RANK ANALYSIS OF INCOMPLETE BLOCK DESIGNS

By R. Bradley, M. E. Terry. 1952

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

Citations in the corpus

Direct Preference Optimization: Your Language Model is Secretly a Reward Model (2023, methodology): “Like existing algorithms, DPO relies on a theoretical preference model (such as the Bradley-Terry model; [5]) that measures how well a given reward function aligns with empirical preference data.”
Deep Reinforcement Learning from Human Preferences (2017, methodology): “(1) We choose ˆ r to minimize the cross-entropy loss between these predictions and the actual human labels: This follows the Bradley-Terry model (Bradley and Terry, 1952) for estimating score functions from pairwise preferences, and is the specialization of the Luce-Shephard choice rule (Luce,…”
Reward learning from human preferences and demonstrations in Atari (2018, methodology): “…)) We train r̂ to minimize the cross-entropy loss between these predictions and the actual judgment labels: loss(r̂) = − ∑ (σ1,σ2,µ)∈A µ(1) log P̂ [σ1 σ2] + µ(2) log P̂ [σ2 σ1] This follows the Bradley-Terry model (Bradley and Terry, 1952) for estimating score functions from pairwise preferences.”
Reliability and Learnability of Human Bandit Feedback for Sequence-to-Sequence Reinforcement Learning (2018, methodology): “We use deep learning models to train reward estimators by regression against cardinal feedback, and by fitting a Bradley-Terry model (Bradley and Terry, 1952) to ordinal feedback.”
Preference-based Reinforcement Learning with Finite-Time Guarantees (2020, unknown): “The comparisons are generated following the Bradley-Terry-Luce model [9]: φ(τ1, τ2) = 1 1+exp(−(r(τ1)−r(τ2))/c) , with c being either 0.”
Squibs: Evaluating Human Pairwise Preference Judgments (2015, methodology): “…earliest generally cited models for pairwise comparisons are the Thurstone model (Thurstone 1927) and the closely related Bradley-Terry (BT) model (Bradley and Terry 1952); these have connections to the IRT models, widely used in analyzing responses to questionnaires, which Hopkins and May (2013)…”
Preferences on a Budget: Prioritizing Document Pairs when Crowdsourcing Relevance Judgments (2022, methodology): “• a method to simulate all possible preference judgments for a document pool; • document pairs selection strategies to decide which pair to judge sequentially, thus being able to stop once a judging budget has been consumed; • dataset transformations that assign scores to documents based on availabl”
Data Analysis Using Regression and Multilevel/Hierarchical Models: Understanding and summarizing the fitted models (2006, background): “Connections between logit/probit regressions and choice models have been studied in psychology, economics, and political science; some important references are Thurstone (1927a, b), Wallis and Friedman (1942), Mosteller (1951), Bradley and Terry (1952), and McFadden (1973). Tobit models are named af”
Default Bayes factors for ANOVA designs (2012, background): “Some psychological process models are members of the generalized linear model class, including BradleyTerry-Luce scaling models (Bradley & Terry, 1952; Luce, 1959) and a wide class of signal-detection models (DeCarlo, 1998).”
Ranking soccer teams on basis of their current strength: a comparison of maximum likelihood approaches (2017, background): “Bradley-Terry (BT) models (Bradley and Terry, 1952) have been designed to predict the outcome of pairwise comparisons.”
One-vs-Each Approximation to Softmax for Scalable Estimation of Probabilities (2016, background): “The fact that the one-vs-each bound in (4) is a product of pairwise probabilities suggests that there is a connection with Bradley-Terry (BT) models [6, 10] for learning individual skills from paired comparisons and the associated multiclass classification systems obtained by combining binary classi”
Deep Networks for Image Super-Resolution with Sparse Prior (2015, methodology): “All the evaluation results can be summarized into a 7×7 winning matrix for 7 methods (including ground truth), based on which we fit a Bradley-Terry [4] model to estimate the subjective score for each method so that they can be ranked.”
GROUPED SPARSE PAIRED COMPARISONS IN THE BRADLEY-TERRY MODEL (2011, background): “To describe the probabilities of the possible outcomes in paired comparisons, Bradley and Terry (1952) suggested that the probability subject i beats subject j be specified by pij = ui ui + uj , (1.1) where the “merit vector” (u0, u1, . . . , ut) represents the merit parameters of t+1 subjects.”
The Sample Complexity of Best-k Items Selection from Pairwise Comparisons (2020, methodology): “We note that many widely used parametric models such as the Bradley-Terry-Luce (Bradley and Terry, 1952; Luce, 2012) (BTL) and Thurstone’s model (Thurstone, 1927) satisfy SST and STI, and thus, the algorithms in this paper can be directly used under these models.”
Handbook of Polytomous Item Response Theory Models (2010, background): “An earlier application of maximum likelihood in estimating binomial response relations appears in Bradley and Terry (1952). They assume the model”