Resource trading strategies with risk selection in collaborative training market

Area of Science:

• Edge Computing
• Artificial Intelligence
• Machine Learning

Background:

• Collaborative training in edge computing and AI is gaining traction.
• Existing research primarily focuses on technical resource allocation, neglecting economic trading mechanisms.

Purpose of the Study:

• To investigate effective budget allocation strategies for computational and data resources in co-training.
• To develop economic models for resource trading and market equilibrium in collaborative training environments.

Main Methods:

• Utilized Constant Proportion Portfolio Investment for payoff maximization under budget constraints.
• Employed Swing Gradient Search Algorithm to determine optimal resource acquisition strategies.
• Developed stepped and smoothed pricing algorithms to maintain market equilibrium.

Main Results:

• The proposed strategies and algorithms demonstrate high algorithmic efficiency and strategic effectiveness.
• The economic framework successfully addresses the coupling between resource acquisition quantities.
• The pricing algorithms effectively maintain dynamic market equilibrium.

Conclusions:

• The study provides a robust economic framework for resource allocation in collaborative training.
• The developed algorithms offer practical solutions for optimizing budget allocation and market stability.
• This research bridges the gap between technical implementation and economic viability in edge AI co-training.