پروژه یادگیری ماشین (54413)

---

شرح پروژه

سلام وقتتون بخیر

ببخشید دیر وقت مزاحم میشم
یه زحمتی براتون داشتم
حدود سه تا تمرین ماشین لرنینگ برای دانشگاهم میخواستم برام انجام بدین
خواستم ببینم باید چیکار کنم؟

تمرین اول:
While we discussed "standard" SGD in class, there are a variety of improvements to SGD, including"Accelerated methods" that can perform significantly better in practice.

For this ILA, you will have the opportunity to attempt to implement these as well as monitor/visualize their performance on a subset of test problems.

Specifically, choose two of the SGD Extensions and Variants listed here: https://en.wikipedia.org/wiki/Stochastic_gradient_descent#Extensions_and_variantsLinks to an external site. and implement them in Python along with standard SGD for optimization of a single objective function with two design variables. For example, you might implement "vanilla" SGD, SGD+Momentum, and RMSProp. Or you might choose others, depending on your interests. Produce an animation similar to the below that compares the solution path of "vanilla" SGD with your two chosen variants on two common 2D test functions (the Rosenbrock Function (https://en.wikipedia.org/wiki/Rosenbrock_functionLinks to an external site.) and the Styblinski–Tang function (https://en.wikipedia.org/wiki/Test_functions_for_optimization#Test_functions_for_singleobjective_optimizationLinks to an external site.)).

تمرین دوم:
Implement Policy Gradients with a linear state space function and a gaussian action potential.

The original paper is: https://papers.nips.cc/paper/1713-policy-gradient-methods-for-reinforcement-learning-with-function-approximation.pdfLinks to an external site..

While Policy Gradient methods are covered in detail in Chapter 13 of Sutton and Barto: http://incompleteideas.net/book/RLbook2018.pdfLinks to an external site. with the simplest method being the REINFORCE Algorithm and more advanced ones from there. Any of the policy gradient approaches is sufficient for this ILA. You only need to do linear function approximations here (w*x) and not more complex non-linear ones such as those that use neural networks (though don't let me stop you if you really want to try that out).

Demonstrate the Policy Gradient implementation on the continuous MountainCar problem: https://en.wikipedia.org/wiki/Mountain_car_problem#Technical_detailsLinks to an external site..

This is most readily implemented in the OpenAI Gym Suite, though you can use alternative implementations if you want to: https://gym.openai.com/envs/MountainCarContinuous-v0/Links to an external site..

تمرین سوم:عکس ها

حدودا بین دو تا سه هفته دیگه نیاز دارم
ممنون میشم بررسی کنید که ایا میتونید انجام بدین یا نه و اینکه حدود قیمت رو هم بفرمایید
ممنونم

این پروژه شامل 2 فایل مهم است، لطفا قبل از ارسال پیشنهاد حتما نسبت به بررسی این فایل اقدام فرمایید.

مهارت ها و تخصص های مورد نیاز

---

---