XuILVQ: A River Implementation of the Incremental Learning Vector Quantization for IoT

Description of an edge-computing algorithm for incremental classification problems

to the number of prototypes it stores in memory.

Asset Description

In Machine Learning, Incremental Learning algorithms provide a solution for models that need to dynamically adapt and react to their context by analysing samples from data streams. These algorithms are especially suitable for Internet of Things (IoT) solutions where devices (like sensors/actuators) have low memory and computation capability. Within this context, we propose an implementation of an Incremental Learning Vector Quantization (ILVQ) algorithm compliant with the well-known River library standards.
This asset provides not the code but a well-explained description of the proposed algorithm including pseudo-code.

Usage

The publication is available in this link and reference:
González Soto, M., Fernández Castro, B., Díaz Redondo, R. P., & Fernández Veiga, M. (2022, October). XuILVQ: A River Implementation of the Incremental Learning Vector Quantization for IoT. In Proceedings of the 19th ACM International Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, & Ubiquitous Networks (pp. 1-8). https://doi.org/10.1145/3551663.3558676

Resources

• Paper available in ACM Digital Library (DL)
• Created by: - Gradiant (Centro Tecnológico de Telecomunicaciones de Galicia)

Acknowledgement

The paper and the associated work was carried out in an internal project developed in Gradiant.