Overview
Explore the world of unstructured data analysis through language models in this comprehensive conference talk. Delve into the fundamental concepts of language modeling, from its historical roots in code-breaking to modern applications in data science. Learn how language models underpin various technologies and discover their practical applications in classification, predictive modeling, and information retrieval. Gain insights into key statistical and information theory concepts, and understand how techniques like word vectors and transfer learning have advanced the field. Through case studies and real-world examples, examine how language models can be used to identify forgeries, predict demographic information, and extract targeted insights from large datasets. Suitable for data scientists looking to expand their toolkit for handling unstructured data, which comprises the majority of available information in today's digital landscape.
Syllabus
Introduction
Unstructured Data and Language Models
Outline
Location
Enigma Machine
Enigma vs Bitcoin
Probability
Language Models
Unstructured Data
Case Studies
Case Study 1 Marvel
Sentiment Analysis
Case Study
Case Study 3
Example
Bag of Words
Mini Example
Real World Example
Taxing
Interpolation
Perplexity
Continuous Space Models
Geoffrey Hinton Course
Building Word Vectors
Neural Networks
Cost
Transfer Learning
Wrapping Up
Question
Taught by
Open Data Science