1 – IBM Scientists Imitate the Functionality of Neurons with a Phase-Change Device
IBM scientists in Zurich, Switzerland have advanced cognitive computing through the research and creation of phase-change-based artificial neurons. While study of phase-change materials for memory applications has been going on for more than a decade, accelerated progress has been made in the past two years. IBM Fellow Evangelos Eleftheriou said,
“We have discovered and published new memory techniques, including projected memory, stored 3 bits per cell in phase-change memory for the first time and now are demonstrating the powerful capabilities of phase-change-based artificial neurons, which can perform various computational primitives such as data-correlation detection and unsupervised learning at high speeds using very little energy.”
The artificial neurons, which are made of the same base materials in re-writable Blu-ray discs, work by imitating two stable states: an amorphous or unstructured state and a crystalline or structured state. IBM’s work was featured this week in the peer-reviewed Nature Nanotechnology.
(Read the full article at the IBM News room)
2 – New Online Course Addresses Data Science Challenges
Beginning October 4, MIT is offering the online course, Data Science: Data to Insights, aimed at data scientists in both startups and larger corporations trying to make sense of big data. Over six weeks, participants will cover a number of topics, including recommendation engines, regressions, network and graphical modeling, anomaly detection, hypothesis testing, machine learning and big data analytics. The course is being offered by MIT Professional Education and the MIT Institute for Data, Systems and Society (IDSS), and will be co-taught be IDSS faculty members. Devavrat Shah, a professor of electrical engineering and computer science who will be co-leading the course, says there are three key components that data scientists need to introduce to modern organizations: a sensing platform for gathering the data, infrastructure to do computations at scale, and statistical and machine-learning methods to extract information. “These three components are fundamentally intertwined. This course looks primarily at the third component, and also provides guidelines as to what is the right data to collect for the given set of decisions,” said Shah.
(Read the full article on MIT News)
3 – The White House Requested Input on Artificial Intelligence, and IBM’s Response is a Great AI 101
IBM recently provided a comprehensive response to the White House’s June RFI, or request for information, organized by sections according to the original RFI questions. In addition to the summary provided in the main, the links ‘see more here’ at the bottom of each subsection lead a more in-depth perspective. Topics covered include the use of AI for public good, social and economic implications of AI, education for harnessing AI technologies, fundamental questions in AI research and gaps, multidisciplinary research, safety and control issues for AI, data sets that can accelerate AI research, legal and governance implications for AI, and other issues (identified as business models). The content is presented and explained in clear language so that most anyone can read it and take away a general understanding of the major issues at hand.
(Read the full article on TechCrunch and IBM’s Response here)
4 – Apple Acquires Turi in Major Exit for Seattle-based Machine Learning and AI Startup
According to multiple sources close to the deal, Apple has acquired the machine learning startup Turi. The Seattle-based Turi, formerly known as Dato and GraphLab and founded by University of Washington professor Carlos Guestrin, was purchased for an officially undisclosed price (though reports of an estimated $200 million have been given by sources). Though Apple’s long-term goals for the Turi acquisition have not been revealed, it’s expected that the Turi team will continue in Seattle while Apple continues expansion of its efforts in data science and artificial intelligence. Turi provides a platform for developers to build apps using AI and machine learning, with a suite of products organizations wanting to make sense of their data. The acquisition is one of several (others include Emotient and Flyby Media) that Apple has completed in 2016.
(Read the full article on GeekWire)
5 – Google Announces First Hardware Contribution to the Open Compute Project
Google announced this week its plans to collaborate with Facebook in the creation of a new data center rack design, made in the image of its own hardware systems. This is a major announcement coming five months after Google joined the Open Computer Project (OCP) to proliferate the adoption of hyperscale hardware (a distributed computing environment). Google’s and Facebook’s Open Rack 2.0 blueprint will be centered on increasing efficiency (a widely-applied Google effort) and saving on technology expenses within an organization. OCP designs are mostly geared toward Fortune 500 enterprises and large service providers that have data center equipped with thousands of machines. Before Google delivers a formal submission of its specification, it will request feedback from the OCP community for improvements to its system.
(Read the full article on siliconAngle and press release at Google Cloud Platform Blog)
Image credit: Wikimedia
