Artificial Intelligence for Smart Cities
By Martin Jezequel, Product Manager Datacenter Solutions and Mohamed Ali Hammami, AI Edge & Iot Program Manager – PNY Technologies
“What are the factors accelerating the use of artificial intelligence and what impact could this technology have on our daily lives?
Many experts describe the current era as a “fourth industrial revolution” driven by artificial intelligence (AI), robotics, genome editing and neurotechnology. More than three billion people now live in megacities. This exponential increase exacerbates the problems of living in cities, leading us to intensively use AI as a predictive and recognition technology to both reduce the cost of data processing and optimize the processes involved in our activities. In the context of a city, a multi-use deployment of IoT Internet of things) will positively impact interactions ans increase the level of safety of residents.
Could you explain some concrete example of the use of AI in a smart City ?
PNY Technologies, as a global provider of AI solutions, is working on artificial intelligence projects at the edge or EDGE AI. This kind of architecture is deployed where the information is created, therefore outside the data center or the cloud. An emblematic example of this type of solution in the context of smar cities is the management of road traffic with the installation of computational units based on NVIDIA Jetson embedded graphics cards or computer models specifically trained to recognize different use cases such as the identification of marks, colors, plates, of vehicles on the road, recognition of risky behaviour of drivers, management and forecasting of road traffic, etc.” – Mohamed Ali Hammami
“How do we like EDGE AI to data center AI and deep learning ?
To get EDGE, you must first have organized raw data, built a neural network, trained a model, in order to make inference in the periphery, such as traffic monitoring, or control, or violations. The data, which in the context of Smart Cities can be photos of cars, traffic signs, road network diagrams, all by the thousand, are organised and labeled data and your specific data, you now have to feed it into your model to train it and get an accurate result. This is where the power and computational efficiency of the hardware is paramount. After this critical training phase, you will have a very valuable result that is your trained model, totally independent of the initial process, ans that can be integrated into an Edge AI device like a smart camera at a crossroads, which will feed itself new information in real time and operate in total autonomy.
What products in your PNY offering address this need ?
GPU-based AI computing technologies have evolved exponentially in recent years. The time savings on computation offered by these solutions, going from several months to only a few days of training, largely justifies their investment. We offer NVIDIA’s AI block based on the world’s leading GPU architecture, the DGX A100. We also offer all the software interface related to NVIDIA hardware, as well as the NVDIDA Networking solutions for which we are direct distributors since April 2021. In addition, we have partnerships with global storage players such as NMS and Net App. The optimization of deep learning and training is conditioned by the performance and good orchestration of this Compute, Networking and Storage package. This ability to offer complete solutions with high added value places PNY as a major player in the field of AI and machine learning.” – Martin Jezequel
