- Intel has introduced some unique solution to enhance scalability, security and privacy of data-centric workloads.
- They are focusing on security, privacy and throughput of AI and blockchain implementations.
The future of computing relies on data-intensive tasks driven by artificial intelligence (AI) algorithms. In the last couple of years, capabilities of machine learning methods have achieved astonishing results, and now are being commercially used in several fields.
Most of the time, machine learning algorithms need access to sensitive information, and limiting their access to correct data could affect their performance. The issue is more complicated in one of the most hyped technologies: blockchain.
In blockchain, the information is stored in infinite number of places. The records are immutable and they are distributed among all users. Each user stores a private cryptographic key. But there is one major problem with blockchain technology: privacy. Your data is not always encrypted in most blockchain applications.
Recently, Intel introduced some unique solution that can help enhance scalability, security and privacy of data-centric workloads such as artificial intelligence and blockchain. They aim to protect both data and algorithms for AI apps, and digital assets and execution of blockchain protocols.
The security in context of artificial intelligence can be implemented in two ways –
- Security for AI, where the primary goal is to protect data, parameters and algorithms.
- AI for security, where AI is used to detect and analyze exploits.
In the first case, Intel leverages multiparty machine learning, in which integrity of algorithms and access to critical data are handled via homomorphic encryption (enables system to do calculations on encrypted data without first decrypting) in hardware-based trusted execution network, such as Intel SGX.
Furthermore, they use federated learning for apps that can’t move data to a centralized location. In this case, data owners work together to tweak (enhance) a shared predication model.
Intel is collaborating with other companies like Duality, Fortanix and Docker to make AI more useful, secure and shareable for federated learning.
In April, 2018, Intel revealed a threat detection technology that improves the results of ‘AI for security’. It utilizes silicon-level telemetry to enhance the detection of advanced cyberthreats and exploits.
Intel has introduced a new technique, called off-chain computing, to better handle both blockchain implementations throughput and privacy. Recently, they announced a collaboration with Enigma – a company that builds secure decentralized applications.
Enigma builds a privacy protocol that utilizes Intel’s security technology and Intel SGX to protect code and data through the use of enclaves that are protected areas of execution in memory. They are integrating this feature for private smart contracts on the Ethereum public ledger.
Intel is already working with SAP to develop a blockchain proof of concept for improving a wide range of applications that involve the exchange of trusted critical information, a crucial application for Intel is the cross-border shipping process.
According to the company, their investments can deliver to meet current requirements of cybersecurity, and they’ll continue to make their chips an active candidate in threat defense lifecycle.