### 机器学习代写|深度学习project代写deep learning代考|Metrics for Evaluation

statistics-lab™ 为您的留学生涯保驾护航 在代写深度学习deep learning方面已经树立了自己的口碑, 保证靠谱, 高质且原创的统计Statistics代写服务。我们的专家在代写深度学习deep learning代写方面经验极为丰富，各种代写深度学习deep learning相关的作业也就用不着说。

• Statistical Inference 统计推断
• Statistical Computing 统计计算
• (Generalized) Linear Models 广义线性模型
• Statistical Machine Learning 统计机器学习
• Longitudinal Data Analysis 纵向数据分析
• Foundations of Data Science 数据科学基础

## 机器学习代写|深度学习project代写deep learning代考|Following metrics are used for evaluation

Accuracy: It is defined as the percentage of correct predictions made by a classifier compared to the actual value of the label. It can also be defined as the average number of correct tests in all tests [13]. To calculate accuracy, we use the equation:
$$\text { Accuracy }=(T N+T P) /(T N+T P+F N+F P) .$$
Here, TP, TN, FP, and FN mean true positives, false negatives, false positives, and false negatives. True positive is a condition where if the class label of a record in a dataset is positive, the classifier predicts the same for that particular record. Similarly, a true negative is a condition where if the class label of a record in a dataset is negative, the classifier predicts the same for that particular record. False-positive is a condition where the class label of a record in a dataset is negative, but the classifier

predicts the class label as positive. Similarly, a false negative is a condition where the class label of a record in a dataset is positive. Still, the classifier predicts the class label as negative for that record [13].

Sensitivity: It is defined as the percentage of true positives identified by the classifier while testing. To calculate it, we use the equation:
Sensitivity $=(T P) /(T P+F N) .$
Specificity: It is defined as the percentage of true negatives which are rightly identified by the classifier during testing. To calculate it we use the equation:
$$\text { Specificity }=(T N) /(T N+F P) .$$

## 机器学习代写|深度学习project代写deep learning代考|Databases Available

Although several databases of fundus images are available publicly, the creation of quality retinal image databases is still in progress to train deep neural networks.

• DRIVE [14] (Digital Retinal Image for vessel Extraction) – This database contains 40 images collected from 400 samples of age 25 to 90 in the Netherland. Out of 40,7 shows mild DR, whereas others are normal. Each set, i.e., training and testing, includes 20 images of different patients. For every image, manual segmentation known as truths or gold standards of blood vessels is provided.
• STARE [15] (Structured Analysis of Retina) – This database contains 20 retinal fundus images taken using a fundus camera. Datasets are divided into two classes or categories, one contains normal images, and the other includes images with various lesions. CHASE [16] – contains 28 images of $1280 * 960$ pixels, taken from multi-ethnic children in England.
• Messidor and Messidor-2 [17] – These databases contain 1200 and 1748 images of the retina, respectively, taken from both eyes. Messidor- 2 is an extension of the Messidor database taken from 874 samples.
• EyePACS-1 [18] – This database contains macula-centred images of 9963 subjects taken from different cameras in May-October 2015 at EyePACS screening sites.
• APTOS [19] (Asia Pacific Tele-Ophthalmology Society) contains 3662 training images and 1928 testing images. Images are available with the ground truths classified based on severity of DR rating on a scale of 0 to 4 .
• Kaggle [20] – contains 88,702 images of the retina with different resolutions and are classified into 5 DR stages. Many images are of bad quality, and also some of the ground truths have incorrect labelling.
• IDRID [21] (Indian Diabetic Retinopathy Image Dataset) contains 516 retinal fundus images captured by a retinal specialist at an Eye Clinic located in Nanded, Maharashtra, India.
• DIARETDBI [22] contains 89 retinal fundus images of size $1500^{*} 1152$ pixels, including 5 normal images and all other 84 DR images.
• $\quad D D R[23]$ (Diagnosis of Diabetic Retinopathy) – contains 13,637 retinal fundus images showing five stages of DR. From the dataset, 757 images show DR lesions.
• Others like E-ophtha [24], HRF [25], ROC [26], and DR2 [27] etc.

