Regularized Frank-Wolfe for Dense CRFs: Generalizing Mean Field and Beyond
This work addresses inference challenges in dense CRFs for tasks like semantic segmentation, offering a unified framework that generalizes existing methods and yields better performance, though it is incremental in nature.
The paper tackles the problem of inference and learning in dense conditional random fields (CRFs) by introducing regularized Frank-Wolfe, a general algorithm that optimizes a nonconvex continuous relaxation. The result shows that several instantiations of this algorithm outperform mean field inference on standard semantic segmentation datasets and produce significant improvements over strong CNN baselines.
We introduce regularized Frank-Wolfe, a general and effective algorithm for inference and learning of dense conditional random fields (CRFs). The algorithm optimizes a nonconvex continuous relaxation of the CRF inference problem using vanilla Frank-Wolfe with approximate updates, which are equivalent to minimizing a regularized energy function. Our proposed method is a generalization of existing algorithms such as mean field or concave-convex procedure. This perspective not only offers a unified analysis of these algorithms, but also allows an easy way of exploring different variants that potentially yield better performance. We illustrate this in our empirical results on standard semantic segmentation datasets, where several instantiations of our regularized Frank-Wolfe outperform mean field inference, both as a standalone component and as an end-to-end trainable layer in a neural network. We also show that dense CRFs, coupled with our new algorithms, produce significant improvements over strong CNN baselines.