We investigate the efficacy of treating all the parameters in a Bayesian neural network stochastically and find compelling theoretical and empirical evidence that this standard construction may be unnecessary. To this end, we prove that expressive predictive distributions require only small amounts of stochasticity. In particular, partially stochastic networks with only $n$ stochastic biases are universal probabilistic predictors for $n$-dimensional predictive problems. In empirical investigations, we find no systematic benefit of full stochasticity across four different inference modalities and eight datasets; partially stochastic networks can match and sometimes even outperform fully stochastic networks, despite their reduced memory costs.
translated by 谷歌翻译
贝叶斯范式有可能解决深度神经网络的核心问题,如校准和数据效率低差。唉,缩放贝叶斯推理到大量的空间通常需要限制近似。在这项工作中,我们表明它足以通过模型权重的小子集进行推动,以便获得准确的预测后断。另一个权重被保存为点估计。该子网推断框架使我们能够在这些子集上使用表现力,否则难以相容的后近近似。特别是,我们将子网线性化LAPLACE作为一种简单,可扩展的贝叶斯深度学习方法:我们首先使用线性化的拉普拉斯近似来获得所有重量的地图估计,然后在子网上推断出全协方差高斯后面。我们提出了一个子网选择策略,旨在最大限度地保护模型的预测性不确定性。经验上,我们的方法对整个网络的集合和较少的表达后近似进行了比较。
translated by 谷歌翻译
Whilst deep neural networks have shown great empirical success, there is still much work to be done to understand their theoretical properties. In this paper, we study the relationship between random, wide, fully connected, feedforward networks with more than one hidden layer and Gaussian processes with a recursive kernel definition. We show that, under broad conditions, as we make the architecture increasingly wide, the implied random function converges in distribution to a Gaussian process, formalising and extending existing results by Neal (1996) to deep networks. To evaluate convergence rates empirically, we use maximum mean discrepancy. We then compare finite Bayesian deep networks from the literature to Gaussian processes in terms of the key predictive quantities of interest, finding that in some cases the agreement can be very close. We discuss the desirability of Gaussian process behaviour and review non-Gaussian alternative models from the literature. 1
translated by 谷歌翻译
Accurate uncertainty quantification is a major challenge in deep learning, as neural networks can make overconfident errors and assign high confidence predictions to out-of-distribution (OOD) inputs. The most popular approaches to estimate predictive uncertainty in deep learning are methods that combine predictions from multiple neural networks, such as Bayesian neural networks (BNNs) and deep ensembles. However their practicality in real-time, industrial-scale applications are limited due to the high memory and computational cost. Furthermore, ensembles and BNNs do not necessarily fix all the issues with the underlying member networks. In this work, we study principled approaches to improve uncertainty property of a single network, based on a single, deterministic representation. By formalizing the uncertainty quantification as a minimax learning problem, we first identify distance awareness, i.e., the model's ability to quantify the distance of a testing example from the training data, as a necessary condition for a DNN to achieve high-quality (i.e., minimax optimal) uncertainty estimation. We then propose Spectral-normalized Neural Gaussian Process (SNGP), a simple method that improves the distance-awareness ability of modern DNNs with two simple changes: (1) applying spectral normalization to hidden weights to enforce bi-Lipschitz smoothness in representations and (2) replacing the last output layer with a Gaussian process layer. On a suite of vision and language understanding benchmarks, SNGP outperforms other single-model approaches in prediction, calibration and out-of-domain detection. Furthermore, SNGP provides complementary benefits to popular techniques such as deep ensembles and data augmentation, making it a simple and scalable building block for probabilistic deep learning. Code is open-sourced at https://github.com/google/uncertainty-baselines
translated by 谷歌翻译
We propose SWA-Gaussian (SWAG), a simple, scalable, and general purpose approach for uncertainty representation and calibration in deep learning. Stochastic Weight Averaging (SWA), which computes the first moment of stochastic gradient descent (SGD) iterates with a modified learning rate schedule, has recently been shown to improve generalization in deep learning. With SWAG, we fit a Gaussian using the SWA solution as the first moment and a low rank plus diagonal covariance also derived from the SGD iterates, forming an approximate posterior distribution over neural network weights; we then sample from this Gaussian distribution to perform Bayesian model averaging. We empirically find that SWAG approximates the shape of the true posterior, in accordance with results describing the stationary distribution of SGD iterates. Moreover, we demonstrate that SWAG performs well on a wide variety of tasks, including out of sample detection, calibration, and transfer learning, in comparison to many popular alternatives including MC dropout, KFAC Laplace, SGLD, and temperature scaling.
