Companies are collecting increasing amounts of data, a trend that is driving the development of better analytical tools and tougher security. Analysis and security are now converging as confidential computing prepares to deliver a critical boost to artificial intelligence.

Intel has been investing heavily in confidential computing as a way to expand the amount and types of data companies will manage through cloud services. According to Intel Fellow Ron Perez, who works on security architecture with the Intel Data Center Group, the company believes the emerging security standard will allow enterprises and large organizations to explore new ways to share the data needed to fuel AI and machine learning.

We see this as a long-term effort, Perez said. But the reason why were investing is that it has the potential to be a huge shift for cloud and utility computing.

Confidential computing is a standard that moves past policy-based privacy and security to implement safeguards on a deeper technical level. By using encryption that can only be unlocked via keys the client holds, confidential computing ensures companies hosting data and applications in the cloud have no way to access underlying data, whether it is stored in a database or passes through an application.

The concept is gaining momentum because it allows data to remain encrypted even as its being processed and used in applications. Because the company hosting the data cant access it, this security standard should prevent hackers from grabbing unencrypted data when it moves to the application layer. It would also theoretically allow companies to share data, even between competitors, to perform security checks on customers and weed out fraud.

In August 2019, Intel became one of the founding members of the Confidential Computing Consortium, an open source effort managed by the Linux Foundation that aims to develop the hardware and software standards needed to further adoption. Companies like IBM,Google, and Microsoft have begun to highlight their work in this area as a way to encourage large enterprises, particularly in areas such as finance and health care, to put more of their sensitive data in the cloud.

Perez leads a group of senior technologists at Intel focused on security architecture through a program dubbed Pathfinding. Perez describes it as the pursuit of interesting challenges that our customers are facing. In Perezs case, the goal is to develop a pipeline of security technologies for Intels datacenter customers.

Intel began its work in this area before the term confidential computing came into vogue, with Perez pointing to the companys launch of software guard extensions in 2015. The SGXs are security coding built directly into Intel processors that create separate memory enclaves where data could be placed to limit access. This idea of using hardware and software to protect data while allowing it to be processed is at the heart of confidential computing.

Microsoft used these Intel processors for its Azure cloud to enable its own confidential computing service. Last month, Intel announced it was expanding these capabilities in a new generation of its Xeon Scalable platform.

Our approach has been to drive continuous innovation and deep collaboration with our technology partners to improve the confidentiality and integrity of all data, wherever it is, Perez said.

Proponents of confidential computing argue that it will lead to a new wave of cloud innovation as companies become more comfortable putting their most sensitive data online. Perez said that helps drive AI and machine learning in a couple of ways.

The first is indirect. AI and ML have advanced in recent years, thanks to the growing datasets available to refine algorithms. Confidential computing, by bringing even more and richer data online, will benefit that development.

The main connection to machine learning and artificial intelligence is the fact that were generating more and more data, Perez said. Were analyzing this data with various machine learning technologies. And that explosion of data is whats really driving the interest in confidential computing, whether its used for machine learning or not. Machine learning just happens to be one of its main uses.

No matter the type of underlying data, if it must be decrypted to be used, the security of algorithms it passes through is critical.

How do you protect these algorithms across this very broad spectrum of use cases? Perez said. We see confidential computing as a paradigm shift for cloud computing. The infrastructure providers are providing the capabilities that allow cloud companies to deliver these services as a utility, and they dont have to take responsibility for the protection of the data themselves.

Beyond that, confidential computing is enabling different types of collaboration around data to drive machine learning. Perez pointed to the example of a brain tumor project at the University of Pennsylvania.

Penns Perelman School of Medicine has teamed up with 29 other health care and research institutions around the world, including in the U.K., Germany, and India. The group is using Intels confidential computing to develop a distributed approach to machine learning that allows them to share patient data, including medical imaging. Because such data can remain encrypted while it is being used for machine learning, the group can safely share that data and collaborate in a way that otherwise might not be possible.

Thats critical because data is urgently needed to train machine learning, but no single institution has enough to achieve this on its own. Previously, Penn Medicine and Intel Labs published a study showing that federated learning (a collaborative approach) could train a machine learning model far more effectively than working alone. In this case, the group believes the combination of confidential computing and federated learning will allow them to make rapid breakthroughs in AI models that identify brain tumors.

Merchants are also tapping the ability to allow new types of collaboration for customer and partner data, as are enterprises. While analysts like Gartner believe the real impact of confidential computing may still be several years away, Perez said it is already helping some sectors accelerate their AI and machine learning capabilities in the short term.

There are multiple aspects of the computing stack that need to be protected, Perez said. Confidential computing solves problems that couldnt be solved before. The concept that I can use any computing capability that may reside in any country around the world and still have some preservation of the privacy and confidentiality of my data, thats pretty powerful.

See original here:

Why Intel believes confidential computing will boost AI and machine learning - VentureBeat