Come on time!!!
Join us as we explore how to get started building models with Keras and then discuss the power of running distributed.
19:00 PM - Using Embedding Layers on the Netflix Prize dataset
Embedding layers are an elegant tool, used to reduce the dimensionality of a complex space in a meaningful manner. A classic example would be word2vec, a family of networks constructed to create short, linguistically meaningful vector representations from text. In this talk, we would follow a relatively simple test-case of constructing a movie recommendation system from the Netflix Prize dataset. We will see that an embedding layer can be built from no more than a sparse list of movie ratings by Netflix users. (The talk would include some live code)
19:30 PM - Distributed TF
The world is talking about Kubernetes so why not try to use it for our TensorFlow model training and predictions.
In this talk, we will discuss the pros and cons of the Distributed mode in TensorFlow, what’s the idea behind it and how we can run it simply on Kubernetes. How we can monitor the services, aggregate all the data from different slaves at one place and get more details about missing links.