Discriminative Embeddings of Latent Variable Models for Structured Data

Kernel classifiers and regressors designed for structured data, such as sequences, trees and graphs, have significantly advanced in a number of interdisciplinary areas such as computational biology and drug design. Typically, kernel functions are designed beforehand for a data type which either exploit statistics of the structures or make use of probabilistic generative models, and then a discriminative classifier is learned based on the kernels via convex optimization. However, such an elegant two-stage approach also limited kernel methods from scaling up to millions of data points, and exploiting discriminative information to learn feature representations. We propose an effective and scalable approach for structured data representation which is based on the idea of embedding latent variable models into feature spaces, and learning such feature spaces using discriminative information. Furthermore, our feature learning algorithm runs a sequence of function mappings in a way similar to graphical model inference procedures, such as mean field and belief propagation. In real world applications involving sequences and graphs, we showed that the proposed approach is much more scalable than alternatives while at the same time produce comparable results to the state-of-the-art in terms of classification and regression.