TDA and Machine Learning: Better Together
Ayasdi, White Paper.
The New Data Analytics Dilemma
Now more than ever, organizations rely upon their data to make informed decisions that can affect millions of lives and billions of dollars of revenue. The collection and analysis of data from transactions, sensors, and biometrics continues to grow at a prodigious rate, taxing the analytic capabilities of even the most sophisticated organizations.
The quantity of possible insights in a given dataset is an exponential function of the number of data points. On top of this, aggregate data growth is an exponential function with time. Unfortunately, we cannot train enough data scientists to meet this runaway, double-exponential demand curve.
This is driving scientists and mathematicians to examine new approaches to improve both the quality and the speed of their analytics engines. Today’s hypothesis-driven analytics will not suffice. High-performance machines and algorithms can examine complex data far faster and seek insights more comprehensively than ever before. However, we need to find exponential improvements in analysis techniques to meet the growing demand.
Introducing Topology and Topological Data Analysis
Topology is a mathematical discipline that studies shape. TDA refers to the adaptation of this discipline to analyzing highly complex data. It draws on the philosophy that all data has an underlying shape and that shape has meaning. Ayasdi’s approach to TDA draws on a broad range of machine learning, statistical, and geometric algorithms. The analysis creates a summary or compressed representation of all of the data points to help rapidly uncover critical patterns and relationships in data. By identifying the geometric relationships that exist between data points, Ayasdi’s approach to TDA offers an extremely simple way of interrogating data to understand the underlying properties that characterize the segments and sub-segments that lie within data.
Sponsored by Ayasdi