Social and ecological factors influencing movement and organizational patterns in sheep Habiba, Caitlin Barale, Ipek Kulahci, Rajmonda Sulo and Khairi.

Slides:



Advertisements
Similar presentations
Yinyin Yuan and Chang-Tsun Li Computer Science Department
Advertisements

Mobile Communication Networks Vahid Mirjalili Department of Mechanical Engineering Department of Biochemistry & Molecular Biology.
Mining Compressed Frequent- Pattern Sets Dong Xin, Jiawei Han, Xifeng Yan, Hong Cheng Department of Computer Science University of Illinois at Urbana-Champaign.
Cluster Analysis: Basic Concepts and Algorithms
An Introduction to Multivariate Analysis
Energy-Efficient Computing for Wildlife Tracking: Design Tradeoffs and Early Experiences with ZebraNet Presented by Eric Arnaud Makita
Overarching Goal: Understand that computer models require the merging of mathematics and science. 1.Understand how computational reasoning can be infused.
Probabilistic & Statistical Techniques Eng. Tamer Eshtawi First Semester Eng. Tamer Eshtawi First Semester
Visualizing the Evolution of Community Structures in Dynamic Social Networks Khairi Reda Department of Computer Science University of Illinois at Chicago.
Flocks, Herds and Schools Modeling and Analytic Approaches.
Patch to the Future: Unsupervised Visual Prediction
Assessment. Schedule graph may be of help for selecting the best solution Best solution corresponds to a plateau before a high jump Solutions with very.
© Tan,Steinbach, Kumar Introduction to Data Mining 4/18/ What is Cluster Analysis? l Finding groups of objects such that the objects in a group will.
Automatic Identification of ROIs (Regions of interest) in fMRI data.
Copyright 2004 David J. Lilja1 What Do All of These Means Mean? Indices of central tendency Sample mean Median Mode Other means Arithmetic Harmonic Geometric.
Probabilistic Aggregation in Distributed Networks Ling Huang, Ben Zhao, Anthony Joseph and John Kubiatowicz {hling, ravenben, adj,
Mutual Information Mathematical Biology Seminar
Applied Geostatistics Geostatistical techniques are designed to evaluate the spatial structure of a variable, or the relationship between a value measured.
Dimensional reduction, PCA
Un Supervised Learning & Self Organizing Maps Learning From Examples
Dimension reduction : PCA and Clustering Slides by Agnieszka Juncker and Chris Workman.
Sampling Distributions
BHS Methods in Behavioral Sciences I
A Framework For Community Identification in Dynamic Social Networks Chayant Tantipathananandh Tanya Berger-Wolf David Kempe Presented by Victor Lee.
Correlation and Regression Analysis
The Context of Forest Management & Economics, Modeling Fundamentals Lecture 1 (03/30/2015)
Models of Influence in Online Social Networks
Objectives of Multiple Regression
Data Mining Techniques
SWARM INTELLIGENCE IN DATA MINING Written by Crina Grosan, Ajith Abraham & Monica Chis Presented by Megan Rose Bryant.
CENTRE FOR INNOVATION, RESEARCH AND COMPETENCE IN THE LEARNING ECONOMY Session 2: Basic techniques for innovation data analysis. Part I: Statistical inferences.
A Distributed and Privacy Preserving Algorithm for Identifying Information Hubs in Social Networks M.U. Ilyas, Z Shafiq, Alex Liu, H Radha Michigan State.
Chapter 2 The Research Enterprise in Psychology. n Basic assumption: events are governed by some lawful order  Goals: Measurement and description Understanding.
Topology Design for Service Overlay Networks with Bandwidth Guarantees Sibelius Vieira* Jorg Liebeherr** *Department of Computer Science Catholic University.
A framework For Community Identification in Dynamic Social Networks Chayant, Tanya Berger-Wolf, David Kempe [KDD’07] Advisor : Dr. Koh Jia-Ling Advisor.
LANGUAGE NETWORKS THE SMALL WORLD OF HUMAN LANGUAGE Akilan Velmurugan Computer Networks – CS 790G.
COMMON EVALUATION FINAL PROJECT Vira Oleksyuk ECE 8110: Introduction to machine Learning and Pattern Recognition.
Why Is It There? Getting Started with Geographic Information Systems Chapter 6.
HOW TO WRITE RESEARCH PROPOSAL BY DR. NIK MAHERAN NIK MUHAMMAD.
Multi-Agent Modeling of Societal Development and Cultural Evolution Yidan Chen, 2006 Computer Systems Research Lab.
Copyright © 2012, SAS Institute Inc. All rights reserved. ANALYTICS IN BIG DATA ERA ANALYTICS TECHNOLOGY AND ARCHITECTURE TO MANAGE VELOCITY AND VARIETY,
First topic: clustering and pattern recognition Marc Sobel.
A Toolkit for Remote Sensing Enviroinformatics Clustering Fazlul Shahriar, George Bonev Advisors: Michael Grossberg, Irina Gladkova, Srikanth Gottipati.
Applications of Spatial Statistics in Ecology Introduction.
Aron, Aron, & Coups, Statistics for the Behavioral and Social Sciences: A Brief Course (3e), © 2005 Prentice Hall Chapter 12 Making Sense of Advanced Statistical.
Why Quantify Landscape Pattern? Comparison (space & time) –Study areas –Landscapes Inference –Agents of pattern formation –Link to ecological processes.
A Framework for Finding Communities in Dynamic Social Networks David Kempe University of Southern California Chayant Tantipathananandh, Tanya Berger-Wolf.
Introduction to Basic Statistical Tools for Research OCED 5443 Interpreting Research in OCED Dr. Ausburn OCED 5443 Interpreting Research in OCED Dr. Ausburn.
Analyzing wireless sensor network data under suppression and failure in transmission Alan E. Gelfand Institute of Statistics and Decision Sciences Duke.
Data Analysis.
Network Community Behavior to Infer Human Activities.
BMTRY 763. Space-time (ST) Modeling (BDM13, ch 12) Some notation Assume counts within fixed spatial and temporal periods: map evolutions Both space and.
Performance of Adaptive Beam Nulling in Multihop Ad Hoc Networks Under Jamming Suman Bhunia, Vahid Behzadan, Paulo Alexandre Regis, Shamik Sengupta.
Dynamics of communities in two fission-fusion species, Grevy's zebra and onager Chayant Tantipathananandh 1, Tanya Y. Berger-Wolf 1, Siva R. Sundaresan.
Analyzing Expression Data: Clustering and Stats Chapter 16.
1 Finding Spread Blockers in Dynamic Networks (SNAKDD08)Habiba, Yintao Yu, Tanya Y., Berger-Wolf, Jared Saia Speaker: Hsu, Yu-wen Advisor: Dr. Koh, Jia-Ling.
Who herds the herders ? Albert Kao and Andrew Berdahl Quis custodiet ipsos custodes? - Juvenal, 1 st century C.E.
Example Apply hierarchical clustering with d min to below data where c=3. Nearest neighbor clustering d min d max will form elongated clusters!
Optimization Techniques for Natural Resources SEFS 540 / ESRM 490 B Lecture 1 (3/30/2016)
Evolution of Cooperation in Mobile Ad Hoc Networks Jeff Hudack (working with some Italian guy)
Chapter 14 EXPLORATORY FACTOR ANALYSIS. Exploratory Factor Analysis  Statistical technique for dealing with multiple variables  Many variables are reduced.
Methods of multivariate analysis Ing. Jozef Palkovič, PhD.
Introduction to Wireless Sensor Networks
Why Quantify Landscape Pattern?
Quantifying Scale and Pattern Lecture 7 February 15, 2005
Making Sense of Advanced Statistical Procedures in Research Articles
Multivariate environmental characterization of samples
CHAPTER 29: Multiple Regression*
An Introduction to Correlational Research
CSE572: Data Mining by H. Liu
Presentation transcript:

