Posted on Categories:CS代写, Machine Learning, 机器学习, 计算机代写

# 机器学习代考_Machine Learning代考_COMP4702 Image Interpretation Using Pixel Attribution

avatest™

## avatest™帮您通过考试

avatest™的各个学科专家已帮了学生顺利通过达上千场考试。我们保证您快速准时完成各时长和类型的考试，包括in class、take home、online、proctor。写手整理各样的资源来或按照您学校的资料教您，创造模拟试题，提供所有的问题例子，以保证您在真实考试中取得的通过率是85%以上。如果您有即将到来的每周、季考、期中或期末考试，我们都能帮助您！

•最快12小时交付

•200+ 英语母语导师

•70分以下全额退款

## 机器学习代考_Machine Learning代考_ Image Interpretation Using Pixel Attribution

Pixel attribution methods highlight the pixels that were relevant for certain image classification by a neural network. Figure 12-6 is an example of an explanation.

Pixel attribution is a special case of feature attribution for images. Feature attribution explains individual predictions by attributing each input feature according to how it changed the prediction (negatively or positively). The features can be input pixels, tabular data, or words. SHAP, Shapley values, and LIME are examples of general feature attribution methods. We consider neural networks that output as prediction a vector of length, $\mathrm{C}$, which includes regression where $\mathrm{C}=1$. The output of the neural network for image, $\mathrm{I}$, is called $\mathrm{S}(\mathrm{I})=[\mathrm{S} 1(\mathrm{I}), \ldots, \mathrm{SC}(\mathrm{I})]$. All these methods take as input $\mathrm{x} \in \mathrm{Rp}$ (can be image pixels, tabular data, words, …) with $\mathrm{p}$ features and output as explanation a relevance value for each of the $p$ input features: $R c=[R c 1, \ldots, R c p]$. The $c$ indicates the relevance for the c-th output SC(I).

There is a confusing amount of pixel attribution approaches. It helps to understand that there are two different types of attribution methods.

• Occlusion- or perturbation-based: Methods like SHAP and LIME manipulate parts of the image to generate explanations (modelagnostic).
• Gradient-based: Many methods compute the prediction gradient (or classification score) with respect to the input features. The gradientbased methods (of which there are many) mostly differ in how the gradient is computed.

## 机器学习代考_Machine Learning代考_Image Interpretation Using Class Activation Maps

Class activation maps (CAM) were introduced in the paper “Learning Deep Features for Discriminative Localization” using global average pooling in CNNs. A CAM in a particular category indicates the discriminative region used by CNN to identify the category.

It has been observed that convolution units of various layers of a convolutional neural network act as an object detector even though no such prior about the location of the object is provided while training the network for a classification task. Even though convolution has this remarkable property, it is lost when using a fully connected layer for the classification task. To avoid using a fully connected network, some architectures like Network in Network (NiN) and GoogLeNet are fully convolutional neural networks. Global average pooling (GAP) is a commonly used layer in such architectures. It is mainly used as a regularizer to prevent overfitting while training. The class activation map simply indicates the discriminative region in the image, which the $\mathrm{CNN}$ uses to classify that image in a particular category. For this technique, the network consists of ConvNet, and right before the softmax layer (for multiclass classification), global average pooling is performed on the convolutional feature maps. The output of this layer is used as a feature for a fully connected layer that produces the desired classification output. Given this simple connectivity structure, we can identify the importance of the image regions by projecting back the weights of the output layer onto the convolutional feature maps.

## MATLAB代写

MATLAB 是一种用于技术计算的高性能语言。它将计算、可视化和编程集成在一个易于使用的环境中，其中问题和解决方案以熟悉的数学符号表示。典型用途包括：数学和计算算法开发建模、仿真和原型制作数据分析、探索和可视化科学和工程图形应用程序开发，包括图形用户界面构建MATLAB 是一个交互式系统，其基本数据元素是一个不需要维度的数组。这使您可以解决许多技术计算问题，尤其是那些具有矩阵和向量公式的问题，而只需用 C 或 Fortran 等标量非交互式语言编写程序所需的时间的一小部分。MATLAB 名称代表矩阵实验室。MATLAB 最初的编写目的是提供对由 LINPACK 和 EISPACK 项目开发的矩阵软件的轻松访问，这两个项目共同代表了矩阵计算软件的最新技术。MATLAB 经过多年的发展，得到了许多用户的投入。在大学环境中，它是数学、工程和科学入门和高级课程的标准教学工具。在工业领域，MATLAB 是高效研究、开发和分析的首选工具。MATLAB 具有一系列称为工具箱的特定于应用程序的解决方案。对于大多数 MATLAB 用户来说非常重要，工具箱允许您学习应用专业技术。工具箱是 MATLAB 函数（M 文件）的综合集合，可扩展 MATLAB 环境以解决特定类别的问题。可用工具箱的领域包括信号处理、控制系统、神经网络、模糊逻辑、小波、仿真等。