Multilinear Compressive Learning

Compressive learning (CL) is an emerging topic that combines signal acquisition via compressive sensing (CS) and machine to perform inference tasks directly on a small number of measurements. Many data modalities naturally have multidimensional or tensorial format, with each dimension tensor mode representing different features such as the spatial temporal information in video sequences spectral hyperspectral images. However, existing CL frameworks, CS component utilizes either random learned linear projection vectorized acquisition, thus discarding structure signals. In this article, we propose multilinear (MCL), framework takes into account nature signals step builds subsequent model structurally sensed Our theoretical complexity analysis shows proposed more efficient compared its vector-based counterpart both memory computation requirement. With extensive experiments, also empirically show our MCL outperforms object classification face recognition tasks, scales favorably when dimensionalities original increase, making it highly for high-dimensional