arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.06559stat.MLcs.LG
收藏

基于随机微分方程的深度贝叶斯神经网络

Infinitely Deep Bayesian Neural Networks with Stochastic Differential Equations

University of Toronto(多伦多大学);Stanford Unviersity(斯坦福大学)

Winnie Xu, Ricky T. Q. Chen, Xuechen Li, David Duvenaud

摘要:

我们在连续深度贝叶斯神经网络中进行可扩展的近似推理。在这个模型类中,每个层中单独权重的不确定性给出了遵循随机微分方程的隐藏单元。我们证明了基于梯度的随机变分推断在这个无限参数设置,产生任意灵活的近似后验。我们还得到了一个新的梯度估计,接近零方差的近似后验权重接近真正的后验。这种方法将连续深度贝叶斯神经网络与离散深度替代方案进行了竞争性比较,同时继承了神经ODE的内存高效训练和可调精度。

英文摘要:

We perform scalable approximate inference in continuous-depth Bayesian neural networks. In this model class, uncertainty about separate weights in each layer gives hidden units that follow a stochastic differential equation. We demonstrate gradient-based stochastic variational inference in this infinite-parameter setting, producing arbitrarily-flexible approximate posteriors. We also derive a novel gradient estimator that approaches zero variance as the approximate posterior over weights approaches the true posterior. This approach brings continuous-depth Bayesian neural nets to a competitive comparison against discrete-depth alternatives, while inheriting the memory-efficient training and tunable precision of Neural ODEs.

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。

↑