Faster Predict-and-Optimize with Three-Operator Splitting
share
In many practical settings, a combinatorial problem must be repeatedly solved with similar, but distinct parameters w. Yet, w is not directly observed; only contextual data d that correlates with w is available. It is tempting to use a neural network to predict w given d, but training such a model requires reconciling the discrete nature of combinatorial optimization with the gradient-based frameworks used to train neural networks. One approach to overcoming this issue is to consider a continuous relaxation of the combinatorial problem. While existing such approaches have shown to be highly effective on small problems (10-100 variables) they do not scale well to large problems. In this work, we show how recent results in operator splitting can be used to design such a system which is easy to train and scales effortlessly to problems with thousands of variables.
READ FULL TEXT
