How TensorFlow Can Help to Perform Natural Language Processing Tasks

by Sophia TurolOctober 12, 2016

With TensorFlow APIs, one is able to accomplish such natural language processing tasks as word embedding, part-of-speech tagging, translation, etc.

tensorflow-meetup-in-silicon-valley-january-2016-v11

At the recent TensorFlow meetup in Silicon Valley, one of the speakers showed some demos that depict natural language processing tasks in action. In addition, the attendees learnt how TensorFlow can be utilized to carry out such tasks.

Table of Contents

Natural language processing with TensorFlow

Junling Hu, a founder and CEO at Question.ai, focused her session on applying TensorFlow to natural language processing (NLP). She enumerated a number of NLP tasks that can be accomplished using TensorFlow APIs:

word embedding
language modeling
part-of-speech tagging
name entity detection
parsing (e.g., using the SyntaxNet framework)
translation

Junling also quickly overviewed recurrent neural networks, mentioning the seq2seq library in TensorFlow for building sequence-to-sequence models, and showed some demos of the NLP tasks in action.

Below is Junling’s presentation from the meetup.

One-click solution for distributed TensorFlow

Ivan Zone of Suchflex talked about the challenges of performing computing tasks on a personal laptop or using cloud services. While the first scenario is time consuming and requires a specialized hardware, the second one may appear expensive or cumbersome. He demonstrated Suchflex’s one-click solution that aims at giving a helping hand to machine learning enthusiasts and gamers.

You can find Ivan’s presentation below.

Want details? Watch the video!

Join our group to stay tuned with the upcoming events!

About the experts

Junling Hu is a leading expert in artificial intelligence and data science and a founder of Question.ai. She was Director of Data Mining at Samsung, building large-scale data-mining solutions for businesses. She has also led data science teams in PayPal and eBay, creating enterprise solutions based on recommender systems, text mining, and predictive modeling.

Ivan Zone is a founder and CEO at SuchFlex. He is a former Quant developer and trader at a Manhattan-based $4B Quant Hedge Fund. Ivan is familiar with building and trading in digital markets. He is passionate about developing a peer-to-peer computing market to make the benefits of distributed computing accessible to everyone.