## 机器学习代写|深度学习project代写deep learning代考|Process of Detection of DR Using Deep Learning

There are various numbers of supervised learning methods and unsupervised learning methods available for detecting Diabetic Retinopathy. Deep learning is one technique widely used in medical imaging applications like image classification, image segmentation, image retrieval, image detection, and registration of images. For the detection and classification of diabetic retinopathy, Deep Learning techniques or deep neural networks have been widely used. Deep neural networks produce outstanding results in the removal of default features and isolation. Unlike machine learning methods, the performance of deep learning methods increases with an increase in the number of training datasets because of an increase in learned features. There is a various number of deep neural networks like CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), LSTM (Long Short Term Memory), GRU (Gated Recurrent Unit), Autoencoders, RBM (Restricted Boltzmann Machine), DBN (Deep Belief Network), DSN (Deep Stacking Network), Self-Organizing Maps, etc. Still, CNN has been widely used in medical imaging and is highly effective [11]. Deep networks are much more powerful by using strategies such as the dropout function that helps the network produce relevant results even when few features are missing in the test dataset. In addition, ReLUs (Direct Line Units) function is used as a transfer function in CNNs that helps in effective training as they do not disappear too much like the sigmoid function and tangent function used by standard ANNs. The basic architecture of CNN is that it works in different layers like Convolutional layers, pooling layers, fully connected layers, Dropout, and Activation function at last. In the convolution layer, different types of filters are used to extract features from the image. The subsampling (or pooling) layer acts as feature selection and makes the network potent to changes in size and orientation of the image. Average pooling and max pooling are mostly used in the pooling layer. A fully connected layer is used to define the whole input image. Several pretrained CNN architectures are present at the moment on ImageNet, such as LeNet, AlexNet, VGG, ResNet, GoogleNet, and more.

## 机器学习代写|深度学习project代写deep learning代考|Following metrics are used for evaluation

准确性 =(吨ñ+吨磷)/(吨ñ+吨磷+Fñ+F磷).

特异性 =(吨ñ)/(吨ñ+F磷).

## 机器学习代写|深度学习project代写deep learning代考|Databases Available

• DRIVE [14]（用于血管提取的数字视网膜图像）——该数据库包含从荷兰 25 至 90 岁的 400 个样本中收集的 40 张图像。在 40,7 中显示轻度 DR，而其他则正常。每组，即训练和测试，包括20张不同患者的图像。对于每张图像，都提供了称为血管真相或黄金标准的手动分割。
• STARE [15]（视网膜结构分析）——该数据库包含 20 张使用眼底照相机拍摄的视网膜眼底图像。数据集分为两类或类别，一类包含正常图像，另一类包含具有各种病变的图像。CHASE [16] – 包含 28 张图片1280∗960像素，取自英格兰的多种族儿童。
• Messidor 和 Messidor-2 [17] – 这些数据库分别包含 1200 和 1748 张从双眼拍摄的视网膜图像。Messidor-2 是从 874 个样本中提取的 Messidor 数据库的扩展。
• EyePACS-1 [18] – 该数据库包含 2015 年 5 月至 2015 年 10 月在 EyePACS 筛查站点从不同相机拍摄的 9963 名受试者的黄斑中心图像。
• APTOS [19]（亚太远程眼科学会）包含 3662 张训练图像和 1928 张测试图像。图像提供了根据 DR 等级的严重程度分类的基本事实，范围为 0 到 4。
• Kaggle [20] – 包含 88,702 张不同分辨率的视网膜图像，分为 5 个 DR 阶段。许多图像质量很差，而且一些基本事实的标签也不正确。
• IDRID [21]（印度糖尿病视网膜病变图像数据集）包含 516 张视网膜眼底图像，由位于印度马哈拉施特拉邦南德的一家眼科诊所的视网膜专家拍摄。
• DIARETDBI [22] 包含 89 个大小的视网膜眼底图像1500∗1152像素，包括 5 个正常图像和所有其他 84 个 DR 图像。
• DDR[23]（糖尿病视网膜病变的诊断）——包含 13,637 张视网膜眼底图像，显示 DR 的五个阶段。从数据集中，757 张图像显示 DR 病变。
• 其他如 E-ophtha [24]、HRF [25]、ROC [26] 和 DR2 [27] 等。

## 有限元方法代写

tatistics-lab作为专业的留学生服务机构，多年来已为美国、英国、加拿大、澳洲等留学热门地的学生提供专业的学术服务，包括但不限于Essay代写，Assignment代写，Dissertation代写，Report代写，小组作业代写，Proposal代写，Paper代写，Presentation代写，计算机作业代写，论文修改和润色，网课代做，exam代考等等。写作范围涵盖高中，本科，研究生等海外留学全阶段，辐射金融，经济学，会计学，审计学，管理学等全球99%专业科目。写作团队既有专业英语母语作者，也有海外名校硕博留学生，每位写作老师都拥有过硬的语言能力，专业的学科背景和学术写作经验。我们承诺100%原创，100%专业，100%准时，100%满意。

## 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。