Can LSH (Locality-Sensitive Hashing) Be Replaced by Neural Network?
This work addresses the problem of improving information-searching performance for developers by proposing a learning-based alternative to traditional hashing, though it appears incremental as it builds on existing neural network applications to data structures.
The paper tackles the problem of replacing traditional locality-sensitive hashing (LSH) with a neural network-based method, called LLSH, to map high-dimensional data to low-dimensional space more efficiently. The result shows that LLSH reduces time and memory consumption while maintaining query accuracy, as demonstrated in experiments on various datasets.
With the rapid development of GPU (Graphics Processing Unit) technologies and neural networks, we can explore more appropriate data structures and algorithms. Recent progress shows that neural networks can partly replace traditional data structures. In this paper, we proposed a novel DNN (Deep Neural Network)-based learned locality-sensitive hashing, called LLSH, to efficiently and flexibly map high-dimensional data to low-dimensional space. LLSH replaces the traditional LSH (Locality-sensitive Hashing) function families with parallel multi-layer neural networks, which reduces the time and memory consumption and guarantees query accuracy simultaneously. The proposed LLSH demonstrate the feasibility of replacing the hash index with learning-based neural networks and open a new door for developers to design and configure data organization more accurately to improve information-searching performance. Extensive experiments on different types of datasets show the superiority of the proposed method in query accuracy, time consumption, and memory usage.