Technology | October 28, 2019
A set of hardware and software systems that enable provisioning, management, and automation of connected devices in industrial settings. Not part of subscription
News Commentary | December 10, 2019
The success of deep learning models like BERT, for natural language processing (NLP), and Mask R‑CNN, for image recognition, has come in part from their massive size. This size, however, presents challenges for training, which in some cases can take days, thereby limiting iterative experimentation ... Not part of subscription
News Commentary | September 22, 2020
Deep learning models have grown from less than a billion parameters to models with nearly 200 billion parameters in just the past few years, leading to significant training challenges. With the updated release of its deep learning optimization library DeepSpeed, Microsoft is announcing a variety of ... Not part of subscription
by Christopher Robinson
Reducing congestion is at the top of many municipalities' lists of top challenges to address, and understanding traffic flow and causes of delays in intersections is crucial data to aid traffic engineers. To date, solutions have consisted of hardware and software solutions, such as Miovision's camera-based systems which can track vehicle, bicycle, and pedestrian data at intersections. Although effective, the solutions cost tens of thousands of dollars per intersection for transportation departments. Inrix's leveraging of anonymous vehicle data will likely be a significantly lower-cost solution, but it will struggle to capture data with the accuracy of camera-based setups.
For the original news article, click here .