In a recent interview at the AI & Big Data Expo, Alessandro Grande, Head of Product at Edge Impulse, delved into the challenges of developing machine learning models for edge devices with limited resources and how to overcome them.
Grande shared valuable insights on the current obstacles, how Edge Impulse is working to address these challenges, and the significant potential of on-device AI.
Primary challenges in edge AI adoption
Grande outlined three main pain points that companies encounter when trying to implement edge machine learning models, including difficulties in determining optimal data collection strategies, a shortage of AI expertise, and communication barriers between hardware, firmware, and data science teams.
“Many companies developing edge devices lack familiarity with machine learning,” Grande noted. “Bringing these two worlds together poses the third challenge of enabling teams to communicate effectively, share knowledge, and collaborate towards common goals.”
Approaches for efficient and lightweight models
When asked about optimizing for edge environments, Grande stressed the importance of minimizing the amount of sensor data required.
“We see many companies struggling with datasets. They are unsure of what data is necessary, which sensors to collect data from, and how to collect the data,” Grande explained.
Choosing efficient neural network architectures and utilizing compression techniques like quantization can help reduce precision without significantly impacting accuracy. It’s crucial to balance sensor and hardware limitations with functionality, connectivity requirements, and software needs.
Edge Impulse aims to empower engineers to validate and verify models independently before deployment using standard ML evaluation metrics, ensuring reliability and speeding up time-to-value. The comprehensive development platform seamlessly integrates with major cloud and ML platforms.
The potential of on-device intelligence
Grande highlighted innovative products that leverage edge intelligence to offer personalized health insights without relying on the cloud, such as the Oura Ring for sleep tracking.
“It has sold over a billion units, demonstrating the power of edge AI that everyone can experience,” Grande emphasized.
Other exciting opportunities include using anomaly detection for preventative maintenance in industrial settings.
Grande envisions on-device AI significantly enhancing utility and user experience in daily life. Edge devices can interpret sensor inputs to provide actionable recommendations and responsive experiences, leading to more practical technology and improved quality of life.
Realizing the potential of AI on edge devices requires overcoming current barriers to adoption. Grande and other industry experts shared profound insights at this year’s AI & Big Data Expo on breaking down these barriers and unlocking the full potential of edge AI.
“I envision a world where our devices are more beneficial to us,” Grande concluded.
Watch the full interview with Alessandro Grande below:
(Photo by Niranjan _ Photographs on Unsplash)
See also: AI & Big Data Expo: Demystifying AI and seeing past the hype
Interested in learning more about AI and big data from industry experts? Explore the AI & Big Data Expo events in Amsterdam, California, and London. These comprehensive events are co-located with Cyber Security & Cloud Expo and Digital Transformation Week.
Discover other upcoming enterprise technology events and webinars powered by TechForge here.
