Human age ranking from pairwise comparison data via HodgeRank

Slides:



Advertisements
Similar presentations
Descriptive Statistics. Descriptive Statistics: Summarizing your data and getting an overview of the dataset  Why do you want to start with Descriptive.
Advertisements

Dong Liu Xian-Sheng Hua Linjun Yang Meng Weng Hong-Jian Zhang.
Topic-Sensitive PageRank Taher H. Haveliwala. PageRank Importance is propagated A global ranking vector is pre-computed.
The Very Small World of the Well-connected. (19 june 2008 ) Lada Adamic School of Information University of Michigan Ann Arbor, MI
Predicting Missing Provenance Using Semantic Associations in Reservoir Engineering Jing Zhao University of Southern California Sep 19 th,
Slide Systems of Linear Equations A system of linear equations consists two or more linear equations.
Liang Xiang, Quan Yuan, Shiwan Zhao, Li Chen, Xiatian Zhang, Qing Yang and Jimeng Sun Institute of Automation Chinese Academy of Sciences, IBM Research.
A Comparative Study of Search Result Diversification Methods Wei Zheng and Hui Fang University of Delaware, Newark DE 19716, USA
MARE 250 Dr. Jason Turner Hypothesis Testing III.
Qualitative and Quantitative Research Quantitative Deductive: transforms general theory into hypothesis suitable for testing Deductive: transforms general.
Review and Validation of ISAT Performance Levels for 2006 and Beyond MetriTech, Inc. Champaign, IL MetriTech, Inc. Champaign, IL.
Aadil Zia Khan and Shahab Baqai LUMS School of Science and Engineering QoS Aware Path Selection in Content Centric Networks Fahad R. Dogar Carnegie Mellon.
Anindya Bhattacharya and Rajat K. De Bioinformatics, 2008.
Chapter 1 Introduction to Statistics. Statistical Methods Were developed to serve a purpose Were developed to serve a purpose The purpose for each statistical.
Empirical Studies as a Vehicle for Software Engineering Projects Claes Wohlin Dept. of Communication Systems Lund University, Sweden
Binxing Jiao et. al (SIGIR ’10) Presenter : Lin, Yi-Jhen Advisor: Dr. Koh. Jia-ling Date: 2011/4/25 VISUAL SUMMARIZATION OF WEB PAGES.
Zibin Zheng DR 2 : Dynamic Request Routing for Tolerating Latency Variability in Cloud Applications CLOUD 2013 Jieming Zhu, Zibin.
Sample Size Determination in Studies Where Health State Utility Assessments Are Compared Across Groups & Time Barbara H Hanusa 1,2 Christopher R H Hanusa.
Yongqin Gao, Greg Madey Computer Science & Engineering Department University of Notre Dame © Copyright 2002~2003 by Serendip Gao, all rights reserved.
Adjudicator Agreement and System Rankings for Person Name Search Mark Arehart, Chris Wolf, Keith Miller The MITRE Corporation {marehart, cwolf,
Learning Photographic Global Tonal Adjustment with a Database of Input / Output Image Pairs.
Scientific Method  Hypothesis  Experiment  Observations  Conclusion.
Learning to Rank: From Pairwise Approach to Listwise Approach Authors: Zhe Cao, Tao Qin, Tie-Yan Liu, Ming-Feng Tsai, and Hang Li Presenter: Davidson Date:
Predicting Consensus Ranking in Crowdsourced Setting Xi Chen Mentors: Paul Bennett and Eric Horvitz Collaborator: Kevyn Collins-Thompson Machine Learning.
Statistics & Evidence-Based Practice
Chapter 8 Introducing Inferential Statistics.
DTC Quantitative Methods Bivariate Analysis: t-tests and Analysis of Variance (ANOVA) Thursday 20th February 2014  
Structure Similarity Index
CHAPTER 12 Statistics.
TYPES OF ACADEMIC ESSAYS
A Melody Composer for both Tonal and Non-Tonal Languages
Descriptive Statistics
Introduction to Statistics
Iwona Skalna Department of Applied Informatics Cracow, Poland
Can Computer Algorithms Guess Your Age and Gender?
Mai Zhou Dept. of Statistics, University of Kentucky Chengwu Cui
12 Inferential Analysis.
Just What Is Science Anyway???
Applying the Analytical Hierarchy Process to determine the optimal
Advanced Artificial Intelligence Evolutionary Search Algorithm
Keller: Stats for Mgmt & Econ, 7th Ed
A Comparative Study for Single Image Blind Deblurring
Enhancing Attack Robustness of Scale-free Networks by Camouflage
Statistical Analysis Error Bars
i/o research methods: final Presentation guidelines
Research methods (2013) Other research methods paper going on the website Inferential statistics pack.
CHAPTER 1 Exploring Data
12 Inferential Analysis.
CHAPTER 1 Exploring Data
CHAPTER 1 Exploring Data
CHAPTER 12 Statistics.
CHAPTER 1 Exploring Data
Analysis for Predicting the Selling Price of Apartments Pratik Nikte
DESCRIPTIVE STATISTICS BORAM KANG
Notes Translations.
An Introductory Tutorial
Creating a Pairwise Comparison Chart
Actively Learning Ontology Matching via User Interaction
Informational Reports
CHAPTER 12 Statistics.
Experiments with More Than Two Groups
Learning to Rank with Ties
CHAPTER 1 Exploring Data
Data analysis LO: Identify and apply different methods of measuring central tendencies and dispersion.
Indicator 3.05 Interpret marketing information to test hypotheses and/or to resolve issues.
Chapter 13: Using Statistics
CHAPTER 1 Exploring Data
Measurement Basics Chapter 13.
CHAPTER 12 Statistics.
Low-Rank Sparse Feature Selection for Patient Similarity Learning
Presentation transcript:

