Upgrade to Pro — share decks privately, control downloads, hide ads and more …

2017 - A Gentle Introduction to Text Classification with Deep Learning

Db2ee812bdc6fd057f8f4209c08b6f63?s=47 PyBay
August 21, 2017

2017 - A Gentle Introduction to Text Classification with Deep Learning


Deep learning has proven very effective for machine learning tasks in the past couple of years, but it is sometimes shrouded in jargon and unnecessary technical detail. This talk will provide a practical introduction to the topic focusing on building an end-to-end text classification system. No machine learning or deep learning experience required. Intermediate knowledge of Python required.


If you wondered what all the hype about deep learning is about but haven't taken the leap to trying it yourself, this talk is for you. First, we'll cover the basics of Deep Learning, focusing on the general intuition of the process. Then get hands-on experience by analyzing common examples of text and training models to predict the category the text belongs to, for example whether a movie review is positive or negative. Along the way, you will learn the necessary machine learning and text processing concepts. We will be using the Kera package which provides a Pythonic way to build deep learning models.


Dr. Brian Spiering is a Data Science Faculty member at GalvanizeU which, in cooperation with the University of New Haven, offers a Master of Science in Data Science. He teaches humans the languages of computers (primarily Python) and teaches computers the languages of humans (through Deep Learning and Natural Language Processing). He is active in the San Francisco tech community through volunteering and mentoring.




August 21, 2017


  1. A Gentle Introduction to Text Classification with Deep Learning Brian

    Spiering http://bit.ly/deep-text
  2. None
  3. Agenda 1. A Gentle Intro to Text Classification and Deep

    Learning (20 Minutes) 2. Demo #1: Learn to classify movie reviews (15 Minutes) 3. Demo #2: Learn to categorize news articles (7.5 Minutes) 4. Questions & Answers (until )
  4. Let's start with foundation: Language

  5. More importantly, humans use language to: - Think - Dream

    - Plan - …
  6. None
  7. Text, digitized language, is the one of the dominant forms

    of communication on the Internet and within organizations
  8. The Internet is just a training set for Deep Learning

  9. Language is hard

  10. Building complex, useful things with computers is hard

  11. None
  12. None
  13. Deep Learning isn't magic, it is just very good at

    finding patterns
  14. Why is Deep Learning taking over Machine Learning?

  15. Neural Networks: Graphs with edges and nodes

  16. A Single Node

  17. Deep Learning: >1 hidden layers

  18. Classification: Discrete Value Prediction

  19. Text Classification: Predict category for a group of words

  20. Neural Network: Given words, weight them to put into categories

  21. Demo #1: Learn to classify movie reviews

  22. Demo #2: Learn to categorize news articles

  23. Summary 1. Text data is everywhere. 2. Deep Learning ❤

    lots of data and has the ability to make sense of it. 3. Keras is approachable way to build Deep Learning systems. 4. Given this start, please continue learning!
  24. Further Study

  25. fast.ai by Jeremy Howard and Rachel Thomas

  26. Deep Learning Specialization from Coursera by Andrew Ng

  27. None
  28. CS224d: Deep Learning for Natural Language Processing from Stanford •

    Course homepage -A complete survey of the field with videos, lecture slides, and sample student projects • Course notes - Probably the best "book" on DL for NLP • Course Lectures
  29. Build Projects

  30. Yesterday's XKCD

  31. None