Search papers, labs, and topics across Lattice.
This paper introduces a maximum-margin Hyperdimensional Computing (HDC) classifier, improving upon standard HDC methods by leveraging a connection between HDC and Support Vector Machines (SVMs). The key result is a significant performance boost on benchmark datasets compared to baseline HDC approaches. The authors establish a formal relationship between HDC and SVMs, providing a theoretical foundation for the proposed method.
HDC, a resource-efficient ML method, gets a major performance boost by borrowing the large-margin principle from SVMs.
Overparameterized machine learning (ML) methods such as neural networks may be prohibitively resource intensive for devices with limited computational capabilities. Hyperdimensional computing (HDC) is an emerging resource efficient and low-complexity ML method that allows hardware efficient implementations of (re-)training and inference procedures. In this paper, we propose a maximum-margin HDC classifier, which significantly outperforms baseline HDC methods on several benchmark datasets. Our method leverages a formal relation between HDC and support vector machines (SVMs) that we established for the first time. Our findings may inspire novel HDC methods with potentially more hardware-oriented implementations compared to SVMs, thus enabling more efficient learning solutions for various intelligent resource-constrained applications.
