Best wrapper library for TensorFlow and Theano.
September 17, 2018

Best wrapper library for TensorFlow and Theano.

Gaurav Yadav | TrustRadius Reviewer
Score 9 out of 10
Vetted Review
Verified User

Overall Satisfaction with Keras

Keras is being used during hackathon in my current company. And it's not used by across the company. Basically, during hackathon lots of people are working on machine learning projects that includes deep learning as well. So, there are lots of people who are using Keras for neural network implementation. And I have used this in my during my college and in company as well. We have used Keras to implement neural network for image recognition and in other things as well.
  • One of the reason to use Keras is that it is easy to use. Implementing neural network is very easy in this, with just one line of code we can add one layer in the neural network with all it's configurations.
  • It provides lot of inbuilt thing like cov2d, conv2D, maxPooling layers. So it makes fast development as you don't need to write everything on your own. It comes with lot of data processing libraries in it like one hot encoder which also makes your development easy and fast.
  • It also provides functionality to develop models on mobile device.
  • As Keras works at a high level of abstraction, it limits the user to use it's own implemented algorithm. It doesn't give complete power to user to modify or implementing their own basic algorithm.
  • Sometimes it is slow on GPU as compared to the pure TensorFlow.
  • Other than the above two cons, I don't think it has any negatives.
  • It made our development faster and easy as well.
  • Sometime, when we need to change the basic algorithm, when we need to configure the neural network configuration then it doesn't allow us to modify that.
  • As it comes with lot of inbuilt features of data processing, it is easy to process the data.
TensorFlow and Caffe are bit hard to learn but they give you power to implement everything by you own. But most of the time it is not required to implement our own algorithm, we can solve the problem with just using the already provided algorithms. As compared to TensorFlow and Caffe, Keras is very easy to use and develop things.
I would recommend it for use when anyone wants to quickly develop a neural network. Or if a user is solving any machine learning problem that includes deep learning. And this kind of problem will be like image recognition, face recognition, doing some text analysis using deep learning which includes LSTM or some other algorithm.