Human age ranking from pairwise comparison data via HodgeRank Lui Go Nam Department of Mechanical and Aerospace Engineering Hong Kong University of Science and Technology

Introduction Pairwise comparison is a comfort way for volunteers to present their opinions. However, when the number of subjects increases, the difficulty of direct pairwise comparison is impossible. Thus, effective pairwise comparison analysis framework is needed.

Introduction HodgeRank on Random Graphs (HRRG) is proposed as a general framework for incomplete pairwise comparison data analysis. Every subjects (portraits e.g.) in comparison is regarded as a graph node. An assessor collects random samples of node pairs or edges independently with a distribution (I.I.D. e.g.).

Content of the project 1. Global score generated by four generalized linear model. (Uniform model, Bradley-Terry, Thurstone-Mosteller, Arcsin) 2.Comparions with real age distribution after standardization. 3.Inconsistency analysis. 4.Error analysis.

Dataset description 14,011 pairwise comparisons from 94 annotators of 30 portrait photos. * I didn’t consider the impact of “hard to judge” option into the model After rearrangement, there are 12778 sets of pairwise comparison left.

Global Score Bradley-Terry model has a similar trend as Thurstone-Mosteller model while uniform model’s trend is likely to angular transform model’s.

Comparison

Inconsistency Analysis , Brad-Terry model has the highest total inconsistency and uniform model has the lowest. Harmonic inconsistency are small for each models

Error Analysis each model has similar RMSE which is around 0.49 and Thurstone-Mosteller model has the lowest which close to 0.48.

Error Analysis This figure shows that the No.21, No.24, No.20 and No.3 is relatively hard to recognize their real age.

Interesting fact The left is No.21, which is 20 years old and the right is No.24, which is 18 years old. Actually it’s hard to tell their real age. :)

Conclusion HHRG is a suitable framework for managing the incomplete pairwise comparison data on the human age ranking. Based on the Error analysis and Inconsistency analysis, I am convinced that Thurstone-Mosteller model is the most appropriate GLM on human age ranking.