Kernel Polytope Faces Pursuit
نویسندگان
چکیده
Polytope Faces Pursuit (PFP) is a greedy algorithm that approximates the sparse solutions recovered by 1 regularised least-squares (Lasso) [4,10] in a similar vein to (Orthogonal) Matching Pursuit (OMP) [16]. The algorithm is based on the geometry of the polar polytope where at each step a basis function is chosen by finding the maximal vertex using a path-following method. The algorithmic complexity is of a similar order to OMP whilst being able to solve problems known to be hard for (O)MP. Matching Pursuit was extended to build kernel-based solutions to machine learning problems, resulting in the sparse regression algorithm, Kernel Matching Pursuit (KMP) [17]. We develop a new algorithm to build sparse kernel-based solutions using PFP, which we call Kernel Polytope Faces Pursuit (KPFP). We show the usefulness of this algorithm by providing a generalisation error bound [7] that takes into account a natural regression loss and experimental results on several benchmark datasets.
منابع مشابه
Group Polytope Faces Pursuit for Recovery of Block-Sparse Signals
Polytope Faces Pursuit is an algorithm that solves the standard sparse recovery problem. In this paper, we consider the case of block structured sparsity, and propose a novel algorithm based on the Polytope Faces Pursuit which incorporates this prior knowledge. The so-called Group Polytope Faces Pursuit is a greedy algorithm that adds one group of dictionary atoms at a time and adopts a path fo...
متن کاملRecovery of Sparse Representations by Polytope Faces Pursuit
We introduce a new greedy algorithm to find approximate sparse representations s of x = As by finding the Basis Pursuit (BP) solution of the linear program min{‖s‖ 1 | x = As}. The proposed algorithm is based on the geometry of the polar polytope P ∗ = {c | à c ≤ 1} where à = [A,−A] and searches for the vertex c ∈ P ∗ which maximizes x c using a path following method. The resulting algorithm is...
متن کاملGeometry and homotopy for l1 sparse representations
We explore the geometry of l1 sparse representations in both the noiseless (Basis Pursuit) and noisy (Basis Pursuit De-Noising) case using a homotopy method. We will see that the concept of the basis vertex c, which has unit inner product with active basis vectors, is a useful geometric concept, both for visualization and for algorithm construction. We derive an explicit homotopy continuation a...
متن کاملChiral extensions of chiral polytopes
polytope Abstract polytope November, 2013 – p. 2 Abstract polytope Abstract polytope −→ combinatorial generalization of convex polytopepolytope Abstract polytope −→ combinatorial generalization of convex polytope November, 2013 – p. 2 Abstract polytope Abstract polytope −→ combinatorial generalization of convex polytopepolytope Abstract polytope −→ combinatorial generalization of convex polytop...
متن کاملOn the number of faces of centrally-symmetric simplicial polytopes
I. Bfirfiny and L. Lovfisz [Acta Math. Acad. Sci. Hung. 40, 323-329 (1982)] showed that a d-dimensional centrally-symmetric simplicial polytope ~ has at least 2 d facets, and conjectured a lower bound for the number f~ of i-dimensional faces o f ~ in terms ofd and the number f0 = 2n of d vertices. Define integers ho . . . . . he by Z f~-1(x 1) d-' = ~ hi xd-'. A. Bj6rner conjectured (uni=O i=O ...
متن کامل