What is a Deep Learning Library? – Ep. 16 (Deep Learning SIMPLIFIED)

What is a Deep Learning Library? – Ep. 16 (Deep Learning SIMPLIFIED)

If you’re coding a deep neural network,
using a Deep Learning software library is a sure-fire way to simplify the development
process. Rather than re-invent the wheel, you can take advantage of well-tested code
that was created by experts in the field. Let’s take a closer look. If you aren’t a developer or if you’re
unfamiliar with the term, a library is a premade set of functions and modules that you can
call through your own programs. Libraries are typically created by highly-qualified
software teams, and popular libraries are regularly maintained. Many libraries are open-source,
and are surrounded by big communities that provide support and contribute to the codebase.
Deep Learning has plenty of great libraries available, several of which were created by
key people in the field. Have you ever tried to code your own deep
net? Did you use a library to help simplify the process? Please comment and share your
thoughts. If you’re building a commercial app that
requires the use of a deep net, your best bet is to use a commercial-grade library like
deeplearning4j, Torch, or Caffe. For educational or scientific projects, you should use a library
like Theano. Another notable library is deepmat, and there are many others. In the next video, we’ll start by looking
at the Theano library.

Only registered users can comment.

  1. At times, you just want the end product, but at others, you want to be able influence the way it works. In those cases, you are left with the task of building your own deep net. In those situations, using a library would help you re-use high-quality code instead of re-inventing the wheel. Enjoy 🙂

  2. Can you please suggest some library for the image captioning purpose, that will involve use of CNN and LSTM. I am confused in between Keras and torch? I will use it for the educational purpose.

  3. What about TensorFlow? It was built by Google, on the hand of Geoffrey Hinton and Jeff Dean as a team leaders improving DistBelief (1st gen of TensorFlow) .

  4. Fantastic channel! It would be even better if it also had an episode on Google's TensorFlow. My personal experiences with that library have been very good, but then I have never worked with the others like Theano, Torch and Caffe.

  5. can you please suggest deep learning library for learning neural network… I mean the library that allow me to build neural net from scratch, but still give some useful tool to avoid reinventing the wheel

  6. Fabulous!!! Please do more videos on explaining some example codes in Deep learning. It will be very useful for beginners.

  7. Been watching your channel lately n it has been very useful ! I'm trying to do Stock market prediction using sentiment analysis.. not sure which neural network model to use… Could you please suggest ? Also which tool ( library ) to use ??? Please do reply ! Thank you 🙂

  8. Since you asked me to comment if I had attempted to code my own neural net, yes I have. I did write my own implementation of a digit classifier for MNIST using numpy whilst working through Andrew Ng's ML course on Coursera. I succeeded porting my Octave code from his course to numpy (including GPU use), but I've decided to use tensorflow in the future since getting matrix operations efficiently coded is tedious even with a great library like numpy. Thanks for these videos, I think they are valuable for people getting started in ML.

Leave a Reply

Your email address will not be published. Required fields are marked *