Call for Papers

The purpose of our Workshop is to bring together researchers, developers, and practitioners in artificial intelligence, adaptive computing, and high performance computing to identify common interests, requirements, capabilities, and solutions in edge sensing and processing.

An increasingly instrumented world has produced massive amounts of data of which only a small fraction is necessary to facilitate knowledge discovery. Managing and gleaning value from the data has become a complex problem demanding new AI-enabled adaptive computational solutions for reducing complexity and accelerating inference at the edge. We are especially interested in approaches that operate in and are robust to highly dynamic environments driven by changes in technology, opportunities, and available data sets. Further, we are especially interested in algorithms that can manage the dynamism on resource-constrained edge devices at the point of need. We need to package and deploy different elements, from data sourcing and processing, trained AI models, to software services and domain-specific applications. The research recognizes that commercial developments contributed to the open source community have yielded dramatic increases in our capabilities; however, most off-the-shelf tools are not designed for the edge. Research has multiple objectives: diminish bottlenecks, minimize data movement, reduce computational complexity, accelerate deployment of new capabilities, and present actionable results with reduced cognitive workload on the end-user.

Large language models (LLMs) and transformers are rapidly expanding in capability and breadth of application. We are interested in LLM / transformer model quantization and other strategies for achieving low latency, minimally lossy inference at the edge; the uses of LLMs for engineering highly adaptive systems for the edge, e.g. code and configuration generation for adaptable data processing; and adapting vision transformer-based approaches for image processing and analysis for the edge. The flexibility of machine learning algorithms presents an opportunity to dynamically adapt task execution and implement results throughout a network to meet changing requirements and available resources. Research is required to utilize adaptive edge capabilities, high performance computing, model reduction, and workflow and service deployment frameworks to produce a converged solution.

We will select the top six papers from the Workshop, edit them to conform to journal requirements, and submit the collection as a special section of an IEEE journal following the Workshop.

Relevant topics include:

•  Adaptive AI for Resource-Constrained Edge Devices

• Robust and Trustworthy AI in Dynamic Edge Environments

• Co-Design of AI Models and Hardware for Edge Acceleration

• Intelligent Data and Workflow Management at the Edge

• Efficient Model Compression and Deployment for the Edge

• Convergence of High-Performance Computing (HPC) and Edge AI

• Synthetic Datasets for model training and finetuning

• Advancing Edge Quantum Processing Units (QPUs) for Hybrid Classical–Quantum Neural Network Architectures

Submission Guidelines:

Papers should be no more than 10 pages in length, 2-column format. Submitted papers must be original work and not previously published or under consideration for another conference or journal.

Click buttons for conference template and paper submission.

Workshop Chairs

• • Dr. J. Michael Barton, Army Research Laboratory DOD Supercomputing Resource Center & Parsons

  • Dr. Venkat Dasari, DEVCOM Army Research Laboratory Research Office

  • Mr.  Michael An, Army Research Laboratory DOD Supercomputing Resource Center & Parsons

  • Dr. Raju Namburu, Army Engineer Research and Development Center

Program Committee

• Ms. Virginia To, Energetics Technology Center

• Prof. Ju Wong, Virginia State University

• Dr. Michael Stokes, SAIC

• Prof. Miachel McGary, University of Texas - El Paso