translated by 谷歌翻译
用于估计模型不确定性的线性拉普拉斯方法在贝叶斯深度学习社区中引起了人们的重新关注。该方法提供了可靠的误差线,并接受模型证据的封闭式表达式,从而可以选择模型超参数。在这项工作中,我们检查了这种方法背后的假设,尤其是与模型选择结合在一起。我们表明,这些与一些深度学习的标准工具(构成近似方法和归一化层)相互作用,并为如何更好地适应这种经典方法对现代环境提出建议。我们为我们的建议提供理论支持,并在MLP,经典CNN,具有正常化层,生成性自动编码器和变压器的剩余网络上进行经验验证它们。
translated by 谷歌翻译
现代深度学习方法构成了令人难以置信的强大工具,以解决无数的挑战问题。然而,由于深度学习方法作为黑匣子运作,因此与其预测相关的不确定性往往是挑战量化。贝叶斯统计数据提供了一种形式主义来理解和量化与深度神经网络预测相关的不确定性。本教程概述了相关文献和完整的工具集,用于设计,实施,列车,使用和评估贝叶斯神经网络,即使用贝叶斯方法培训的随机人工神经网络。
translated by 谷歌翻译
我们引入了重新定性,这是一种数据依赖性的重新聚集化,将贝叶斯神经网络(BNN)转化为后部的分布,其KL对BNN对BNN的差异随着层宽度的增长而消失。重新定义图直接作用于参数,其分析简单性补充了宽BNN在功能空间中宽BNN的已知神经网络过程(NNGP)行为。利用重新定性,我们开发了马尔可夫链蒙特卡洛(MCMC)后采样算法,该算法将BNN更快地混合在一起。这与MCMC在高维度上的表现差异很差。对于完全连接和残留网络,我们观察到有效样本量高达50倍。在各个宽度上都取得了改进,并在层宽度的重新培训和标准BNN之间的边缘。
translated by 谷歌翻译
深度神经网络易于对异常值过度自信的预测。贝叶斯神经网络和深度融合都已显示在某种程度上减轻了这个问题。在这项工作中,我们的目标是通过提议预测由高斯混合模型的后续的高斯混合模型来结合这两种方法的益处,该高斯混合模型包括独立培训的深神经网络的LAPPALL近似的加权和。该方法可以与任何一组预先训练的网络一起使用,并且与常规合并相比,只需要小的计算和内存开销。理论上我们验证了我们的方法从训练数据中的培训数据和虚拟化的基本线上的标准不确定量级基准测试中的“远离”的过度控制。
translated by 谷歌翻译
对于神经网络的近似贝叶斯推断被认为是标准培训的强大替代品,通常在分发数据上提供良好的性能。然而,贝叶斯神经网络(BNNS)具有高保真近似推断的全批汉密尔顿蒙特卡罗在协变速下实现了较差的普遍,甚至表现不佳的经典估算。我们解释了这种令人惊讶的结果,展示了贝叶斯模型平均值实际上如何存在于协变量的情况下,特别是在输入特征中的线性依赖性导致缺乏后退的情况下。我们还展示了为什么相同的问题不会影响许多近似推理程序,或古典最大A-Bouthiori(地图)培训。最后,我们提出了改善BNN的鲁棒性的新型前锋,对许多协变量转变来源。
translated by 谷歌翻译
Compared to point estimates calculated by standard neural networks, Bayesian neural networks (BNN) provide probability distributions over the output predictions and model parameters, i.e., the weights. Training the weight distribution of a BNN, however, is more involved due to the intractability of the underlying Bayesian inference problem and thus, requires efficient approximations. In this paper, we propose a novel approach for BNN learning via closed-form Bayesian inference. For this purpose, the calculation of the predictive distribution of the output and the update of the weight distribution are treated as Bayesian filtering and smoothing problems, where the weights are modeled as Gaussian random variables. This allows closed-form expressions for training the network's parameters in a sequential/online fashion without gradient descent. We demonstrate our method on several UCI datasets and compare it to the state of the art.
