Practical Algorithms for Learning Near-Isometric Linear Embeddings

We propose two practical non-convex approaches for learning near-isometric, linear embeddings of finite sets of data points. Given a set of training points , we consider the secant set that consists of all pairwise difference vectors of , normalized to lie on the unit sphere. The problem can be formulated as finding a symmetric and positive semi-definite matrix that preserves the norms of all the vectors in up to a distortion parameter . Motivated by non-negative matrix factorization, we reformulate our problem into a Frobenius norm minimization problem, which is solved by the Alternating Direction Method of Multipliers (ADMM) and develop an algorithm, FroMax. Another method solves for a projection matrix by minimizing the restricted isometry property (RIP) directly over the set of symmetric, postive semi-definite matrices. Applying ADMM and a Moreau decomposition on a proximal mapping, we develop another algorithm, NILE-Pro, for dimensionality reduction. FroMax is shown to converge faster for smaller while NILE-Pro converges faster for larger . Both non-convex approaches are then empirically demonstrated to be more computationally efficient than prior convex approaches for a number of applications in machine learning and signal processing.
View on arXiv