GT-PCA: Effective and Interpretable Dimensionality Reduction with General Transform-Invariant Principal Component Analysis
This addresses the need for interpretable and transformation-invariant dimensionality reduction in fields like image and time series analysis, though it appears incremental as it builds on PCA and autoencoders.
The paper tackles the problem of dimensionality reduction lacking robustness to transformations like rotations or shifts, introducing GT-PCA as an effective and interpretable alternative that significantly outperforms PCA and autoencoders in experiments on synthetic and real data.
Data analysis often requires methods that are invariant with respect to specific transformations, such as rotations in case of images or shifts in case of images and time series. While principal component analysis (PCA) is a widely-used dimension reduction technique, it lacks robustness with respect to these transformations. Modern alternatives, such as autoencoders, can be invariant with respect to specific transformations but are generally not interpretable. We introduce General Transform-Invariant Principal Component Analysis (GT-PCA) as an effective and interpretable alternative to PCA and autoencoders. We propose a neural network that efficiently estimates the components and show that GT-PCA significantly outperforms alternative methods in experiments based on synthetic and real data.