逻辑学习差异卷积神经网络
Magombe Yasin1, Mehmet Sarıgül2, Mutlu Avci3
1Islamic University in Uganda, Kumi Road, P.O. BOX 2555, Mbale, 256, Eastern, Uganda.
概括
对数式学习集成增强差异卷积神经网络 (CNN) 以更快的图像分类. 这种新的方法提高了准确性,并大大减少了训练时间,克服了以前的计算成本缺点.
科学领域:
- 计算机视觉 计算机视觉
- 机器学习 机器学习
- 深度学习 (Deep Learning) 是一种深度学习.
背景情况:
- 卷积神经网络 (CNN) 在图像分类中至关重要,但可能会产生高的计算成本.
- 不同的CNN提供了性能改进,但仍然面临计算挑战.
研究的目的:
- 将对数式学习引入差分CNN,以提高性能和减少计算开销.
- 开发一种新的逻辑差异卷积神经网络 (LDiffCNN),以实现更有效的图像分类.
主要方法:
- 将LogRelu激活函数集成到CNN和差异CNN中.
- 开发一个对数成本函数和一个独特的对数学习方法.
- 使用各种数据集和优化器 (SGD/Adam) 的评估.
主要成果:
- 通过LogRelu集成,CNN和差异CNN的性能提高了1.61%5.44%.
- 使用LogRelu的ResNet架构显示了增强的top-1准确性 (3.07%9.96%).
- 该LDiffCNN比标准CNN高达3.02%的精度,并将训练代减少了38%.
结论:
- 对数式学习集成有效地解决了差分CNN的计算成本缺点.
- 拟议的LDiffCNN展示了卓越的性能,更快的融合和更短的培训时间.
- 该研究验证了对图像分类深度学习中的对数方法的效率和好处.
相关概念视频
Convolution: Math, Graphics, and Discrete Signals
262
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...
262
Convolution Properties II
203
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...
203
Linear Approximation in Frequency Domain
91
Linear systems are characterized by two main properties: superposition and homogeneity. Superposition allows the response to multiple inputs to be the sum of the responses to each individual input. Homogeneity ensures that scaling an input by a scalar results in the response being scaled by the same scalar.
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....
91
Convolution Properties I
152
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:
152
Gain
181
Gain and phase shift are properties of linear circuits that describe the effect a circuit has on a sinusoidal input voltage or current. The circuit's behavior that contains reactive elements will depend on the frequency of the input sinusoid. As a result, it is observed that the gain and phase shift will all be frequency functions.
Gain:
Suppose Vin is the input and Vout is the output signal to a circuit.
Gain:
Suppose Vin is the input and Vout is the output signal to a circuit.
181
Bode Plots
663
Bode plots are graphical tools that use logarithmic scales for frequency on the x-axis and gain in decibels on the y-axis. This logarithmic method allows a wide range of frequencies to be compactly displayed, enabling the analysis of component effects on circuit behavior across a broad frequency spectrum.
A network function represents the ratio of a system's output to its input, with the magnitude and phase angle derived from the complex network function. The decibel logarithmic gain is...
A network function represents the ratio of a system's output to its input, with the magnitude and phase angle derived from the complex network function. The decibel logarithmic gain is...
663


