Using the Singular Value Decomposition to Generate Composite NFTs

Abstract

Blockchain and Artificial Intelligence (AI) are two rapidly emerging technologies. The intersection between them typically focuses on areas where the blockchain can improve trust and transparency in the decisions made by AI models. Yet few practical use cases exist that demonstrate where AI can be used to benefit blockchain. In this paper, we identify Non-Fungible Tokens (NFTs) as a concrete example of how techniques developed in AI can enhance the functionality of blockchain systems and enable new use cases. Specifically, we explore how singular value decomposition (SVD), a fundamental tool in the field of machine learning, can be used in several ways to generate composite NFTs. We show how using the SVD to generate NFTs can allow their more efficient storage on public blockchains as low-fidelity thumbnails that can be provably linked to the original full image. Building on this design, we outline how composite NFTs can be constructed by combining different components of an NFT image derived using the SVD. These components can either be made public or kept private to allow for complex functionality in novel NFT protocols.