Intel is co-sponsoring the virtual European Dependable Computing Conference (EDCC) 2020, a leading conference featuring research, industrial practice and innovations in dependable and secure computing, on September 7-10. This year’s conference is free of charge for all participants due to the sponsorship and support of Intel Germany and assistance from Fraunhofer’s Institute for Cognitive Systems IKS.
Ignacio Alvarez, a senior research scientist in the Autonomous Driving Research Lab at Intel Labs, will present a keynote speech on September 8 entitled "Towards Universal Safety Guarantees of Decision Making in Automated Vehicles." At Intel Labs, Alvarez develops software, system architectures, and simulation tools to accelerate the adoption of safe automated driving technologies. His research focuses on the development of safe intelligent connected automated vehicles (AV) to augment human mobility.
In his keynote, Alvarez will discuss how mass deployment of highly automated driving technology on the road requires the industry to come to an agreement on measurement and guarantees of safety in AV decision making. He will introduce Intel Labs' solutions and recent technical contributions toward the measurement and standardization of safety, as well as the democratization of efforts within the academic, industry, and regulatory communities.
Two other industry leaders will present keynotes at EDCC 2020. On September 9, Peter Schlicht, an artificial intelligence architect at Volkswagen, will present "Challenges for Artificial Intelligence in Automated Driving." On September 10, the keynote address on "Public Transport: Challenges and Opportunities for Dependability" will be presented by Martin Rothfelder, a research group head at Siemens, also addressing Intel customers’ needs.
On September 8, researchers Xiaoming Du, Cong Li, and Shen Zhou from Intel, and Mao Ye and Jing Li from Tencent will present their paper entitled, "Predicting Uncorrectable Memory Errors for Proactive Replacement: An Empirical Study on Large-Scale Field Data."
The research team found that uncorrectable memory errors are the leading cause of server failures in data centers. Predicting uncorrectable errors (UEs) using the historical correctable error (CE) information helps for proactive replacement of memory hardware before catastrophic events happen. The researchers performed an empirical study of UE prediction on large-scale field data from more than 30,000 contemporary servers in Tencent data centers over an 8-month period. They demonstrate that the traditional approach based on CE rate works poorly with low precision. They leveraged the detail micro-level CE information to design several new predictors. The comparative study shows that the new predictor based on column fault identification boosts the baseline precision for a factor of more than 300% and at the same time substantially improves the baseline recall.
Intel Germany worked on EDCC committees with other industry and academic leaders to help organize the event. Michael Paulitsch, a dependability research architect at Intel Germany, served as a general chair of the organizing committee alongside Mario Trapp of Fraunhofer IKS. Paulitsch also participated on the program and steering committees. In addition Kerstin Alexander, business operations manager at Intel Labs Europe, served as a local organization chair with Veronika Seifried of Fraunhofer IKS. Intel Germany intern-alumni Nikolaj Schack Holmkvist Pedersen acted as web chair.
Astrid Elbe is the managing director of Intel Labs Europe.