I have a question regarding batche gardient descent, stochastic gradient descent and mini batches gradient descent
3 visualizaciones (últimos 30 días)
Mostrar comentarios más antiguos
I have a non-linear equation, with 8 variables. the goal obviously is to to find the right variable values. My data set is a vector sized (1,124).
To optimize, I apply the gradient descent (so far I do not know with one is suitable for my case).
To do so, I avoided the non-linearity by applying the taylor expansion to my cost function. Indeed it gave me a linear equation.
Still, I do not know with gradient descent to apply.
my code is devided as follows:
- The first .m (MiniBatches_SGD_Trial) file to calculate the cost function, its derivatives and sum up everything (in case of batch gradient descent).
- The second .m (MiniBatches_SGD_Trial2) file is to run the gradient descent algorithm.
The unusual things are:
- My cose function does not decrease
- I found a hard time to choose the initial guess
I update my two .m files.
Please could you assist me at what exactly is relatively wrong with my code.
Thank you very much in advance
0 comentarios
Respuestas (0)
Ver también
Categorías
Más información sobre Biological and Health Sciences en Help Center y File Exchange.
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!