Top Tech Highlights: GKE Autopilot, Power Fx, Quantum Assembly, and More

GKE Autopilot for Kubernetes

Google announced GKE Autopilot, a managed Kubernetes service that automates node provisioning, maintenance, and lifecycle management, locking nodes to prevent unsupported changes and optimizing clusters for production workloads.

Microsoft Power Fx Low‑Code Language

Microsoft plans to launch Power Fx, a low‑code language based on Excel formulas for the Power Platform, expected at the Microsoft Ignite 2021 conference to enable non‑programmers to develop on Power Platform.

Quantum Assembly Language (JAQAL)

Researchers at Sandia National Laboratories introduced JAQAL (Just Another Quantum Assembly Language), an open‑source assembly language for quantum computers running on the QSCOUT platform, offering fine‑grained control over quantum operations.

JD Cloud “Jinggang” Virtualization Architecture

JD Cloud unveiled Jinggang, a self‑developed hardware‑software integrated virtualization architecture that supports cloud hosts, bare metal, and native containers, featuring a smart chip that offloads virtualization tasks to hardware for high‑speed data forwarding.

Google‑Intel Cloud‑Native 5G Collaboration

Google Cloud and Intel announced a joint effort to develop cloud‑native 5G services and technologies, including pre‑integrated network function platforms, edge service blueprints, and collaboration with software vendors for validation and testing.

Nvidia Jarvis 1.0 Public Beta

Nvidia released Jarvis 1.0, a GPU‑based framework for building conversational AI services, featuring ASR, NLU, and TTS models trained on thousands of hours of speech data, and enabling end‑to‑end deployment with a single line of code.

WWW 2021: Knowledge‑Infused BERT Attention Model

A paper presented at WWW 2021 proposes a simple and effective method to incorporate external knowledge into BERT for text semantic matching, improving performance without altering BERT’s architecture.

FACS Regional Detection

A study introduces a region‑based attention network for facial action unit (AU) detection, dividing the face into three zones and using hard and soft attention masks along with SE modules to improve AU detection accuracy.