arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2303.03027stat.MLcs.LG
收藏

采用Bures-Wasserstein损失训练的生成式深度线性网络的临界点与收敛性分析

Critical Points and Convergence Analysis of Generative Deep Linear Networks Trained with Bures-Wasserstein Loss

Max Planck Institute for Mathematics in the Sciences(马克斯·普朗克数学科学研究所);Duke University(杜克大学)

Pierre Bréchet, Katerina Papagiannouli, Jing An, Guido Montúfar

AI总结:

该研究针对生成式场景下采用Bures-Wasserstein损失的深度矩阵分解模型,刻画了其临界点与极小值点,解决了低秩处Hessian发散的分析难题,建立了梯度流与有限步长梯度下降的收敛性结果。

摘要:

我们研究采用Bures-Wasserstein距离训练的协方差矩阵深度矩阵分解模型。尽管近期研究在超参数化低秩矩阵逼近的优化问题上取得进展,但大多聚焦于判别式场景与平方损失。相比之下,我们的模型采用另一类损失,且与生成式场景相关联。我们刻画了秩有界矩阵空间上Bures-Wasserstein距离的临界点与极小值点。该损失在低秩矩阵处的Hessian矩阵理论上可能发散,给梯度优化方法的收敛性分析带来挑战。我们利用损失的光滑扰动版本建立了梯度流的收敛结果,还在初始权重满足特定假设的条件下,建立了有限步长梯度下降的收敛结果。

英文摘要:

We consider a deep matrix factorization model of covariance matrices trained with the Bures-Wasserstein distance. While recent works have made advances in the study of the optimization problem for overparametrized low-rank matrix approximation, much emphasis has been placed on discriminative settings and the square loss. In contrast, our model considers another type of loss and connects with the generative setting. We characterize the critical points and minimizers of the Bures-Wasserstein distance over the space of rank-bounded matrices. The Hessian of this loss at low-rank matrices can theoretically blow up, which creates challenges to analyze convergence of gradient optimization methods. We establish convergence results for gradient flow using a smooth perturbative version of the loss as well as convergence results for finite step size gradient descent under certain assumptions on the initial weights.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