|
Abstract
|
Hashing has progressed from a simple constant-time lookup mechanism to a diverse set of algorithmic paradigms that underpin efficiency, security, and large-scale data management in modern computing. This survey organizes existing approaches into six principal categories: division-based hashing, dynamic hashing, cryptographic hashing, geometric and robust hashing, Bloom filter methods, and deep hashing. For each category, the paper outlines the key design principles, operational objectives, and characteristic performance trade-offs. The discussion connects these families of algorithms to their main application areas, including authentication, multimedia forensics, distributed storage, web systems, and approximate nearest-neighbour retrieval. Through comparative analysis, the survey emphasizes that the choice of hashing strategy is inherently context-dependent, shaped by constraints such as collision tolerance, memory efficiency, and semantic accuracy. Distinct from earlier reviews, this work brings together conventional and learning-based hashing techniques within a single analytical framework, highlighting the emergence of hybrid models that balance scalability, security, and similarity preservation in AI-driven and resource-limited environments.
|
|
Keywords
|
Hashing algorithms, dynamic hashing, cryptographic hashing, Bloom filters, deep hashing, robust hashing.
|