Optimizing Machine Learning at Scale: A Review of Efficient Data Structures

As a result of the exponential growth of machine learning application, it presents new challenges to satisfy the needs of computational efficiency, scalability and memory optimization. With the growing datasets getting large and complex, data structures to store this data impact the performance of the entire machine learning systems. In this paper, we provide a thorough review of such efficient data structures appropriate for large scale machine learning. In the analysis we present tree based, graph based, hashing, and diffused data structures paying special attention to computational complexity, memory usage and scalability. In addition, we compare different data structures and their appropriateness for given machine learning problems. We present our findings based on an extensive scarification study from the literature review and the case studies, and discuss the advantages and disadvantages of each approach, and identify the future research direction. The goal of this review is to provide a useful resource for researchers and practitioners in finding the most appropriate way to handle large scale machine learning workloads.