translated by 谷歌翻译
不确定性估计(UE)技术 - 例如高斯过程(GP),贝叶斯神经网络(BNN),蒙特卡罗辍学(MCDropout) - 旨在通过为每个分配估计的不确定性值来提高机器学习模型的可解释性他们的预测输出。然而,由于过高的不确定性估计可以在实践中具有致命的后果,因此本文分析了上述技术。首先,我们表明GP方法始终会产生高不确定性估计(OOD)数据。其次,我们在2D玩具示例中显示了BNN和MCDRopout在OOD样品上没有提供高不确定性估计。最后,我们凭经验展示了这种BNNS和MCDRopout的陷阱也在现实世界数据集中持有。我们的见解(i)提高了对深度学习中目前流行的UE方法更加谨慎使用的认识,(ii)鼓励开发UE方法,这些方法近似于基于GP的方法 - 而不是BNN和MCDROPOUT,以及我们的经验设置可用于验证任何其他UE方法的ood性能。源代码在https://github.com/epfml/unctemationsiapity-娱乐中获得。
translated by 谷歌翻译
本文研究了用于训练过度参数化制度中的贝叶斯神经网络(BNN)的变异推理(VI),即当神经元的数量趋于无穷大时。更具体地说,我们考虑过度参数化的两层BNN,并指出平均VI训练中的关键问题。这个问题来自于证据(ELBO)的下限分解为两个术语:一个与模型的可能性函数相对应,第二个对应于kullback-leibler(KL)差异(KL)差异。特别是,我们从理论和经验上都表明,只有当根据观测值和神经元之间的比率适当地重新缩放KL时,在过度参数化制度中,这两个术语之间存在权衡。我们还通过数值实验来说明我们的理论结果,这些实验突出了该比率的关键选择。
translated by 谷歌翻译
贝叶斯神经网络和深度集合代表了深入学习中不确定性量化的两种现代范式。然而,这些方法主要因内存低效率问题而争取,因为它们需要比其确定性对应物高出几倍的参数储存。为了解决这个问题,我们使用少量诱导重量增强每层的重量矩阵,从而将不确定性定量突出到这种低尺寸空间中。我们进一步扩展了Matheron的有条件高斯采样规则,以实现快速的重量采样,这使得我们的推理方法能够与合并相比保持合理的运行时间。重要的是,我们的方法在具有完全连接的神经网络和RESNET的预测和不确定性估算任务中实现了竞争性能,同时将参数大小减少到$单辆$ \ LEQ 24.3 \%$的参数大小神经网络。
translated by 谷歌翻译
目前,难以获得贝叶斯方法深入学习的好处,这允许明确的知识规范,准确地捕获模型不确定性。我们呈现先前数据拟合网络(PFN)。 PFN利用大规模机器学习技术来近似一组一组后索。 PFN唯一要求工作的要求是能够从先前分配通过监督的学习任务(或函数)来采样。我们的方法将后近似的目标重新定为具有带有值的输入的监督分类问题:它反复从先前绘制任务(或功能),从中绘制一组数据点及其标签,掩盖其中一个标签并学习基于其余数据点的设定值输入对其进行概率预测。呈现来自新的监督学习任务的一组样本作为输入,PFNS在单个前向传播中对任意其他数据点进行概率预测,从而学习到近似贝叶斯推断。我们展示了PFN可以接近完全模仿高斯过程,并且还可以实现高效的贝叶斯推理对难以处理的问题,与当前方法相比,多个设置中有超过200倍的加速。我们在非常多样化的地区获得强烈的结果,如高斯过程回归,贝叶斯神经网络,小型表格数据集的分类,以及少量图像分类,展示了PFN的一般性。代码和培训的PFN在https://github.com/automl/transformerscandobayesianinference发布。
translated by 谷歌翻译
神经线性模型(NLM)是深度贝叶斯模型,通过从数据中学习特征,然后对这些特征进行贝叶斯线性回归来产生预测的不确定性。尽管他们受欢迎,但很少有作品专注于有条理地评估这些模型的预测性不确定性。在这项工作中,我们证明了NLMS的传统培训程序急剧低估了分发输入的不确定性,因此它们不能在风险敏感的应用中暂时部署。我们确定了这种行为的基本原因,并提出了一种新的培训框架,捕获下游任务的有用预测不确定性。
translated by 谷歌翻译
