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 2202.11481math.OC
收藏

具有ReLU激活函数的人工神经网络非全局局部极小值的无穷多实现函数的存在性

On the existence of infinitely many realization functions of non-global local minima for artificial neural networks with ReLU activation

University of Münster(明斯特大学);The Chinese University of Hong Kong, Shenzhen (CUHK-Shenzhen)(香港中文大学(深圳))

Shokhrukh Ibragimov, Arnulf Jentzen, Timo Kröger, Adrian Riekert

AI总结:

该研究针对带ReLU激活的浅层人工神经网络,证明了非全局局部极小值的无穷多实现函数存在性,同时给出了特定条件下临界点实现函数数量有限的结论。

摘要:

梯度下降(GD)类优化方案是训练带修正线性单元(ReLU)激活函数的全连接前馈人工神经网络(ANN)的标准工具,可视为梯度流(GF)微分方程解的时间离散化形式。已知对于浅层ReLU ANN,在适当假设下,每条有界GF轨迹的风险都会收敛到临界点的风险,即对应梯度函数的零点。这促使研究GF和GD收敛分析中临界点的充要条件及其对问题参数(如目标函数)的依赖性。第一个主要结果证明,对于浅层ReLU ANN,对任意维度$d \in \mathbb{N}$、任意$a \in \mathbb{R}$、$b \in (a, \infty)$及任意正$\delta \in (0, \infty)$,存在Lipschitz连续目标函数$f \colon [a, b]^d \to \mathbb{R}$,使得对任意隐藏层宽度$\mathfrak{h} \in \mathbb{N} \cap (1, \infty)$,存在不可数多个非全局局部极小值点的不同实现函数,其风险超出风险函数下确界的幅度超过$\delta$。第二个主要结果表明,对于每层仅含一个神经元且目标函数为连续分段多项式的浅层ReLU ANN,临界点的不同实现函数至多有有限个。

英文摘要:

Gradient descent (GD) type optimization schemes are standard tools for training fully connected feedforward artificial neural networks (ANNs) with rectified linear unit (ReLU) activation and can be considered as temporal discretizations of solutions of gradient flow (GF) differential equations. It is known that, for shallow ReLU ANNs, the risk of every bounded GF trajectory converges under suitable assumptions to the risk of a critical point, a zero of the corresponding gradient function. This motivates studying necessary and sufficient conditions in GF and GD convergence analysis for critical points and their dependence on problem parameters, such as the target function. In the first main result, we prove for shallow ReLU ANNs that for every dimension $d \in \mathbb{N}$, every $a \in \mathbb{R}$, $b \in (a, \infty)$, and every positive $δ\in (0, \infty)$, there exists a Lipschitz continuous target function $f \colon [a, b]^d \to \mathbb{R}$ such that, for every hidden-layer width $\mathfrak{h} \in \mathbb{N} \cap (1, \infty)$, there are uncountably many different realization functions of non-global local minimum points whose risks exceed the infimum of the risk function by more than $δ$. In the second main result, we show that, for shallow ReLU ANNs with only one neuron on each layer and a continuous piecewise polynomial target function, there exist at most finitely many different realization functions of critical points.

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。

↑