This Week’s Tech Highlights: AI‑Driven Chip Production, New Blockchain Standard & More

Industry News

China's first financial blockchain international standard approved.

Intel launches 11th‑generation low‑power Core Tiger Lake processors.

Linux From Scratch releases version 10.

Microsoft releases AI deep‑fake video detection tool.

Chinese scientists build the world’s largest neuron‑scale brain‑inspired computer.

Level 3 network outage caused a 3.5% global traffic drop.

TSMC uses AI/ML for chip‑production data processing.

Advances in Bayesian neural networks.

SIGIR 2020 paper on a two‑stage matching model for similar questions.

Academic News

At the ITU‑T SG16 plenary, the Financial Distributed Ledger Technology Application Guideline, led by the Digital Currency Research Institute of the People’s Bank of China together with the China Academy of Information and Communications Technology and Huawei, was approved, marking China’s first international financial blockchain standard.

Intel recently unveiled its 11th‑generation low‑power Core Tiger Lake processors built on 10 nm SuperFin technology. The CPUs feature the new Willow Cove architecture, up to four cores and eight threads, and can reach frequencies of 4.8 GHz, surpassing the 3.9 GHz of the previous Ice Lake Sunny Cove.

The Linux From Scratch project released version 10, offering a step‑by‑step guide for building a custom Linux distribution from source, which helps users deepen their Linux knowledge while creating a tailored OS.

Microsoft introduced a video authentication tool that performs frame‑by‑frame analysis to expose deep‑fake manipulations. The tool leverages Azure components to add digital hashes and certificates to videos, providing practical anti‑spoofing capabilities despite the rapid evolution of AI‑generated media.

Zhejiang University and the Zhijiang Laboratory successfully built China’s first brain‑inspired computer, the Darwin Mouse, based on domestically developed Darwin‑2 chips. The system integrates 792 chips, supports 120 million neurons and nearly a hundred billion synapses, consumes only 350–500 W, and represents the largest neuron‑scale brain‑inspired machine worldwide.

On August 30, a misconfigured Flowspec rule at CenturyLink/Level 3 prevented BGP announcements, causing a network failure that reduced global internet traffic by 3.5 %. The issue was resolved after resetting equipment and restoring a clean BGP routing table, a process that took about seven hours.

TSMC disclosed that it is applying artificial intelligence and machine learning to process data generated during chip manufacturing, enabling the company to identify production bottlenecks, predict defect sources, and improve yield for its advanced 7 nm process.

Bayesian neural networks, which can extend existing knowledge through probabilistic inference, are highlighted as a promising direction for next‑generation neural networks, and recent advances are summarized in a new review article.

A SIGIR 2020 paper from the Chinese Academy of Sciences proposes a two‑stage matching model for similar questions, using archived answers as bridges to improve question similarity detection, achieving better performance than previous state‑of‑the‑art methods.