Author ORCID Identifier
0009-0001-7436-9586
Document Type
Dissertation
Date of Award
8-31-2024
Degree Name
Doctor of Philosophy in Mathematical Sciences - (Ph.D.)
Department
Mathematical Sciences
First Advisor
Wenge Guo
Second Advisor
Sunil Kumar Dhar
Third Advisor
Ji Meng Loh
Fourth Advisor
Antai Wang
Fifth Advisor
Guiling Wang
Abstract
In many machine learning applications, such as image tagging, document classi-fication, and medical diagnosis, a data instance can be associated with multiple classes in parallel so that each instance is associated with multiple response variables simultaneously defining multi-label classification. Standard multi-label classification methods that provide point predictions have been developed. They lack in quantifying the uncertainty of predictions. These methods also lack in accounting for label dependencies and are very computationally expensive. This dissertation develops two methods of multi-label classification using conformal prediction that quantify the uncertainty of predictions. Chapter 1 introduces notations and tools that have been used in the dissertation. Chapter 2 proposes the tree-based method using conformal prediction and the techniques of multiple hypothesis testing for multi-label classification. This is a non-parametric method that is shown to control type I errors at a pre-specified level. Two versions of the method, namely fixed-alpha and adaptive-alpha, have been developed to overcome the conservative behavior of the fixed-alpha approach. Chapter 3 introduces maximal conformity score (mxc) methods based on conformal prediction that provide prediction sets with proven statistical guarantees. This method has been shown to control k-FWER at a given significance level. It has been explicitly shown that any multiple testing procedure wrapped with hyper-parameter tuning produces an adaptive version, which overcomes the conservativeness of the fixed-alpha method. Experiments have been conducted on synthetic data and real-life data sets and the results have been reported. Finally, Chapter 4 summarizes the contributions and suggests possible future work.
Recommended Citation
Tyagi, Chhavi, "Multi-label classification using conformal prediction" (2024). Dissertations. 1769.
https://digitalcommons.njit.edu/dissertations/1769
