FairMLancer: Dynamic Resource Allocation and Fairness Optimization in Decentralized Systems for Machine Learning Freelancing

Abstract

Decentralized systems have caused a paradigm change in several industries, including freelancing. Decentralized systems have the ability to improve transactional efficiency, lower costs, and promote transparency because of their disruptive potential, these platforms have piqued the interest of both scholars and practitioners. Centralized freelance platforms are frequently chastised for a lack of openness and trustworthiness, which can result in unfavorable encounters for both clients and freelancers. According to, traditional freelance platforms frequently struggle with gaining consumer trust since there is no guarantee that clients' data will be handled safely or that the work will be of high quality. This can lead to a lack of confidence in the platform, resulting in lower user engagement and significant revenue loss. Furthermore, payment issues and job accessibility are common issues for freelancers using centralized platforms .The emergence of decentralized systems backed by blockchain technology has transformed a variety of sectors, including machine learning freelancing. "FairMLancer" is a decentralized Ethereum-based platform for machine learning freelancing. It handles dynamic resource allocation and fairness optimization in tabular data projects. Freelancers may be assigned to tasks, and the platform assesses models using IPFS test datasets. Notably, FairMLancer uses co-operative game theory and the IPF method to distribute fair remuneration based on RMSE results. The top three models are chosen to maximize customer outcomes. This study presents a unique method for fair compensation in decentralized systems, which is enhanced using a tabular GAN model to produce synthetic data for training datasets. FairMLancer's mission is to promote equity and transparency in the machine learning freelancing industry or the gig economy.