PCA is deep learning

R deep learning libraries


I was wondering if there are any good R libraries out there for deep learning neural networks. I know there are, and, but none of them seem to implement deep learning methods.

I'm particularly interested in unsupervised followed by supervised learning and the use of dropouts to prevent co-adjustment.

/ edit: After a couple of years I found the h20 deep learning package to be very well designed and easy to install. I also like the mxnet package which is (a little) more difficult to install but supports things like Covnets, runs on GPUs and is very fast.






Reply:


OpenSource h2o.deepLearning () is a package for deeplearning in R from h2o.ai here is a description http://www.r-bloggers.com/things-to-try-after-user-part-1-deep-learning - with-h2o /

And code: https://gist.github.com/woobe/3e728e02f6cc03ab86d8#file-link_data-r


There is a package called "darch"

http://cran.um.ac.ir/web/packages/darch/index.html

Quote from CRAN:

darch: Package for deep architectures and restricted Bolzmann machines

The Darch package is based on the code from GE Hinton and RR Salakhutdinov (available under Matlab Code for Deep Believe networks: last visit: 01.08.2013). This package is used to generate neural networks with many layers (deep architectures) and to train them with the method described in the publications "A Fast Learning Algorithm for Deep Believe Networks" (GE Hinton, S. Osindero, YW Teh) and "Reduction of dimensionality" was presented by data with neural networks "(GE Hinton, RR Salakhutdinov). This method includes a preliminary training with the method of contrastive divergence published by GE Hinton (2002) and a fine-tuning with common known training algorithms such as backpropagation or conjugates Gradient.







There is another new package for deep networks in R: deepnet

I haven't tried it yet, but it's already included in the Caret package.


To answer my own question, I wrote a small package in R for RBMs: https://github.com/zachmayer/rbm

This package is still under development and I know very little about RBMs, so I would appreciate your feedback (and pull requests!). You can install the package with devtools:

The code is similar to Andrew Landgraf's implementation in R and Edwin Chen's implementation in Python, but I've written the function to be similar to the pca function in Base R and include functions to stack. I think it's a bit more user friendly than the darch package which I never got to figure out how to use (even before it was removed from CRAN).

If you have installed the gputools package, you can use your GPU for matrix operations with the rbm_gpu function. That speeds things up a lot! Additionally, most of the work in an RBM is done with matrix operations, so installing a good BLAS like openBLAS also speeds things up a lot.

Here's what happens when you run the code in Edwin's sample dataset:








To add another answer:

mxnet is amazing and i love it. It's a little difficult to install, but it supports GPUs and multiple CPUs. If you want to learn more about R (especially with regard to images), I recommend that you start with mxnet.



While I haven't found a specific deep learning library for R, I've encountered similar discussions with R bloggers. The focus of the discussion is the use of RBM (Restricted Boltzman Machines). Take a look at the link below:

http://www.r-bloggers.com/restricted-boltzmann-machines-in-r/ (newly posted by "alandgraf.blogspot.com")

The author did a really good job encapsulating a self implemented algorithm in R. It has to be said that I haven't checked the validity of the code yet, but there is at least a tinge of deep learning looming in R.

I hope it helps.







We use cookies and other tracking technologies to improve your browsing experience on our website, to show you personalized content and targeted ads, to analyze our website traffic, and to understand where our visitors are coming from.

By continuing, you consent to our use of cookies and other tracking technologies and affirm you're at least 16 years old or have consent from a parent or guardian.

You can read details in our Cookie policy and Privacy policy.