Improving Classification When a Class Hierarchy is Available Using a Hierarchy-Based Prior

  title={Improving Classification When a Class Hierarchy is Available Using a Hierarchy-Based Prior},
  author={Babak Shahbaba and Radford M. Neal},
  journal={Bayesian Analysis},
We introduce a new method for building classification models when we have prior knowledge of how the classes can be arranged in a hierarchy, based on how easily they can be distinguished. The new method uses a Bayesian form of the multinomial logit (MNL, a.k.a. “softmax”) model, with a prior that introduces correlations between the parameters for classes that are nearby in the tree. We compare the performance on simulated data of the new method, the ordinary MNL model, and a model that uses the… 

Figures and Tables from this paper

Restricted Boltzmann Machine for Classification with Hierarchical Correlated Prior

This paper proposes a hierarchical correlated RBM for classification problem, which generalizes the classification RBM with sharing information among different classes and introduces orthogonal restrictions to the objective function.

Large-scale Structured Learning

This thesis studies large-scale structured learning in the context of supervised, unsupervised and semisupervised settings and proposes two frameworks that can leverage hierarchical dependencies using Bayesian priors and a non-Bayesian Risk minimization framework that incorporates hierarchical and graphical dependencies into the regularization structure.

Bayesian models for Large-scale Hierarchical Classification

A set of Bayesian methods to model hierarchical dependencies among class labels using multivariate logistic regression, where the parent-child relationships are modeled by placing a hierarchical prior over the children nodes centered around the parameters of their parents.

Gene function classification using Bayesian models with hierarchy-based priors

Together, these results show that gene function can be predicted with higher accuracy than previously achieved, using Bayesian models that incorporate suitable prior information.

Multimedia Tools and Applications for Environmental & Biodiversity Informatics

The goal of this introductory chapter is to give a global picture of that domain and to overview the research works presented in this book.

On the role of classification in patent invalidity searches

A similarity measure for hierarchically-ordered patent classes and subclasses is examined and returned a ranked list of candidate patents, using a similarity measure that has demonstrated its effectiveness when applied to WordNet ontologies.

A Survey on Visual Transfer Learning using Knowledge Graphs

A broad overview of knowledge graph embedding methods is provided and several joint training objectives suitable to combine them with high dimensional visual embeddings are described, to help researchers find meaningful evaluation benchmarks.

Solving Long-tailed Recognition with Deep Realistic Taxonomic Classifier

Deep realistic taxonomic classifier (Deep-RTC) is proposed as a new solution to the long-tail problem, combining realism with hierarchical predictions, and outperforms the state-of-the-art methods in longtailed recognition, hierarchical classification, and learning with rejection literature using the proposed correctly predicted bits (CPB) metric.

A Multi-Layer Perceptron With a Hierarchical Prior for Operating State Recognition of Ship Propulsion Systems

To define more clearly vibration-related problems of ship propulsion systems, a procedure incorporating operating state recognition into conventional vibration analysis is proposed in this paper.



A statistical learning approach to document image analysis

This paper develops a new software environment for manual page image segmentation and labeling, and uses it to create a dataset containing 932 page images from academic journals, and develops a physical layout analysis algorithm based on a logistic regression classifier, which is found to outperform existing algorithms of comparable complexity.

Hierarchical document categorization with support vector machines

A novel hierarchical classification method that generalizes Support Vector Machine learning and that is based on discriminant functions that are structured in a way that mirrors the class hierarchy is proposed.

Application of Statistical Pattern Recognition to Document Segmentation and Labelling

  • Master’s Thesis,
  • 2005

Bayesian learning for neural networks

Econometric Models for Probabilistic Choice Among Products

I understand the discipline of marketing exists to answer questions such as: "Will housepersons buy more Brand A soap if its perfume content is increased?" Traditional econometric demand analysis

Additive similarity trees

A computer program, ADDTREE, for the construction of additive trees is described and applied to several sets of data, and some empirical and theoretical advantages of tree representations over spatial representations of proximity data are illustrated.

Automatic information structuring and retrieval.

Applied Regression Analysis, Linear Models, and Related Methods

PART ONE: PRELIMINARIES Statistics and Social Science What Is Regression Analysis? Examining Data Transforming Data PART TWO: LINEAR MODELS AND LEAST SQUARES Linear Least-Squares Regression

Functions of the gene products of Escherichia coli

  • M. Riley
  • Biology
    Microbiological reviews
  • 1993
A scheme to categorize cellular functions is used and the occurrence in the E. coli genome of redundant pairs and groups of genes of identical or closely similar function, as well as variation in the degree of density of genetic information in different parts of the genome are discussed.

Comparison of functional annotation schemes for genomes

This survey highlights many issues related to the design and implementation of gene product functional classifications, which are discussed in the light of emerging 'second-generation' schemes.