Glassoformer: A Query-Sparse Transformer for Post-Fault Power Grid Voltage Prediction

Here, we propose GLassoformer, a novel and efficient transformer architecture leveraging group Lasso regularization to reduce the number of queries of the standard self-attention mechanism. Due to the sparsified queries, GLassoformer is more computationally efficient than the standard transformers. On the power grid post-fault voltage prediction task, GLasso-former shows remarkably better prediction than many existing benchmark algorithms in terms of accuracy and stability.. Authors: Zheng, Yunling [Univ. of California, Irvine, CA (United States); Purdue University]; Hu, Carson [Univ. of California, Irvine, CA (United States)]; Lin, Guang [Purdue Univ., West Lafayette, IN (United States)]; Yue, Meng [Brookhaven National Lab. (BNL), Upton, NY (United States)]; Wang, Bao [Univ. of Utah, Salt Lake City, UT (United States)]. DOE Contract: SC0021142. Subjects: 24 POWER TRANSMISSION AND DISTRIBUTION; efficient transformer; group lasso; power grid prediction; query sparsity