Machine learning 2 -- ...
$30-250 AUD
Pagado a la entrega
Your task is to prepare data reader for the training code and evaluate several training configurations, which are:
1-UNet vs LinkNet vs MobileUNet comparison
2-Increased filter size (from 3x3 to e.g. 7x7)
3-Location augmentation as described here: [login to view URL]
4-SegCaps - 4th architecture to test (as an addition to 3 architectures from #1) which should be taken from here: [login to view URL]
_This means you need to perform at least 6 training (3 for #1 and 1 for each of #2, #3 and #4). Make sure you optimize hyperparams before the actual training.
-Hopt module should help in this, but you can obviously perform the optimizations your own way. If you decide to stick to hopt, you can find some explanations here: [login to view URL]
after you finish, you have to run the code on my laptop and show the comparison between each model which one the best and why .. by create reports and put all the results for each model with graphs . please put comments inside the code to make everything clear
Nº del proyecto: #20052502
Sobre el proyecto
7 freelancers están ofertando un promedio de $144 por este trabajo
hello sir, I have 3 years experienced of machine learning. I will help you in your project and run in your laptop and also show you the comparison of the algorithm. if you have any questions, please come to inbox. tha Más
I have better knowledge about machine learning and data preprocessing and also have certificate of this..i am pursuing my bachelor degree and proficient in python.. please give me one chance
Deal with hyperparameters is what I've done during the whole last year on my master thesis. I'm pretty fresh on the subject. Is your code in python? Just give it to me.
Hello I'm Mohammad, I'm an electronic engineer and I have a wide knowledge in the field of microcontrollers, choose me now so we can start immediately.
Hello, I am a bioinformatician working in a government organization and I have worked in the field of segmentation of X-rays and other DICOM standard images. I have previously used UNet model for a Kaggle competition Más
