arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2008.11245stat.MLcs.LGmath.OC
收藏

深度网络与多重流形问题

Deep Networks and the Multiple Manifold Problem

Columbia University(哥伦比亚大学);Harvard University(哈佛大学)

Sam Buchanan, Dar Gilboa, John Wright

摘要:

我们研究了多流形问题,这是一个基于机器视觉应用的二进制分类任务,其中训练了一个深度全连接神经网络来分离单位球面的两个低维子流形。我们提供了一个分析的一维情况下,证明了一个简单的流形配置,当网络深度$L$是大的相对于某些几何和统计属性的数据,网络宽度$n$增长作为一个足够大的多项式在$L$,和i.i.d.流形样本在$L$中是多项式的,随机初始化梯度下降快速学习以高概率完美地分类两个流形。我们的分析在实际模型问题的背景下展示了深度和宽度的具体好处:深度作为拟合资源,较大的深度对应于更平滑的网络,可以更容易地分离类流形,宽度作为统计资源,使随机初始化的网络及其梯度集中。讨论的中心是神经切核及其在训练过参数化神经网络的非渐近分析中的作用;对于这篇文献,我们贡献了深度全连接网络的神经切核的本质上最优的集中率,需要宽度$n \gtrsim L\,\mathrm{poly}(d_0)$以在单位球$\mathbb{S}^{n_0-1}$的$d_0$维子流形上实现初始核的均匀集中,和一个非渐近框架,用于建立在NTK制度与结构化数据训练的网络泛化。证明大量使用鞅集中度来最佳地处理初始随机网络各层之间的统计依赖性。这种方法应该是用于建立其他网络架构的类似结果。

英文摘要:

We study the multiple manifold problem, a binary classification task modeled on applications in machine vision, in which a deep fully-connected neural network is trained to separate two low-dimensional submanifolds of the unit sphere. We provide an analysis of the one-dimensional case, proving for a simple manifold configuration that when the network depth $L$ is large relative to certain geometric and statistical properties of the data, the network width $n$ grows as a sufficiently large polynomial in $L$, and the number of i.i.d. samples from the manifolds is polynomial in $L$, randomly-initialized gradient descent rapidly learns to classify the two manifolds perfectly with high probability. Our analysis demonstrates concrete benefits of depth and width in the context of a practically-motivated model problem: the depth acts as a fitting resource, with larger depths corresponding to smoother networks that can more readily separate the class manifolds, and the width acts as a statistical resource, enabling concentration of the randomly-initialized network and its gradients. The argument centers around the neural tangent kernel and its role in the nonasymptotic analysis of training overparameterized neural networks; to this literature, we contribute essentially optimal rates of concentration for the neural tangent kernel of deep fully-connected networks, requiring width $n \gtrsim L\,\mathrm{poly}(d_0)$ to achieve uniform concentration of the initial kernel over a $d_0$-dimensional submanifold of the unit sphere $\mathbb{S}^{n_0-1}$, and a nonasymptotic framework for establishing generalization of networks trained in the NTK regime with structured data. The proof makes heavy use of martingale concentration to optimally treat statistical dependencies across layers of the initial random network. This approach should be of use in establishing similar results for other network architectures.

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。

↑