4
$\begingroup$

I know the concept of a neural network, and I followed the Machine Learning course by Andrew Ng on Coursera, so I also coded some simple ones. However, I miss all the new tricks which are useful to prevent overfitting of Deep networks, such as for example:

  • using ReLU instead than tanh neurons
  • using dropout
  • more advanced learning methods than just vanilla stochastic gradient descent

etc. I would like to follow a MOOC which tells me how to use Keras for Deep Learning (I like Keras very much because IMO is much easier to understand than other packages, but I'm open to suggestions). I would also be content with a book, but I'd really prefer a MOOC. Doesn't have to be free. Can you indicate me one? The application is Data Science, but generic Deep Learning would do.

EDIT: to provide more context to the question, my main applications would be Internet of Things Analytics, i.e., applications on cloud platforms which collect real-time, streaming sensor data from industrial machines and allow to estimate their actual performance, predict the probability of a failure and the time before it happens, detect anomalies, etc. I don't need to develop the cloud platform: I just need to develop the "core" Analytics. Think of it as just applying Deep Learning to Time Series or to Classification problems. However, methods which can easily be retrained when new data arrive, without having to go through the full dataset again, would be preferred.

| improve this question | |
$\endgroup$
  • 1
    $\begingroup$ Keras encapsulates details in a way that makes MOOCs looking at the underlying theory less likely to use it (e.g. you don't implement dropout, just declare it). As a result, you can learn to use it with only a very lightweight guide to the underlying theory. If you have the endurance for a bit more maths than Andrew Ng's course, then coursera.org/learn/neural-networks might be good addition, although it covers some things that are not used in practice in modern NNs, whilst it is 5 years old so missing some recent things like VAEs, GANs, LSTM etc. $\endgroup$ – Neil Slater Jul 2 '17 at 12:41
  • $\begingroup$ @NeilSlater ouch! I can live without GANs, but I would need LSTMs...my applications are mainly time series. I'll add that detail in the question. Isn't there anything more recent? If no MOOC, I could accept a book/lecture slides as an answer, but nothing huge such as Murphy's book. $\endgroup$ – DeltaIV Jul 2 '17 at 12:46
2
$\begingroup$

These 2 are pretty popular:

https://de.udacity.com/course/deep-learning--ud730/ - Tensorflow-Course by Google http://www.fast.ai/ - Deep Learning by a Kaggle hero

The first is easy to follow and nicely presented, the second takes quite some time.

| improve this answer | |
$\endgroup$
3
$\begingroup$

Since you like Keras, the main author has written a book on deep learning with Python. You can check out the sections to see if it covers all the topics you want.

| improve this answer | |
$\endgroup$
  • $\begingroup$ This is excellent! I do like Keras, yes. What do you think of it? As @NeilSlater noted, it allows practitioners like me to use modern Deep Learning without reinventing the wheel. I thought it was quite popular in the Python community. $\endgroup$ – DeltaIV Jul 2 '17 at 17:15
  • 1
    $\begingroup$ I'm a big fan of Keras. I use it all the time. The best resource for me has been the machinelearningmastery series. For example: machinelearningmastery.com/… $\endgroup$ – StatsSorceress Jul 2 '17 at 17:59
  • $\begingroup$ I liked your answer a lot (+1), but I chose the other one because in the question I said I would prefer a MOOC. That doesn't mean I won't come back to your book after the MOOC ;) $\endgroup$ – DeltaIV Jul 3 '17 at 7:19
  • 1
    $\begingroup$ I'm reading the pre-released version of this book. Pretty awesome. $\endgroup$ – piggybox Jul 12 '17 at 1:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.