Sequence Models course certificate on Coursera
If you’re familiar with the course, you’ll notice that I didn’t really do things in order, again. This course is the last in the series. I have three more to do.
But I thought it was important to do this course now so that I could gain understanding to work on the tools we need at Mediawen which deal with audio processing and, more specifically, voice.
The course taught me how to build and train Recurrent Neural Networks (RNNs), and variants such as LSTMs and GRUs. I also learned how to apply sequence models to natural language problems (Word2vec, Glove) and audio processing such as speech recognition.
On the course, I also learned how to deal with Keras, especially its functional API, because several of the programming assignments required some Keras knowledge. Thankfully, Keras is very well documented and there are lots of examples online.
In fact, I just bought Deep Learning with Python by François Chollet. François works on deep learning at Google in Mountain View and is the creator of Keras, as well as a contributor to the TensorFlow machine-learning framework.
I still have to complete three more courses, as I’ve already completed the first one in this specialization: “Neural Networks and Deep Learning”. I’ll write about it soon. I think I’m close to completing the second one: “Improving Deep Neural Networks”. I’ll share more information about it as soon as I finish it.
Now, I’m eager to start the course on CNNs…
Copyright © 2017-2019 Philippe Anel