Social and ecological factors influencing movement and organizational patterns in sheep Habiba, Caitlin Barale, Ipek Kulahci, Rajmonda Sulo and Khairi Reda complementary approaches from ecology and computer science

How do we identify the key individuals in a group? Personality 'types’ Group social dynamics Group movement patterns

Approaches Ecology Direct observation Information on: Proximity Individual movement patterns Behavior Interactions Individual state Computer Science Remote sensing Information on: Proximity Group-level movement patterns Clustering Velocity Turning angle

How can we use concepts from computer science to study animal behavior? Association visualizations Static networks GPS clustering algorithms Dynamic networks Social network analysis Agent-based modeling

GPS Data  Data Collection  General Statistics  Map with GPS readings overlaid (Khairi)

Personality Analysis PCA to generate personality ‘types’ Static social networks Association visualizations

Push/pull Push/pushed Followed pulls Pull/follower Push/follower Lactation state Join time in the mornings Group joining Alone grazing Location- front Location- back Location- edge Location- front during movement Personality components

Push/pull Push/pushed Followed pulls Pull/follower Push/follower Lactation state Join time in the mornings Group joining Alone grazing Location- front Location- back Location- edge Location- front during movement Personality components PCA component 1 - spatial PCA component 2 - behavioral

PCA 2 Push/pull Push/pushed Followed pulls PCA1 Location- back Location- edge Location- front during movement Personality scores

Group movement can be initiated either by pulling or pushing

Pull-follow and displacement networks What does the network look like from the perspective of individuals?

