Topological data analysis: New perspectives on machine learning from abstract mathematics
When looked at from the right perspective, many of the ideas and algorithms involved in machine learning/data science can be thought of as discovering geometric patterns and shapes in a collection of data points. I will explain how this perspective applies to some of the basic algorithms, then explore some ways in which mathematicians are using concepts from higher-dimensional abstract geometry and topology to better understand the structure of data.
About Jesse Johnson:
Before starting at Google, Jesse was a math professor studying abstract geometry and topology. He became interested in data after learning that many problems in machine learning can be posed in terms of geometry. In addition to developing a number of new algorithms by translating existing techniques from abstract geometry, Jesse began writing the Shape of Data blog to explore how geometric thinking can make the ideas behind machine learning/data mining accessible to non-experts.
Check out his blog here: https://shapeofdata.wordpress.com/