15.8 C
New York
Sunday, October 2, 2022

Buy now

AI on the edge: 5 trends to watch

Edge AI offers opportunities for multiple applications. See what organizations are doing to integrate it today and in the future.

Virtual screen with stock chart and polygonal arrow hologram.
Image: Who is Danny/Adobe Stock

AI at the edge continues to evolve. Edge AI applications include: multiple: Autonomous vehicles, art, healthcare, personalized advertising and customer service can all benefit from it. Ideally, edge architecture delivers low latency as it is closer to the requests.

SEE: Don’t hold back your enthusiasm: Trends and challenges in edge computing (TechRepublic)

Astute Analytics Predicts the edge AI market will grow from $1.4 million in 2021 to $8 million in 2027, a CAGR of 29.8%. They expect this growth will come in large part from AI for the Internet of Things, wearable consumer devices and a need for faster computing in 5G networks, among other factors. These offer both opportunities and reservations because edge AI’s real-time data is vulnerable to cyber-attacks.

Take a look at five trends likely to shape edge AI in the coming year.

Top 5 edge AI trends

Separating AI from the cloud

One of the current changes at sea is the ability to perform AI processing without a cloud connection. Arm recently released two new chip designs that can bring the processing power for IoT devices to the edge, bypassing a remote server or the cloud. Their current Cortex-M processor can handle object recognition, while other capabilities such as gesture or speech recognition come into play with the addition of ARMs Ethos-U55. google’s Corala toolkit to build products with local AI also promises hefty AI processing “offline”.

Machine learning options

NVIDIA predicts that best practices in machine learning operations will prove to be a valuable business process for edge AI. It needs a new lifecycle for IT production – at least that’s the speculation as MLOps evolve. MLOps can help organize and push the data flow to the edge. A continuous cycle of updates can prove effective as more organizations discover what works best for them when it comes to edge AI.

Data scientists working on designing algorithms, choosing the model architectures, and implementing and monitoring ML on a day-to-day basis can benefit from simplified ML practices.

That means “it is possible for” neural nets to design neural netsGoogle CEO Sundar Pichai said.

Auto ML requires a lot of memory and computing power, so its implementation at the edge goes hand in hand with other considerations for continuous processing.

Specialized chips

To do more processing at the edge, companies need custom chips to deliver enough power. Last year, startup DeepVision made headlines with a $35 million Series B funding round for its video analytics and natural language processing chip for the edge.

“We expect 1.9 billion edge devices to ship with” deep learning accelerators in 2025,” explains Linley Group’s lead analyst Linley Gwennap.

DeepVision’s AI accelerator chip is coupled with a software suite that essentially converts AI models into math graphs. IBM has their first gear hardware in 2021, intended to combat fraud.

New computer vision usage scenarios and capabilities

Computer vision remains one of the prominent uses of edge AI. NVIDIA’s partner network for its application framework and set of developer tools today spans more than 1,000 members.

A major development in this area is multimodal AI, which draws from multiple data sources to go beyond natural language understanding by analyzing poses and performing inspection and visualization. This can be useful for AI that works seamlessly with humans, such as store assistants.

Higher-order vision algorithms can now classify objects using more detailed functions. Instead of recognizing a car, he can go deeper to locate the make and model.

Training a model to recognize which detailed features are unique to each object can be difficult. Approaches such as function representations with fine-grained information, segmentation to extract specific features, algorithms that normalize an object’s pose, and multi-layered convolutional neural networks are all current ways of making this possible.

Business use cases in their infancy include quality control, live supply chain tracking, identifying an internal location using a snapshot and detecting deep counterfeits.

Increased growth of AI on 5G

5G and beyond are almost there. Satellite networks and 6G waiting on the horizon for telecom providers. For the rest of us, there’s still some time to transition between core 4G networks that work with some 5G services before we can fully jump to the next generation.

Where does this touch the edge of AI? AI on 5G could be deliver better performance and security for AI applications. It can provide some of that low-latency edge AI, as well as open up new applications such as factory automation, toll collection and vehicle telemetry, and smart supply chain projects. Mavenir introduced edge AI with 5G for video analytics in Nov 2021.

There are more emerging trends in edge AI than we can fit into one list. In particular, its spread may also require some change on the human side. NVIDIA predicts that edge AI management will become a task for IT, probably using Kubernetes. Using IT resources instead of letting the industry manage edge solutions can: optimize costsGartner reported.

Source link

Related Articles


Please enter your comment!
Please enter your name here

Stay Connected


Latest Articles