long as there is data can automatically generate the model.

Daily economic news