VILSS: Discriminative Feature Learning for Large-scale Data

Mengyang Yu, Northumbria University

Computation on large-scale data spaces has been involved in many active problems in computer vision and pattern recognition. However, in realistic applications, most existing algorithms are heavily restricted by the huge number and the high dimension of feature descriptors in data spaces. Generally speaking, there are two main ways to speed up the algorithms: (1) projecting features onto a lower-dimensional subspace; (2) embedding features into a Hamming space. In this talk, I will present our recent work on the dimensionality reduction and the binarization of features for various applications. First, I will show a novel subspace learning algorithm which realizes the discriminant analysis for large-scale local feature descriptors, and a generalized orthogonalization method leading to a more compact and less redundant subspace. Next, local feature based hashing for similarity search will be introduced. Most existing hashing methods for image search and retrieval are based on global representations, e.g., Fisher vectors and VLAD, which lack the analysis of the intrinsic geometric property of local features and heavily limit the effectiveness of the hash code. Finally, I will present how to efficiently reduce very high-dimensional representations to medium-dimensional binary codes with a small memory cost and the low coding complexity.