PhD or Post-Doc on "Decentralised AI for resource-constrained Edge Systems"
National Research Council (CNR) of Italy
Pisa, Italy
21h fa
source : Euraxess

The Ubiquitous Internet Research Unit of IIT-CNR (Pisa, Italy) is scouting for talented candidates for PhD and post-doc topics.

The scouting is at both PhD and post-doc levels for this specific topic ( Decentralised AI for resource-constrained Edge Systems ).

Interested people are requested to send an expression of interest to the provided contact as described in the following, for possibly scheduling a specific session of p2p discussion on the topic.

Decentralised AI for resource-constrained Edge Systems

Deep Neural Networks are knowingly powerful yet resource-demanding machine learning tools. Currently, machine learning systems are moving the execution of both training and inference tasks from powerful and remote data centres where all data is available in a centralized fashion to more pervasive and distribute / decentralised systems where devices are far less powerful and can access only a limited quantity of data (typically the one generated in their physical proximity).

Performing AI tasks on resource-constrained devices is a cutting-edge research area that poses several exciting methodological and technological challenges.

Such decentralised AI schemes are appropriate for all scenarios where data cannot be moved from where they are generated, for privacy, confidentiality or data sovereignty reasons.

Typical examples range from industrial applications to privacy-preserving ML.

We are looking for a PhD student working on this topic. Specifically, the PhD thesis will target both aspects : i) the efficient distributed / decentralised training of models between constrained devices ii) the optimization of the training and inference of a model on resource-constrained devices.

The challenges connected to this problem include the development of compression algorithms pruning) for DNNs, the definition of new lightweight Neural Network models, the development of new training algorithms for lightweight models and the development solution to perform real-time inference.

The PhD thesis will focus on combinations of the following aspects

  • Cooperative decentralised / distributed ML algorithms with or without central controllers
  • Definition of new lightweight Neural Network architectures working on resource-constrained devices
  • Compression of ML structures to work on resource-constrained devices
  • The activities will be carried out in the framework of the European projects (Social Explainable AI) and H2020

    Ubiquitous Internet Research Unit

    The Ubiquitous Internet RU is part of the Institute for Informatics and Telematics of CNR, located in Pisa, Italy. Pisa hosts the biggest CNR campus in Italy and is the home of three prestigious universities, the University of Pisa, Scuola Normale Superiore, S.

    Anna School of Advanced Studies. Pisa research institutions have a long heritage of excellence in Computer Science, dating back from the first Italian computer (CEP, 1961) and the first Italian connection to the Internet (1986).

    The activities of the Ubiquitous Internet Research Unit range over multiple topics related to the design and analysis of the Next Generation Internet (NGI) networking and computing systems, including edge computing, Internet of People, decentralised AI, human-centric networked systems, online / mobile social networks, data-centric and self-organising networks.

    Verticals of interest include Industry e-health, energy efficiency, smart mobility. The RU has a strong track record of successful activities in national and European projects, from FP6 to H2020, which is reflected in the many international collaborations in the EU and USA activated by the researchers of the RU.

    Currently, active reference projects include :

  • AI and BigData : H2020 HumanAI-Net, SoBigData++, CHIST-ERA SAI (Social Explainable AI)
  • Edge computing & decentralised AI : H2020 MARVEL, PON-MIUR OK-INSAID
  • Next-Generation Internet Infrastructures : H2020 SLICES-DS, SLICES-SC
  • Quantum Computing & Networking : H2020 HPC-QS, PON-MIUR QUANCOM
  • Please refer to the publication records of the contact persons for more specific information.

    Offer Requirements

  • REQUIRED EDUCATION LEVEL Computer science : Master Degree or equivalent
  • Skills / Qualifications

    Ideal candidates should have or about to obtain an MSc degree (for the PhD topic) or PhD degree (for the Post-doc topic) in Computer Science, Computer Engineering, Mathematics, or closely related disciplines, and a proven track record of excellent University grades (PhD topic) or publications in relevant top-tier conferences and journals (Post-doc topic).

    Preferably, the MSc / PhD thesis topic should be in one of the relevant research areas (Artificial Intelligence, BigData analytics, distributed systems).

    Good written and spoken communication skills in English are required.

    Segnala questo annuncio

    Thank you for reporting this job!

    Your feedback will help us improve the quality of our services.

    Invia candidatura
    La mia Email
    Cliccando su “Continua”, autorizzo neuvoo ad utilizzare i miei dati ed inviarmi avvisi email come menzionato nella sezione Politica sulla Privacy di neuvoo. Posso ritirare il mio consenso e cancellare la registrazione in qualsiasi momento.
    Modulo di candidatura