Recent advancements in edge AI are focusing on enhancing efficiency and adaptability across various applications, particularly in resource-constrained environments. Techniques like model stitching for multi-DNN inference are enabling systems to dynamically recombine model subgraphs, significantly reducing service level objective violations and improving throughput. On-board processing solutions, such as those utilizing compact segmentation networks for satellite imagery, are addressing the challenges of data transmission by generating actionable insights directly in orbit, thus minimizing energy costs. Furthermore, frameworks that facilitate on-device training of deep learning models are emerging, allowing for real-time adaptation while preserving privacy. Innovations in retrieval-augmented generation are also being tailored for edge devices, enhancing personalization in noisy environments. Collectively, these developments are poised to solve commercial challenges related to latency, energy efficiency, and data privacy, paving the way for more robust and responsive edge AI systems across industries.
Modern edge applications increasingly require multi-DNN inference systems to execute tasks on heterogeneous processors, gaining performance from both concurrent execution and from matching each model ...
Efficient and adaptable deep learning models are an important area of deep learning research, driven by the need for highly efficient models on edge devices. Few-shot learning enables the use of deep ...
General aviation fault diagnosis and efficient maintenance are critical to flight safety; however, deploying deep learning models on resource-constrained edge devices poses dual challenges in computat...
Accurate sea ice mapping is essential for safe maritime navigation in polar regions, where rapidly changing ice conditions require timely and reliable information. While Sentinel-1 Synthetic Aperture ...
On-device tuning of deep neural networks enables long-term adaptation at the edge while preserving data privacy. However, the high computational and memory demands of backpropagation pose significant ...
Personalized virtual assistants powered by large language models (LLMs) on edge devices are attracting growing attention, with Retrieval-Augmented Generation (RAG) emerging as a key method for persona...
This paper introduces DMind-3, a sovereign Edge-Local-Cloud intelligence stack designed to secure irreversible financial execution in Web3 environments against adversarial risks and strict latency con...
\emph{Integrated communication and computation} (IC$^2$) has emerged as a new paradigm for enabling efficient edge inference in sixth-generation (6G) networks. However, the design of IC$^2$ technologi...
Smart shipping operations increasingly depend on collaborative AI, yet the underlying data are generated across vessels with uneven connectivity, limited backhaul, and clear commercial sensitivity. In...
The rapid evolution toward 6G and beyond communication systems is accelerating the convergence of digital twins and world models at the network edge. Traditional digital twins provide high-fidelity re...