Improved graph-based SFA: Information preservation complements the slowness principle
This work addresses a specific bottleneck in supervised feature learning for tasks like age estimation, offering an incremental improvement over existing methods.
The paper tackled the problem of premature information loss in hierarchical graph-based slow feature analysis (HGSFA) networks, which leads to suboptimal feature extraction, and proposed an extension called hierarchical information-preserving GSFA (HiGSFA) that improves performance. On the MORPH-II database for age estimation from facial photos, HiGSFA achieved a mean absolute error of 3.50 years, setting a new state-of-the-art result.
Slow feature analysis (SFA) is an unsupervised-learning algorithm that extracts slowly varying features from a multi-dimensional time series. A supervised extension to SFA for classification and regression is graph-based SFA (GSFA). GSFA is based on the preservation of similarities, which are specified by a graph structure derived from the labels. It has been shown that hierarchical GSFA (HGSFA) allows learning from images and other high-dimensional data. The feature space spanned by HGSFA is complex due to the composition of the nonlinearities of the nodes in the network. However, we show that the network discards useful information prematurely before it reaches higher nodes, resulting in suboptimal global slowness and an under-exploited feature space. To counteract these problems, we propose an extension called hierarchical information-preserving GSFA (HiGSFA), where information preservation complements the slowness-maximization goal. We build a 10-layer HiGSFA network to estimate human age from facial photographs of the MORPH-II database, achieving a mean absolute error of 3.50 years, improving the state-of-the-art performance. HiGSFA and HGSFA support multiple-labels and offer a rich feature space, feed-forward training, and linear complexity in the number of samples and dimensions. Furthermore, HiGSFA outperforms HGSFA in terms of feature slowness, estimation accuracy and input reconstruction, giving rise to a promising hierarchical supervised-learning approach.