Intel Publishes Artificial Intelligence and Internet of Things Reference Kits

Intel has announced the release of its first set of four open-source AI reference kits , developed in collaboration with Accenture and specifically designed to make AI more accessible. The kits aim to facilitate the work of organizations in on-premises, edge, and cloud environments.

Built with components from Intel's comprehensive portfolio of AI software, these reference kits will enable millions of developers and data scientists to quickly and easily introduce AI into their applications or to energize their existing intelligent solutions.

Dr. Wei Li, vice president and general manager of AI and analytics at Intel.

By including AI model code, comprehensive machine learning pipeline instructions, libraries, and Intel oneAPI components, these kits bring increased performance across different architectures. Each of them offers different capabilities. There's service health assessment, visual quality control, customer chatbots, and intelligent document indexing.

The Internet of things on the railways

Rising demand, high customer expectations, capacity constraints, and legacy infrastructure are challenges facing the rail sector . Operators can modernize their core technology and infrastructure with the help of the Internet of Things (IoT) and artificial intelligence (AI), along with deep learning capabilities and vast amounts of data and insights. This enables operators to overcome these challenges.

Smart devices can measure important data generated by rail journeys, such as vibration and temperature. They also act as cameras for vehicles and stations, as security systems, and as digital signage. When their data is used with edge computing , actionable insights can be extracted. AI can also be used to help rail operators gain a competitive edge over other modes of transport and prepare for the future.

These artificial intelligence kits can make it easier for railway operators to transition to a modern and technological network. Combining data center technologies, wired and wireless networks, security, and software is one of the options it enables. Another rail-focused solution is a secure platform that is deployed at the edge and integrates information and operations technologies into a single system.

Using the cloud and AI for fraud detection in academia

Websites like El Rincón del Vago demonstrated that students, if they can, share notes and even plagiarize entire papers. With the shift to distance learning, it's essential to prevent students from plagiarizing and copying on exams. This has become a challenge for universities offering these types of courses . The Swiss distance learning university Fernfachhochschule Schweiz has been offering distance learning for 20 years, yet it lacked an effective and efficient way to monitor distance exams.

An AI-powered video technology solution optimized with Intel's OpenVino toolkit was used to monitor exams. The solution, hosted on Intel DevCloud, uses automation to detect suspicious activity during student exam recordings.

The new system extracts videos of questionable activity and assembles them into a two- to four-minute reel. It is then reviewed by a member of the faculty. The AI-assisted system frees up an incredible amount of time for the proctor, who previously had to review one to two hours of video of students taking exams.

Improving cybersecurity in financial institutions

Nationwide, the world's largest building society and the UK's second-largest mortgage provider, must maintain the integrity and security of the data and accounts of more than 15 million people. Know Your Customer (KYC) protocols, designed to help protect financial institutions from fraud and corruption, are typically implemented with the support of credit rating agencies. However, it is a slow and highly bureaucratic process .

To streamline and improve the accuracy of this process, they have begun digitizing their protocol, starting with a proof of concept to explore machine learning on coded datasets. They have employed confidential computing techniques that support machine learning and computing on sensitive third-party datasets.

While successful, the results of consultations working with real-world customer experiences and production environments did not perform as expected. Therefore, Nationwide turned to confidential computing powered by Intel Software Guard Extensions, which offers hardware-based memory encryption capable of isolating application-specific code and data in memory.

This second method based on Intel SGX allowed the encrypted data to be securely decrypted within an enclave. This way it can be scaled more easily than the previous method. There is now a clear case for using Intel SGX-protected data exchange to help detect and mitigate financial fraud and money laundering.