Social Network metrics Degree Betweenness Closeness Cluster coefficient Eigenvector

Key individuals – pull network Size: # of times an individual pulls Color: PCA category Lactating Not-lactating

Key individuals – push network Size: # of times an individual is pushed Color: PCA category Lactating Not-lactating

Key individuals – nearest neighbor network Size: # of times two individuals are neighbors while aligned-grazing Color: PCA category Lactating Not-lactating

Moving from static networks to dynamic ones

Static Vs Dynamic Networks T = i … T = i+1T = i+2T = i+3T = i+4 …

GPS clustering  What is an edge ?  Use k-means algorithm to spatially cluster gps readings. Any two sheep in the same cluster have an edge between them.  Use the Simple Structure Index (SSI) to determine the number of optimal clusters in each time interval

 Using different graph theoretic measures, the goal is to identify sampling rates at which the measures contain minimal noise while still being informative of the underlying dynamics in the network. What is the right temporal resolution?  Apply information theoretic concepts to quantify noise and information and compute the best trade- off across different sampling rates.

Algorithm TWIN  For w < w max * Compute the interaction network. * Compute the different structure measures such as density, and centrality, etc.  Quantify the amount of noise and information inherent in the network measure.  Compute a measure of goodness of fit based on a trade-off between the two.

TWIN Results Optimal interval = 3min Optimal interval = 5min

GPS data - issues  Data Quality  Discrepancy of sensor readings on the same sheep  Missing Values

Static Vs Dynamic Networks T = i … T = i+1T = i+2T = i+3T = i+4 …

Diffusion Process Independent Cascade Diffusion Model pr

Diffusion Process Linear Threshold Diffusion Model өiөi өjөj өkөk өlөl b i,j b j,l b i,l b k,l b i,k

Independent cascade

Text Linear threshold model

Community Identification A dynamic community is a subset of individuals that stick together over time. NOTE: Communities ≠ Groups t=1 t=2 t=3 t=4 t=5

Approach: Assumptions Individuals and groups represent exactly one community at a time. Concurrent groups represent distinct communities. Desired Required Conservatism: community affiliation changes are rare. Group Loyalty: individuals observed in a group belong to the same community. Parsimony: few affiliations overall for each individual.

Approach: Color = Community V alid coloring: distinct color of groups in each time step

Dynamic structure measures

Interesting behavioral aspects to model –Social foraging Alignment Sub-group formation from local interactions “Exploratory” random walks –Foraging strategy How do sheep find good food in patchy landscape? How does information about food resources spread in the herd? Competition –Pull and push Agent-based model

Interesting behavioral aspects to model –Social foraging Alignment Sub-group formation from local interactions “Exploratory” random walks –Foraging strategy How do sheep find good food in patchy landscape? How does information about food resources spread in the herd? Competition –Pull and push Agent-based model

–Sheep watch closest N neighbors and react to their behavior –No personality variance among agents –Agents assumed to be “hungry”. They graze continuously without rest –Agents have 3 states: GRAZE, WALK, PANIC –Low vegetation quality and patchy landscape; agents need to move frequently to maximize gain Assumptions

Agents switch back and forth between grazing and social walking State transition governed by a Gaussian timer –Grazing bouts: mean=6 sec, SD=2 sec –Social walking: mean=2-8 sec (depending on isolation), SD=2.5 sec Certain events cause agents to switch state –Agents minimize grazing and take longer walks if they are isolated from subgroup Timing

Local rules Collision avoidanceCohesionAlignment Random biased-walks P = 0.05 (when walking) Pulling

Agents that find themselves drifting away turn back and rejoin the subgroup Subgroup awareness

Demo

Conclusions

Sheep 15 – an influential individual?? Central in the observational networks –High betweeness and degree –Highest number of successful pulls Role in correlational plots –Often an outlier –Makes trends more significant GPS – doesn’t stand out Why?? Missing data…

Sheep 19 – a loner?? Static networks Not well connected –>>> Animation Little interaction with others in herd GPS – consistently low spread values

Are social network measures of individuals consistent across the three networks? Push & pull networks: + correlation (betweenness & closeness) Push & nearest neighbor networks: - correlation (eigenvector & closeness)

Network comparison  Is there consistency between observational subgroups and GPS clustering data?  Is there consistency between personality analysis obtained from observations and that obtained from GPS data aggregated at the “optimal” time interval?

Bite rate correlates with relative position p < 0.005, R 2 = 0.47

CS future directions Community structure evolution. Compare the observational data with “unsupervised” data collection and interpretation methods. Network comparison of different population to discover their social dynamics independent of their environment and other factors.