arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2304.03385cs.LGcs.NEmath.PR
收藏

宽神经网络:从初始化时的非高斯随机场到训练的 NTK 几何

Wide neural networks: From non-gaussian random fields at initialization to the NTK geometry of training

Luís Carvalho, João Lopes Costa, José Mourão, Gonçalo Oliveira

AI总结:

本文研究了宽神经网络在有限宽度下的行为,计算了初始化时的非高斯性修正,控制了训练期间输出的演化并改进了 NTK 的衰减率估计,同时估计了非高斯性偏差随训练的演化。

摘要:

近期参数量超过 $n=10^{14}$ 的人工神经网络应用发展,使得研究此类网络的大 $n$ 行为变得极为重要。大多数研究宽神经网络的工作聚焦于此类网络的无限宽度 $n \to +\infty$ 极限,并表明在初始化时,它们对应于高斯过程。在本工作中,我们将研究它们在大但有限 $n$ 下的行为。我们的主要贡献如下:(1) 计算以 $n^{-\frac{1}{2}}$ 渐进级数表示的非高斯性修正。该展开中的系数由参数初始化的统计特性和激活函数决定。(2) 通过计算与极限无限宽度情况(此时网络通过线性流演化)的偏差,控制有限宽度 $n$ 网络在训练期间输出的演化。这改进了先前的估计,并得出了(有限宽度)NTK 在 $n$ 方向上更尖锐的衰减率,且在整个训练过程中均有效。作为推论,我们还证明,以任意高的概率,足够宽的神经网络的训练会收敛到相应二次损失函数的全局最小值。(3) 估计非高斯性偏差如何随训练以 $n$ 表达式演化。具体而言,使用测度空间中的某个度量,我们发现沿着训练,所得测度与对应于无限宽度网络的时变高斯过程(该过程显式地由将初始高斯过程与无限宽度极限下对应训练的线性流进行复合给出)之间的距离在 $n^{-\frac{1}{2}}(\log n)^{1+}$ 范围内。

英文摘要:

Recent developments in applications of artificial neural networks with over $n=10^{14}$ parameters make it extremely important to study the large $n$ behaviour of such networks. Most works studying wide neural networks have focused on the infinite width $n \to +\infty$ limit of such networks and have shown that, at initialization, they correspond to Gaussian processes. In this work we will study their behavior for large, but finite $n$. Our main contributions are the following: (1) The computation of the corrections to Gaussianity in terms of an asymptotic series in $n^{-\frac{1}{2}}$. The coefficients in this expansion are determined by the statistics of parameter initialization and by the activation function. (2) Controlling the evolution of the outputs of finite width $n$ networks, during training, by computing deviations from the limiting infinite width case (in which the network evolves through a linear flow). This improves previous estimates and yields sharper decay rates for the (finite width) NTK in terms of $n$, valid during the entire training procedure. As a corollary, we also prove that, with arbitrarily high probability, the training of sufficiently wide neural networks converges to a global minimum of the corresponding quadratic loss function. (3) Estimating how the deviations from Gaussianity evolve with training in terms of $n$. In particular, using a certain metric in the space of measures we find that, along training, the resulting measure is within $n^{-\frac{1}{2}}(\log n)^{1+}$ of the time dependent Gaussian process corresponding to the infinite width network (which is explicitly given by precomposing the initial Gaussian process with the linear flow corresponding to training in the infinite width limit).

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。

↑