NeuroHash: A Hyperdimensional Neuro-Symbolic Framework for Spatially-Aware Image Hashing and Retrieval

Sanggeon Yun,Ryozo Masukawa,SungHeon Jeong,Mohsen Imani
2024-05-23
Abstract:Customizable image retrieval from large datasets remains a critical challenge, particularly when preserving spatial relationships within images. Traditional hashing methods, primarily based on deep learning, often fail to capture spatial information adequately and lack transparency. In this paper, we introduce NeuroHash, a novel neuro-symbolic framework leveraging Hyperdimensional Computing (HDC) to enable highly customizable, spatially-aware image retrieval. NeuroHash combines pre-trained deep neural network models with HDC-based symbolic models, allowing for flexible manipulation of hash values to support conditional image retrieval. Our method includes a self-supervised context-aware HDC encoder and novel loss terms for optimizing lower-dimensional bipolar hashing using multilinear hyperplanes. We evaluate NeuroHash on two benchmark datasets, demonstrating superior performance compared to state-of-the-art hashing methods, as measured by mAP@5K scores and our newly introduced metric, mAP@5Kr, which assesses spatial alignment. The results highlight NeuroHash's ability to achieve competitive performance while offering significant advantages in flexibility and customization, paving the way for more advanced and versatile image retrieval systems.
Computer Vision and Pattern Recognition
What problem does this paper attempt to address?
### Problems the Paper Aims to Solve The paper aims to address the problem of efficiently and accurately retrieving similar images in large-scale image datasets, particularly in terms of preserving spatial relationships within the images. Traditional deep learning-based hashing methods often fail to adequately capture spatial information and lack transparency. To tackle these issues, the paper proposes **NeuroHash**, a novel neuro-symbolic framework that combines High-Dimensional Computing (HDC) to enable highly customizable spatial-aware image retrieval. Specifically, NeuroHash integrates pre-trained deep neural network models with an HDC-based symbolic model, allowing flexible manipulation of hash values to support conditional image retrieval. The method includes a self-supervised context-aware HDC encoder and introduces a new loss term for optimizing low-dimensional bipolar hashing. Experimental results demonstrate that NeuroHash performs excellently on two benchmark datasets, showing significant advantages over existing hashing methods in the mAP@5K metric and introducing a new evaluation metric, mAP@5K r, to assess spatial alignment effectiveness. These results highlight NeuroHash's significant advantages in flexibility and customizability, paving the way for more advanced and versatile image retrieval systems.