AI‑Powered Supply Chain, LinkedIn’s Dagli ML Library, and Latest GNN Research

JD Opens AI‑Driven Intelligent Supply Chain Decision Engine

During the World Science and Development Forum in Beijing (Nov 8‑9), JD Group’s technology committee chair Dr. Zhou Bowen announced the full release of the JD Intelligent Supply Chain Decision Engine, highlighting its role in addressing VUCA‑era challenges such as price volatility, extended delivery cycles, and unclear product competitiveness. The engine was also listed on the China Association for Science and Technology’s “Science Innovation China” platform.

LinkedIn Open‑Sources Dagli Java Machine‑Learning Library

LinkedIn has released Dagli, an open‑source library for Java (and other JVM languages) that simplifies building, reading, modifying, maintaining, and deploying machine‑learning model pipelines. Dagli aims to reduce bugs and improve readability for experienced data scientists while providing an API that helps less‑experienced engineers avoid common logical errors.

PayPal Enables All U.S. Users to Trade Cryptocurrencies

PayPal announced that all eligible U.S. account holders can now buy, sell, and hold Bitcoin, Litecoin, Ethereum, and Bitcoin Cash, expanding beyond the limited rollout that began in October.

SpaceX Approved to Offer Starlink Internet Service in Canada

Canada’s regulator has granted SpaceX permission to provide Starlink satellite internet services within the country, marking the first expansion of the beta‑tested service outside the United States.

AppleWatch App NightWare Can Interrupt Nightmares Without Waking Users

The FDA‑approved NightWare app for Apple Watch detects REM‑stage movements and heart‑rate patterns indicative of a nightmare, then vibrates the wristband to interrupt the dream without waking the wearer. A prescription is required, and the app is intended to complement other treatments.

Intel Launches First Server‑Grade Independent GPU for Cloud Gaming and Streaming

Intel introduced its first server‑focused GPU, code‑named SG1, based on the Xe LP low‑power architecture. Designed for high‑density, low‑latency cloud gaming and media streaming, the product reflects Intel’s “XPU” vision of integrating CPUs, GPUs, FPGAs, and other accelerators.

Academic Paper: Graph Neural Networks for Multivariate Time‑Series Forecasting (KDD 2020)

The paper argues that multivariate time‑series forecasting requires modeling inter‑variable dependencies, which traditional statistical methods (VAR, GP) and many deep‑learning approaches fail to capture effectively. By representing each variable as a node in a graph, Graph Neural Networks can model these dependencies, offering a promising direction for accurate forecasting.

Academic Paper: Graph Convolutional Networks for Decoding Molecular Generation

The study proposes a two‑step decoder for molecular generation: first, a fully‑connected network predicts a molecular formula from a latent vector; second, a Graph Convolutional Network constructs bonds between the predicted atoms, enabling efficient molecular auto‑encoding.