Introduction to Topological Data Analysis

25
TOPOLOGICAL DATA ANALYSIS HENDRI KARISMA

Transcript of Introduction to Topological Data Analysis

Page 1: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

HENDRI KARISMA

Page 2: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

HENDRI KARISMA

▸ Senior Software Development Engineer at blibli.com

▸ Part of Fulfillment Team

[email protected]

Page 3: Introduction to Topological Data Analysis

WHAT WE THINK IF WE HEARD THE WORD “DATA”?

Page 4: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

PROVIDE LARGE DATA

▸ Complicated dataset to analyse

▸ How to make sense of large array of numbers, of column and rows

Page 5: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

???

▸ Analogy a group of data is a node

▸ Each node have relationship and we call it edge

▸ We can see unstructured massive number, we have network and shape. Shape of network representing the shape of our data

▸ Use visual system to look the data and identify the feature, in the network which correspond with the data.

▸ Extracting knowledge of data.

Page 6: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

WHY TDA

▸ Data has shape

▸ shape has meaning

Page 7: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

TDA HISTORY (1)

▸ Topology sub field in math the concern with the study of the shape

▸ The origin in 18 century with swiss mathematician, Leonhard Euler.

▸ Eurler become aware of a challenge problem, concerning the seven bridges crossing the river.

▸ He took all the information about the bridges and the river, and the island, and the land mass. Then converted it all into simple network, and he found that in fact that is not possible to cross the bridge exactly one.

Page 8: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

TDA HISTORY (2)

▸ The last 15 years, there has application to many different real problem.

▸ One of those, the analysis and understanding of high dimensional and complex dataset.

▸ This area study is called topological data analysis, it’s changing the way that people are able to understand and analyze their data

Page 9: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

Page 10: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

COMPLEXITY OF THE DATA

▸ Large data set can be simple in extractor

▸ Small data set can be complex

Page 11: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

THE PROPERTIES OF TDA

▸ Three big concept that give it’s big power for analysing and understanding shape

▸ Coordinate invariance

▸ Deformation Invariance

▸ Compressed Representations

Page 12: Introduction to Topological Data Analysis
Page 13: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

DATA COMES WITH THE WRONG COORDINATE

Page 14: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

DATA COMES WITH THE WRONG COORDINATE

Page 15: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

DATA COMES WITH THE WRONG COORDINATE

Page 16: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

COMPLEX DATA

Page 17: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

TOPOLOGICAL NETWORK CAPTURE SHAPE

Page 18: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

TOPOLOGICAL NETWORK CAPTURE SHAPE

Page 19: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

CAPTURING COMPLEX SHAPE

Page 20: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

EXAMPLE FRAUD

Page 21: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

TOPOLOGY YIELDS PRINCIPLED LOCAL MODELS

Page 22: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

PCA FIND 3 CLUSTER

Page 23: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

THE ACTUAL IS 4

Page 24: Introduction to Topological Data Analysis

TOPOLOGICAL DATA ANALYSIS

SUMMARY

▸ The three properties combine and very striking ways to allow one to analyse and understand very large and complicated data sets

▸ Topological data analysis represent a fundamental advances in machine learning

▸ In the near future machine will help humans organise simplify and understand their very large and complicated data sets

Page 25: Introduction to Topological Data Analysis

THANKS