1: \begin{abstract}

2: This paper studies a sparse signal recovery task in time-varying

3: (time-adaptive) environments. The contribution of the paper to

4: sparsity-aware online learning is threefold; first, a Generalized

5: Thresholding (GT) operator, which relates to both convex and non-convex

6: penalty functions, is introduced. This operator embodies, in a unified way,

7: the majority of well-known thresholding rules which promote

8: sparsity. Second, a non-convexly constrained, sparsity-promoting, online

9: learning scheme, namely the Adaptive Projection-based Generalized

10: Thresholding (APGT), is developed that incorporates the GT operator with a

11: computational complexity that scales linearly to the number of

12: unknowns. Third, the novel family of partially quasi-nonexpansive mappings

13: is introduced as a functional analytic tool for treating the GT

14: operator. By building upon the rich fixed point theory, the previous class

15: of mappings helps us, also, to establish a link between the GT operator and

16: a union of linear subspaces; a non-convex object which lies at the heart of

17: any sparsity promoting technique, batch or online. Based on such a

18: functional analytic framework, a convergence analysis of the APGT is

19: provided. Furthermore, extensive experiments suggest that the APGT exhibits

20: competitive performance when compared to computationally more demanding

21: alternatives, such as the sparsity-promoting Affine Projection Algorithm

22: (APA)- and Recursive Least Squares (RLS)-based techniques. 

23: \end{abstract}