集团卷积神经网络的VC维度
Philipp Christian Petersen1, Anna Sepliarskaia1
1University of Vienna, Faculty of Mathematics and Research Network Data Science@ Uni Vienna, Kolingasse 14-16, 1090 Wien, Austria.
概括
我们研究了组卷积神经网络的概括能力. 我们的研究结果表明,即使是简单的两参数家族也可以拥有无限的VC维度,从而挑战了关于它们的表示权力的假设.
科学领域:
- 机器学习 机器学习
- 人工智能的人工智能
- 深度学习理论 深度学习理论
背景情况:
- 组卷积神经网络 (GCNNs) 设计用于具有对称性的数据.
- 了解GCNNs的概括能力对于其有效应用至关重要.
研究的目的:
- 准确估计简单的GCNNs集的VC尺寸.
- 分析GCNNs的概括能力,特别是在无限组的背景下.
主要方法:
- 对GCNNs的VC维度进行理论分析.
- 确定特定GCNN架构的精确估计.
主要成果:
- 对于简单的GCNNs集,确定了精确的VC维度估计.
- 具有无限组和特定内核的GCNN的双参数家族表现出无限的VC维度.
- 尽管网络对集团的行动是不变的,但这一点是正确的.
结论:
- 即使对于简单的模型,GCNNs的概括能力也可能令人惊地高.
- 不变性属性并不一定限制GCNN的VC维度.
- 这些发现对对称深度学习模型的设计和理解有影响.
相关概念视频
Convolution Properties I
155
Convolution computations can be simplified by utilizing their inherent properties.
The commutative property reveals that the input and the impulse response of an LTI (Linear Time-Invariant) system can be interchanged without affecting the output:
The commutative property reveals that the input and the impulse response of an LTI (Linear Time-Invariant) system can be interchanged without affecting the output:
155
Convolution Properties II
210
The important convolution properties include width, area, differentiation, and integration properties.
The width property indicates that if the durations of input signals are T1 and T2, then the width of the output response equals the sum of both durations, irrespective of the shapes of the two functions. For instance, convolving two rectangular pulses with durations of 2 seconds and 1 second results in a function with a width of 3 seconds.
The area property asserts that the area under the...
The width property indicates that if the durations of input signals are T1 and T2, then the width of the output response equals the sum of both durations, irrespective of the shapes of the two functions. For instance, convolving two rectangular pulses with durations of 2 seconds and 1 second results in a function with a width of 3 seconds.
The area property asserts that the area under the...
210
Convolution: Math, Graphics, and Discrete Signals
268
In any LTI (Linear Time-Invariant) system, the convolution of two signals is denoted using a convolution operator, assuming all initial conditions are zero. The convolution integral can be divided into two parts: the zero-input or natural response and the zero-state or forced response, with t0 indicating the initial time.
To simplify the convolution integral, it is assumed that both the input signal and impulse response are zero for negative time values. The graphical convolution process...
To simplify the convolution integral, it is assumed that both the input signal and impulse response are zero for negative time values. The graphical convolution process...
268
Dimensionless Groups in Fluid Mechanics
342
Dimensionless groups in fluid mechanics provide simplified ratios that help analyze fluid behavior without relying on specific units. The Reynolds number (Re), which represents the ratio of inertial to viscous forces, distinguishes between laminar and turbulent flows, making it essential in the design of pipelines and aerodynamic surfaces. The Froude number (Fr), the ratio of inertial to gravitational forces, is particularly useful in predicting wave formation and hydraulic jumps in...
342
Deconvolution
165
Deconvolution, also known as inverse filtering, is the process of extracting the impulse response from known input and output signals. This technique is vital in scenarios where the system's characteristics are unknown, and they must be inferred from the observable signals.
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
Deconvolution involves several mathematical techniques to derive the impulse response. One common approach is polynomial division. In this method, the input and output sequences are treated as coefficients of...
165
Collisions in Multiple Dimensions: Introduction
5.4K
It is far more common for collisions to occur in two dimensions; that is, the initial velocity vectors are neither parallel nor antiparallel to each other. Let's see what complications arise from this. The first idea is that momentum is a vector. Like all vectors, it can be expressed as a sum of perpendicular components (usually, though not always, an x-component and a y-component, and a z-component if necessary). Thus, when the statement of conservation of momentum is written for a...
5.4K


