Computational Complexity Workshop

In this mini-workshop we will explore how we can measure complexity computationally, in particular when it comes to time series and biosignals.

Water close-up

About the Workshop

We will explore the concept of complexity from definitions and mathematical foundations to computationally practical approximations applicable to research at RITMO. We will focus on discussing applications to biosignals, motion data, and dynamics in collective systems.

Program

10:15 - 10:20

H?ffding/Glette 

Intro

10:20 - 11:00

Fernando Rosas (Imperial College)

Complexity analysis in time series data

11:00 - 11:15

Coffee break

 

11:15 - 12:15

RITMO researchers

Research topics related to or requesting input on complexity analysis (around 15 min each):

 

Julian Fuhrer & Alejandro Blenkmann 

Measuring complexity of iEEG responses

 

Victor Sanchez 

Complexity and fractal organization of human movement

 

Olivier Lartillot 

Complexity in musical scores

12:15 - 13:00

Lunch

 

13:00 - 13:30

Fernando Rosas

Complexity as complicatedness, Lempel-Ziv methods, IIT

13:30 - 15:00

All

Discussions

Abstract

Fernando Rosas: Complexity analysis in time series data

Complexity has many faces and interpretations; two interesting ones are "how hard is something to be described" and "how intertwined are the parts together." We will discuss various information-theoretic approaches to assess both aspects, with a particular emphasis in brain data. For the former category, our favourite measure is Shannon's entropy rate, and estimators based in either compression (Lempel-Ziv algorithm, Context Tree Weighting) and state-space approaches. For the latter, we will consider Granger Causality and Transfer Entropy, IIT (versions 0.1 and 2.0), and recent novel approaches based on Partial Information Decomposition.

Organizer

Kyrre Glette, Simon H?ffding and Julian Fuhrer
Published Feb. 26, 2020 12:47 PM - Last modified Feb. 19, 2024 6:12 PM