IEEE calls for papers on autonomous optimization in networked AI

The IEEE Signal Processing Society is inviting submissions for a forthcoming particular difficulty of the IEEE Journal of Selected Topics in Signal Processing (JSTSP), specializing in what it describes as a rising subject: Autonomous and Evolutive Optimization in Networked AI.

In accordance with the decision for papers, the subject “represents a transformative paradigm for Sign Processing and Synthetic Intelligence (AI) communities”, combining conventional sign processing strategies with fashionable deep studying approaches.

The organizers say the strategy integrates “conventional knowledge-based adaptive sign processing strategies and data-centric deep-neural community fashions”, enabling techniques to “dynamically purchase high-quality information within the steady inferences of networked AI fashions”.

The idea facilities on networked AI techniques able to self-optimization by adaptive suggestions mechanisms. These techniques can “optimize each particular person mannequin by adaptively producing corresponding rewards and pseudo-labels on-line”, a course of the organizers say mirrors how advanced organizations evolve over time.

The decision for papers additionally highlights the potential to unify totally different machine studying paradigms, noting that such techniques “can unify supervised and reinforcement studying within the networking techniques of AI, by the adaptive sign processing”.

A key focus is on multi-agent techniques, the place distributed AI fashions work together dynamically. These interactions are mentioned to allow “autonomous self-optimization and evolution of networked AI, guaranteeing sturdy efficiency in time-varying environments with out human interventions”.

The scope of the particular difficulty spans a number of disciplines, together with sign processing, communications, and industrial automation. Advised utility areas embrace giant language fashions, autonomous driving techniques, and real-time 3D reconstruction.

The organizers state that the problem goals “to consolidate and develop the foundational rules of the adaptive and on-line optimization for networked AI fashions, and foster its developments in clever sign processing techniques”.

Submissions are open till June 15, 2026, with publication scheduled for January 2027. The particular difficulty is being led by Liang Tune of Fudan College, alongside visitor editors from establishments in Canada, Israel, Greece, and China.

Further details, together with submission guidelines, can be found through the IEEE Sign Processing Society.