This is Part 3 in our series on Generative AI for Edge Computing. Click here to read Part 1 and here to read Part 2.
From generating training data to co-designing advanced new model architectures, Generative Artificial Intelligence (generative AI) promises to expand the possibilities in the Edge AI landscape both now and in the future. We have already begun testing and exploring what we can do with this technology to elevate the user experience and empower a new generation of machine learning on edge devices like we have never seen before.
“We‘ve started experimenting using generative AI for processing and training data to create more off-the-shelf DEEPCRAFT™ Ready Models,” says Alexander Samuelsson, CTO at Imagimob, an Infineon Technologies company. “We have even started incorporating the tool into our research and development efforts as we test new possibilities and imagine future versions of our development platform.”
Here we will continue our conversation with Alexander as we walk through some of the ways we are leveraging the benefits of generative AI for customers both today and in the future.
The possibility of generating and extracting synthetic data with generative AI is expected to benefit DEEPCRAFT™ Ready Model customers as well as users of DEEPCRAFT™ Studio who want to develop their own applications.
“Generative AI is helping us accelerate the output and quality of the next generation of Ready Models, which will result in even better performance and a greater variety of models to choose from for customers who don’t want to invest in developing from scratch,” says Alexander.
“In the future, DEEPCRAFT™ Studio users will be able to accelerate development too using our generative AI-enhanced Data Augmentation functionality which will help them generate more data variants for the training scenarios they need, and ultimately create more robust machine learning models.”
Collecting, synthesizing, and verifying data can sometimes be difficult, and is often time consuming. However, generative AI is expected to help bridge the gap and this is currently being explored as a potential new functionality in DEEPCRAFT™ Studio.
“This is where we see huge potential value for customers, because we can generate more data, and more variations of that data,” says Alexander. “If we can remove this barrier, we can really level up the quality of data they can source and, in turn, the breadth of application possibilities they can explore.”
DEEPCRAFT™ Studio users can look forward to more on-demand assistance during the modeling process. We are currently looking into ways to incorporate chat-like AI functionalities into the platform to provide guidance for users, making it easier for them to quickly navigate obstacles and improve their models. For example, if our auto-ML wizard was a chat interface with an LLM, you could give it a better understanding of your use case for more intelligent model generation.
“We believe that having an assistant that helps you determine next steps when you encounter difficulties can be invaluable for our customers,” says Alexander. “You can ask it questions about how to improve your model or brainstorm ideas. For example, you can inquire, 'What should I do next?' or 'How can I enhance my model?' This will significantly simplify our customers' workflow and the quality of their Edge AI applications.”
Looking ahead at future versions of development tools, we are also researching new model architectures and new types of network layers that could be derived from generative AI in order to help improve code and model generation.
“DEEPCRAFT™ customers will benefit from these kinds of advancements in our tools in the future which will lead to their models being generally improved,” says Alexander.
Our exploration of ways to integrate generative AI into our development tools and offerings is ongoing, and the future looks exciting.
Harnessing the capabilities of generative AI may not only help us optimize our Edge AI development tools and expand our DEEPCRAFT™ Ready Model offerings, but it may also enable our users to tackle complex challenges with greater confidence and launch powerful new Edge AI applications to the market faster than ever before.
The future holds exciting possibilities for machine learning on edge devices, and we are dedicated to ensuring that our users are well-equipped to make the most of it.
Don't miss out on the latest blogs, news and more from Imagimob!
Subscribe to our monthly newsletter to stay up-to-date on all the latest blogs, news, events, webinars and more from Imagimob.