PAISE 2023


5th Workshop on Parallel AI and Systems for the Edge

PAISE 2023 will be an interaction-focused workshop, co-conducted with IPDPS 2023 on 19 May in St. Petersburg, Florida USA.


Past Editions:

2022 2021 2020 2019

From applications to hardware platforms, Edge computing is rapidly transforming the computing landscape. Taking advantage of the waning pandemic, this year PAISE will adopt a hybrid format, prioritizing interaction focused sessions over traditional technical talks. Toward this, PAISE invites two-page extended abstracts in addition to full papers. The objective of the new format is to enable impromptu opinionated discussions that augment traditional paper presentations, resulting in post-workshop position papers authored by the workshop participants. Thus, the workshop will provide a critically needed opportunity to discuss the current trends and issues, to share visions and opinions, to collect feedback and to discuss solutions covering the following areas of edge computing:

  • applications — computer vision, machine learning, analytics, IoT;
  • data flows — processing pipeline of data from ingestion to archival, pipeline of AI from learning to inference;
  • control flows — parallel and distributed programming models and runtimes for managing constrained resources, cybersecurity; and
  • infrastructure — storage, compute, and connectivity conducive to resource-constrained and harsh edge environments.

Topics

Full technical papers, and extended abstracts on opinions, visions, positions, and solutions along the following topics are welcome:

    • AI and IoT Applications at the Edge
    • Collaborative and Distributed Learning at the Edge
    • Cyber-Security and Privacy in Edge Computing
    • DevOps across Edge and Cloud
    • Data and AI Lifecycle Mgmt across Edge and Cloud
    • Edge-driven HPC, and HPC-steered Edge Computing
    • Energy Efficient Edge Hardware and Software
    • Serverless and other Programming Models for Edge
    • Multitenancy at the Edge

Paper Submission, Paper Style, and Proceedings

All papers must be original. The papers submitted to the workshop will be peer reviewed by a minimum of 3 reviewers.

The following paper categories are welcome:

  • Full Papers: Full research papers should describe original work and not simultaneously submitted to another journal or conference, and be 7-8 pages in length. The full papers will be presented as 20 min talks.
  • Short Papers: Position papers, visions, concepts, demo descriptions, or practice reports, 2 pages in length, should contain enough information for the program committee to evaluate whether the submission will generate discussion at the Workshop. Recently published work that is relevant to the Workshop is also welcome, although those submissions will not be included in the proceedings. The short papers will be presented as 5 min talks to seed the discussions following them.
  • Accepted papers will be included in the IPDPS workshop proceedings. With the objective of building the community, the organizers will facilitate piecing together a position paper reflecting the workshop discussions, co-authored by the participants and published at a suitable venue after the workshop. For more details on paper submission instructions, please visit the workshop webpage (https://paise.org).

    Templates for MS Word and LaTeX provided by IEEE eXpress Conference Publishing are available for download. See the latest versions here.

    Here is a link to the EasyChair CFP. Upload your submission to EasyChair submission server in PDF format. Accepted manuscripts will be included in the IPDPS workshop proceedings.

    Important Dates:

    • January 23rd AOE, 2023: Submission deadline.
    • February 20th, 2023: Notification of acceptance.
    • February 28th, 2023: Camera ready papers due.
    • May 19th, 2023: Workshop.

    Background

    Applications involving voluminous data often necessitate the computing to be performed as close to the data source as possible, due to communication constraints, latency requirements, privacy and sensitivity of data as well as costs associated with moving it. Given the recent advances in algorithms and techniques, as well as the increasing and improving last-mile wireless connectivity with the emergence of 5G and Wi-Fi 6, more and more application scenarios with the above requirements are becoming a reality. However, this reality comes with its own set of challenges for the applications as well as the wide range of edge computing platforms that support them.

    First of all, these applications have various runtime requirements (e.g., continuous versus event-driven) and resource demands (e.g., GPUs). Similarly, the platforms they run on are diverse in terms of their architectures, hardware capabilities and programming models, spanning from intelligent embedded devices (e.g., smart cameras) to on-premise systems (e.g., small-scale server racks). Due to the usually limited capacity at the edge for computation, network bandwidth and energy, sharing such heterogeneous resources among applications with different, and sometimes conflicting, requirements becomes an imminent challenge with multi-tenancy considerations.

    Second, the allocation and orchestration of these limited computing and network resources will experience many challenges currently encountered in cloud computing, but with the complexity and heterogeneity of the edge. Perhaps, the largest DevOps and management problems for the infrastructure will be seen while devising mechanisms requiring cooperation and coordination of various parts of the software stack, including the data and control plane of the applications to fine-tune their behavior and change their operational parameters. Coupling of these applications with their centrally located cloud and HPC counterparts will increase their effectiveness but also create new challenges.

    Furthermore, as we push more toward edge-enabled networks of devices, we inherit a setting where resources are deployed away from the safety of secure indoor spaces, often in the midst of a bustling urban canyon, and exposed to physical and cybersecurity threats. Deployed and interconnected predominantly over public networks, these systems have to be designed with cybersecurity as a first-class design citizen, ensuring not only the integrity and confidentiality of the data, but also the correct and accountable processing of it.

    At PAISE, we aim to discuss these topics from the perspectives of different members of the edge community, ranging from software to hardware researchers as well as from academia to industry. Our ultimate goal is to identify gaps in our thinking and to foster collaboration for addressing those gaps.

    Tentative Program Committee (based on 2022)

    • Utku Gunay Acer, Nokia Bell Labs
    • Paarijaat Aditya, Nokia Bell Labs
    • Marco Brocanelli, Wayne State University
    • Kevin Chan, Army Research Laboratory
    • Ruichuan Chen, Nokia Bell Labs
    • Lucy Cherkasova, ARM Research
    • Nicolas Erdody, Open Parallel
    • Nicola Ferrier, University of Chicago
    • Dhiraj Joshi, IBM Research
    • Takayuki Katsuki, IBM Research Tokyo
    • Hana Khamfroush, University of Kentucky
    • Dawei Li, Amazon Inc.
    • Eric Matson, Purdue University
    • Chulhong Min, Nokia Bell Labs
    • Alessandro Montanari, Nokia Bell Labs
    • Priya Panda, Yale University
    • Michael Papka, Northern Illinois University
    • Sekou Remy, IBM Research Africa
    • Koichi Shinoda, Tokyo Institute of Technology
    • Eric Van Hensbergen, ARM Research
    • Blesson Varghese, Queen's University,Belfast
    • Yuxiong Wang, University of Illinois
    • Wei Wang, The Hong Kong University of Science and Technology
    • Feng Yan, University of Nevada, Reno
    • Kazutomo Yoshii, Argonne National Laboratory


    General Chairs

    Workshop Organizers