Dimensional Encoding

Dimensional encoding focuses on representing data, particularly high-dimensional or complex data like images and time series, as lower-dimensional vectors to facilitate machine learning tasks. Current research explores various encoding methods, including those based on convolutional neural networks, random linear codes, and novel approaches tailored to specific data types (e.g., Laplace kernels for binary data, successive data injection for quantum computing). These techniques aim to improve efficiency, accuracy, and interpretability in applications ranging from anomaly detection and classification to reinforcement learning and feature selection, impacting fields like healthcare and communication networks.

Papers