我们研究了回归中神经网络(NNS)的模型不确定性的方法。为了隔离模型不确定性的效果,我们专注于稀缺训练数据的无噪声环境。我们介绍了关于任何方法都应满足的模型不确定性的五个重要的逃亡者。但是,我们发现,建立的基准通常无法可靠地捕获其中一些逃避者,即使是贝叶斯理论要求的基准。为了解决这个问题,我们介绍了一种新方法来捕获NNS的模型不确定性,我们称之为基于神经优化的模型不确定性(NOMU)。 NOMU的主要思想是设计一个由两个连接的子NN组成的网络体系结构,一个用于模型预测,一个用于模型不确定性,并使用精心设计的损耗函数进行训练。重要的是,我们的设计执行NOMU满足我们的五个Desiderata。由于其模块化体系结构,NOMU可以为任何给定(先前训练)NN提供模型不确定性,如果访问其培训数据。我们在各种回归任务和无嘈杂的贝叶斯优化(BO)中评估NOMU,并具有昂贵的评估。在回归中,NOMU至少和最先进的方法。在BO中,Nomu甚至胜过所有考虑的基准。
translated by 谷歌翻译
随机梯度马尔可夫链蒙特卡洛(SGMCMC)被认为是大型模型(例如贝叶斯神经网络)中贝叶斯推断的金标准。由于从业人员在这些模型中面临速度与准确性权衡,因此变异推理(VI)通常是可取的选择。不幸的是,VI对后部的分解和功能形式做出了有力的假设。在这项工作中,我们提出了一个新的非参数变分近似,该近似没有对后验功能形式进行假设,并允许从业者指定算法应尊重或断裂的确切依赖性。该方法依赖于在修改的能量函数上运行的新的langevin型算法,其中潜在变量的一部分是在马尔可夫链的早期迭代中平均的。这样,统计依赖性可以以受控的方式破裂,从而使链条混合更快。可以以“辍学”方式进一步修改该方案,从而导致更大的可扩展性。我们在CIFAR-10,SVHN和FMNIST上测试RESNET-20的计划。在所有情况下,与SG-MCMC和VI相比,我们都会发现收敛速度和/或最终精度的提高。
translated by 谷歌翻译
本文确定了贝叶斯深度学习中近似MCMC的几个特征。它提出了针对神经网络的近似采样算法。通过类似于从大数据集中采样数据批次,提出了从高维的神经网络参数空间中采样参数子组。尽管在文献中已经讨论了Minibatch MCMC的优势,但阻止的Gibbs抽样在贝叶斯深度学习中受到了研究的关注。
translated by 谷歌翻译
A longstanding goal in deep learning research has been to precisely characterize training and generalization. However, the often complex loss landscapes of neural networks have made a theory of learning dynamics elusive. In this work, we show that for wide neural networks the learning dynamics simplify considerably and that, in the infinite width limit, they are governed by a linear model obtained from the first-order Taylor expansion of the network around its initial parameters. Furthermore, mirroring the correspondence between wide Bayesian neural networks and Gaussian processes, gradient-based training of wide neural networks with a squared loss produces test set predictions drawn from a Gaussian process with a particular compositional kernel. While these theoretical results are only exact in the infinite width limit, we nevertheless find excellent empirical agreement between the predictions of the original network and those of the linearized version even for finite practically-sized networks. This agreement is robust across different architectures, optimization methods, and loss functions.
translated by 谷歌翻译