## 计算机代写|深度学习代写deep learning代考|MDA522

2022年12月27日

couryes-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
• Advanced Probability Theory 高等概率论
• Advanced Mathematical Statistics 高等数理统计学
• (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
couryes™为您提供可以保分的包课服务

## 计算机代写|深度学习代写deep learning代考|Evolutionary Strategies with DEAP

Evolutionary strategies are an expansion to evolutionary and genetic methods that add controlling sub-genes or phenotypes called strategies. These strategies are nothing more than an additional vector that controls or influences the mutation operator. This provides ES for more efficient solving of various complex problems including function approximation.

In this next notebook, we are going to explore a function approximation problem we will revisit when we look at evolution with deep learning. To keep things simple, we will look at approximating function parameters of a known continuous polynomial solution. Then we will move on to more complex discontinuous solutions and see how well ES performs.

ES differs from vanilla genetic algorithms in that an individual carries an additional gene sequence or vector called a strategy. Over the course of the evolution, this strategy vector learns to adjust and apply better more fine-tuned mutation to an individual evolution.

As we discovered previously in chapter 3, mutation and mutation rate are like the learning rate in deep learning. Mutation controls the variability of the population during evolution. The higher the mutation rate the more variable and diverse the population. The ability to control and learn this mutation rate over iterations allows for more efficient determination of solutions.

In this next notebook, we are going to set up an ES algorithm to approximate to known solutions. We will see how learning to optimize the mutation over time allows a population to better converge and approximate solutions. Let’s start by opening notebook EDL_4_4_ES.ipynb in Google Colab and running the whole notebook.

1. Evolutionary strategies is an extension to $\mathrm{GA}$ and as such much of the code we need to use DEAP is like what we have seen before. We will look over the key differences focusing on how ES is implemented starting with the hyperparameter definitions. The IND_SIZE value controls the dimensionality of the solved polynomial function or effectively the gene size. The MAX_TIME hyperparameter is for controlling the total amount of time to run the evolution for. An effective way to control how long an evolution runs for instead of relying on number of generations. Lastly, the strategy allocation hyperparameterSMIN_VALUE, MAX_VALUE, MIN_STRATEGY and MAX_STRATEGY control the mutation vector and will be examined further below.

## 计算机代写|深度学习代写deep learning代考|Differential Evolution with DEAP

Deep learning systems are often described as a simply good function or convex approximators. By no means is function approximation limited to deep learning, but it currently ranks as top favorite for most solutions.

Fortunately, EC encompasses several methods not limited to continuous solutions but instead can solve discontinuous solutions as well. One such method focused on function approximation for continuous and discontinuous solutions is Differential Evolution. DE is not calculus based but instead relies on reducing the difference in optimized solutions.

In our next notebook we are going to employ DE to approximate a known continuous polynomial solution, from our last exercise, as well as basic examples of discontinuous and complex functions. This will give us another tool in our EC toolbelt when we look at building combined solutions with DL later.

Differential Evolution has more in common with particle swarm optimization than genetic algorithms or programming. In DE we maintain a population of agents each of some equal vector size. Like PSO agents are long-running and don’t produce offspring but their component vector is modified using difference comparisons from other random agents to produce new and better agents.

Figure $4.11$ shows the basic workflow for DE. At the start of this figure, 3 agents are randomly selected from a pool of agents. These 3 agents are then used to modify a target $\mathbf{Y}$ for each index value in the agent by taking the first agent $\mathbf{a}$ and adding its value to a scaled difference between agents $\mathbf{b}$ and $\mathbf{c}$. The resulting $\mathbf{Y}$ agent is evaluated for fitness and if that value is better then that agent is replaced with the new agent $\mathbf{Y}$.

functions which often need to blend results as in deep learning or generalize results like genetic evolution, DE does a component-wise differentiation.

In deep learning the gradient optimization method, we use to backpropagate errors or differences during training is a global optimization problem. DE extracts the optimization into component wise differentiation of values and is therefore not limited by global methods. This means DE can be used to approximate discontinuous or difficult functions as we will see.
For the next scenario, open notebook EDL_4_5_DE.ipynb in Colab and run all the cells. This example works on the same problem set from the last exercise. As such, we have 3 problems we can run this sample against, a polynomial, the absolute, and step functions. For comparison, we will begin by running the example of the same polynomial function approximation problem we just looked at.

## 计算机代写|深度学习代写deep learning代考|Evolutionary Strategies with DEAP

ES 与普通遗传算法的不同之处在于，个体携带一个额外的基因序列或向量，称为策略。在进化过程中，这个策略向量学会调整和应用更好更微调的突变到个体进化。

1. 进化策略是对G一种因此，我们需要使用 DEAP 的大部分代码就像我们之前看到的一样。我们将从超参数定义开始，重点关注 ES 的实现方式。IND_SIZE 值控制求解的多项式函数的维数或有效地控制基因大小。MAX_TIME 超参数用于控制运行进化的总时间。一种控制进化运行多长时间而不依赖于世代数的有效方法。最后，策略分配超参数 SMIN_VALUE、MAX_VALUE、MIN_STRATEGY 和 MAX_STRATEGY 控制变异向量，将在下面进一步检查。

## 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。