Visualizing Large, Complex Data



Similar documents
VISUALIZING HIERARCHICAL DATA. Graham Wills SPSS Inc.,

Hierarchical Data Visualization. Ai Nakatani IAT 814 February 21, 2007

Hierarchical Data Visualization

Hank Childs, University of Oregon

An Analytical Framework for Particle and Volume Data of Large-Scale Combustion Simulations. Franz Sauer 1, Hongfeng Yu 2, Kwan-Liu Ma 1

Big Data: Rethinking Text Visualization

Voronoi Treemaps in D3

Graph/Network Visualization

By LaBRI INRIA Information Visualization Team

Large Vector-Field Visualization, Theory and Practice: Large Data and Parallel Visualization Hank Childs + D. Pugmire, D. Camp, C. Garth, G.

Protein Protein Interaction Networks

HPC & Visualization. Visualization and High-Performance Computing

Visualizing Large Graphs with Compound-Fisheye Views and Treemaps

DIY Parallel Data Analysis

Visualization methods for patent data

Social Network Discovery based on Sensitivity Analysis

NEXT-GENERATION. EXTREME Scale. Visualization Technologies: Enabling Discoveries at ULTRA-SCALE VISUALIZATION

Machine Learning over Big Data

Cray: Enabling Real-Time Discovery in Big Data

Final Project Report

Visual Analysis Tool for Bipartite Networks

Interactive Exploration of Decision Tree Results

Visualization and Visual Analytics

Visualizing Data: Scalable Interactivity

CHAPTER 1 INTRODUCTION

Towards running complex models on big data

Clustering & Visualization

PACE Predictive Analytics Center of San Diego Supercomputer Center, UCSD. Natasha Balac, Ph.D.

Ordered Treemap Layouts

Interactive Data Mining and Visualization

COPYRIGHTED MATERIAL. Contents. List of Figures. Acknowledgments

Space-filling Techniques in Visualizing Output from Computer Based Economic Models

Information Visualization Multivariate Data Visualization Krešimir Matković

NStreamAware: Real-Time Visual Analytics for Data Streams to Enhance Situational Awareness

Parallel Analysis and Visualization on Cray Compute Node Linux

VisCG: Creating an Eclipse Call Graph Visualization Plug-in. Kenta Hasui, Undergraduate Student at Vassar College Class of 2015

Treemaps for Search-Tree Visualization

NEW VERSION OF DECISION SUPPORT SYSTEM FOR EVALUATING TAKEOVER BIDS IN PRIVATIZATION OF THE PUBLIC ENTERPRISES AND SERVICES

KATE GLEASON COLLEGE OF ENGINEERING. John D. Hromi Center for Quality and Applied Statistics

Parallel Simplification of Large Meshes on PC Clusters

Network Metrics, Planar Graphs, and Software Tools. Based on materials by Lala Adamic, UMichigan

Improvements of Space-Optimized Tree for Visualizing and Manipulating Very Large Hierarchies

HierarchyMap: A Novel Approach to Treemap Visualization of Hierarchical Data

NodeXL for Network analysis Demo/hands-on at NICAR 2012, St Louis, Feb 24. Peter Aldhous, San Francisco Bureau Chief.

University Turbine Systems Research 2012 Fellowship Program Final Report. Prepared for: General Electric Company

Review of Modern Techniques of Qualitative Data Clustering

Topic Maps Visualization

NakeDB: Database Schema Visualization

Principles of Data Mining by Hand&Mannila&Smyth

Spatio-Temporal Mapping -A Technique for Overview Visualization of Time-Series Datasets-

Christopher W. Muelder

Topological Properties

Visual Analysis of People s Calling Network from CDR data

The Design and Implement of Ultra-scale Data Parallel. In-situ Visualization System

SQL Server 2012 End-to-End Business Intelligence Workshop

Big Graph Processing: Some Background

Complex Network Visualization based on Voronoi Diagram and Smoothed-particle Hydrodynamics

Social and Economic Networks: Lecture 1, Networks?

. Learn the number of classes and the structure of each class using similarity between unlabeled training patterns

Lavastorm Analytic Library Predictive and Statistical Analytics Node Pack FAQs

QoS EVALUATION OF CLOUD SERVICE ARCHITECTURE BASED ON ANP

Visualizing e-government Portal and Its Performance in WEBVS

Visualization and Data Analysis

Hadoop MapReduce and Spark. Giorgio Pedrazzi, CINECA-SCAI School of Data Analytics and Visualisation Milan, 10/06/2015

A Property & Casualty Insurance Predictive Modeling Process in SAS

Visualization Techniques in Data Mining

A SIMULATOR FOR LOAD BALANCING ANALYSIS IN DISTRIBUTED SYSTEMS

Regular TreeMap Layouts for Visual Analysis of Hierarchical Data

Fast Multipole Method for particle interactions: an open source parallel library component

