Proximal Methods for Hierarchical Sparse Coding

Sparse coding consists in representing signals as sparse linear combinations of atoms selected from a dictionary. We consider an extension of this framework where the atoms are further assumed to be embedded in a tree. This is achieved using a recently introduced tree-structured sparse regularization norm, which has proven useful in several applications. This norm leads to regularized problems that are difficult to optimize, and we propose in this paper efficient algorithms for solving them. More precisely, we show that the proximal operator associated with this norm is computable exactly via a dual approach that can be viewed as the composition of elementary proximal operators. Our procedure has a complexity linear, or close to linear, in the number of atoms, and allows the use of accelerated gradient techniques to solve the tree-structured sparse approximation problem at the same computational cost as traditional ones using the L1-norm. Our method is efficient and scales gracefully to millions of variables, which we illustrate in two types of applications: first, we consider fixed hierarchical dictionaries of wavelets to denoise natural images. Then, we apply our optimization tools in the context of dictionary learning, where learned dictionary elements naturally organize in a prespecified arborescent structure, leading to a better performance in reconstruction of natural image patches. When applied to text documents, our method learns hierarchies of topics, thus providing a competitive alternative to probabilistic topic models.

💡 Research Summary

The paper tackles the problem of hierarchical sparse coding, where the atoms of a dictionary are organized in a tree‑like structure and the sparsity pattern must respect this hierarchy. Traditional sparse coding relies on an ℓ1‑norm regularizer that promotes element‑wise sparsity but ignores any relationships among atoms. In many practical settings—wavelet transforms, hierarchical feature extraction, topic modeling—such relationships are intrinsic, and enforcing them can lead to more interpretable and accurate representations.

To encode hierarchy, the authors adopt a recently introduced tree‑structured sparsity norm:
\