Design Exploration of Hardware Accelerators For The K-NN Algorithm

Increasingly, machine-learning algorithms are playing an important role in the context of embedded and real-time systems. Applications such as wireless sensor networks, security, and commercial enterprises rely increasingly on machine-learning algorithms to efficiently make pre- dictive decisions based on the large volumes of data these systems collect. Most supervised machine-learning algorithms, however, require relatively large amounts of runtime to perform training and/or classification due to the size and dimensionality of the data they must work with. Therefore, there is a need to accelerate the runtime of these algorithms, especially for real-time applications. In this thesis, several different hardware accelerators are proposed and compared for the K-Nearest Neighbor (K-NN) classification algorithm. These accelerators are developed using Xilinx Vivado High-Level Synthesis (HLS) and Cadence Tensilica tools, and represent different (tightly coupled versus semi-tightly coupled) architectures. The experimental results, based on several benchmarks, show that hardware speedups for an HLS range from 48x-168x, while those obtained using Cadence Tensilica tools range from 86x-650x.