Data Sharing Markets

Abstract

With the growing use of distributed machine learning techniques, there is a growing need for data markets that allows agents to share data with each other. Nevertheless data has unique features that separates it from other commodities including replicability, cost of sharing, and ability to distort. We study a setup where each agent can be both buyer and seller of data. For this setup, we consider two cases: bilateral data exchange (trading data with data) and unilateral data exchange (trading data with money). We model bilateral sharing as a network formation game and show the existence of strongly stable outcome under the top agents property by allowing limited complementarity. We propose ordered match algorithm which can find the stable outcome in O(N 2 ) (N is the number of agents). For the unilateral sharing, under the assumption of additive cost structure, we construct competitive prices that can implement any social welfare maximizing outcome. Finally for this setup when agents have private information, we propose mixed-VCG mechanism which uses zero cost data distortion of data sharing with its isolated impact to achieve budget balance while truthfully implementing socially optimal outcomes to the exact level of budget imbalance of standard VCG mechanisms. Mixed-VCG uses data distortions as data money for this purpose. We further relax zero cost data distortion assumption by proposing distorted-mixed-VCG. We also extend our model and results to data sharing via incremental inquiries and differential privacy costs.Preprint. Under review.

Knowledge Sharing in AI Services: A Market-Based Approach

“…Solving Problem 1 corresponds to finding a solution that maximizes the utility of each buyer s subject to budget constraints and clears 4 the market. Such an outcome is denoted as the tuple ( p, Ã), where the vector p = ( p1 , .…”

“…for instance [4]. Despite its simplicity, data sharing is often impractical due to multiple reasons, including privacy concerns related to the source data [5] or the problematic process of finding an agreement between different parties [6].…”

Learn to Match with No Regret: Reinforcement Learning in Markov Matching Markets

“…Broadly speaking, our work is also related to the research on learning economic models via RL. In particular, Kandasamy et al (2020); Rasouli andJordan (2021) studied VCG mechanisms, andGuo et al (2021) study exchange economies. Although similar in spirit, these models differ from our Markov matching markets in their mathematical structure, and thus have different solution concepts and planning methods.…”

On the Design of Decentralised Data Markets

“…2) Trust and Honesty Assumptions: Another possible categorisation of prior work relates to the trust assumptions made in the system. The work in [50] assumes that upon being shared, the data is reported truthfully and fully. In practise, this assumption rarely holds, and a mitigation for malicious behaviour in shared systems must be considered.…”