Network VisualizationS

HPC technology and future architecture

Specific Usage of Visual Data Analysis Techniques

Using an In-Memory Data Grid for Near Real-Time Data Analysis

Survey of Web Testing Techniques

USE OF EIGENVALUES AND EIGENVECTORS TO ANALYZE BIPARTIVITY OF NETWORK GRAPHS

a Data Science Univ. Piraeus [GR]

SciDAC Visualization and Analytics Center for Enabling Technology

HT2015: SC4 Statistical Data Mining and Machine Learning

REGULATIONS FOR THE DEGREE OF MASTER OF SCIENCE IN COMPUTER SCIENCE (MSc[CompSc])

BIG DATA VISUALIZATION. Team Impossible Peter Vilim, Sruthi Mayuram Krithivasan, Matt Burrough, and Ismini Lourentzou

EXPLORE THE WORLD OF BIG DATA

EM Clustering Approach for Multi-Dimensional Analysis of Big Data Set

Scientific Computing Programming with Parallel Objects

Sanjeev Kumar. contribute

International Journal of Computer Science Trends and Technology (IJCST) Volume 3 Issue 3, May-June 2015

Transcription:

Visualizing Large, Complex Data

Outline Visualizing Large Scientific Simulation Data Importance-driven visualization Multidimensional filtering Visualizing Large Networks A layout method Filtering methods

Supernova

Supernova

The Large Data Problem Shared resources Supercomputer Storage Visualization Machine

A Turbulent Lifted Autoignitive Ethylene/air Jet Flame HO 2 and mixture fraction isosurface HO 2 and OH

Complex, Multi-Scale Nature of Turbulent Flow Small eddies are hidden in the multi-layer flow

Feature-directed Data Reduction and Visualization

Feature-directed Data Reduction and Visualization Achieved over 80% saving In situ data reduction and triage can facilitate following data analysis and visualization!

In Situ Methods Enables Seeing all the data and capturing transient events at highest possible detail More effective data reduction More efficient postprocessing analysis and viz Monitoring/debugging of the simulation (ensuring the calculation is running well) Steering the simulation and driving the simulation with interactive analysis Tuning and optimizing the performance of the simulation/machine

Fusion

Multidimensional Particles Filtering

Multidimensional Particles Filtering Trapped particles that change direction frequently

Visualizing Large, Complex Networks Data are created and collected for a variety of purposes Internet is a source of massive data Cyber security Homeland security Business transactions Mobile device user data Health care data Email Relations in these data are often represented with graph/ networks for analysis To visualize a network, we need to lay it out

The Graph Layout Problem The cost of displaying a graph The hairball problem of large graph layouts Large, dense graphs become a mess Inefficient use of space Details cluttered Solutions Filtering Clustering Abstraction Focus+context California data 6,107 nodes 15,160 edges High dimensional embedding method

Space Partitioning Based Layout California data 6,107 nodes 15,160 edges Treemap Hibert curve Linlog Method 10,737s Radial Treemap Gosper curve

Space Filling Curve Based Layout Layout defined by clustering Space filing Interaction is very fast O( V ) Scales to large graphs Effective for Focus+Context Guaranteed aspect ratios Nodes don t become colinear Rendering is slower than layout A protein homology graph. Color corresponds to depth in the clustering hierarchy. V = 28,854, E = 1,180,816

Visualizing Internet Connectivity

Centrality Sensitivity Centralities (degree, between-ness, closeness, eigenvector, Markov, ) indicate how important a node is in a network. Studying the sensitivity and stability of a network in terms of different metrics for centrality allow us to Filter the network Search and explore in the network Obtain an overview of the network Compute sensitivity as the derivative of the centrality function, approximate derivatives of centrality using finite difference, and validate by computing the mean square error of the linear fit between the approximated and analytical values

Centrality Sensitivity Minimum spanning tree as the core network with centrality derivatives as edge weights Central nodes remain central Network of protein-protein interaction (~1500 nodes)

Overview of Sensitivity Friendster social network Links exhibit negative sensitivity (red) between cluster centers Astrophysics co-author network One competitive network (red) and one collaborative network (blue)

Summary Visualization, as a tool complementing conventional data analysis and mining methods, enhances our ability to utilize and communicate with data and knowledge In situ visualization and data reduction/triage is the most plausible solution to extreme scale scientific supercomputing Visual-based network analysis has become an essential tool, and interactivity is the key to understanding complex networks. The 1 st IEEE Symposium on Large Data Analysis and Visualization (LDAV), Oct 23-24, Providence, RI The 6 th Ultrascale Visualization Workshop, Supercomputing Conference (SC11), November 13, Seattle, WA