Semester of Graduation
Summer 2026
Degree Type
Dissertation/Thesis
Degree Name
Data Science and Analytics
Department
School of Data Science and Analytics
Committee Chair/First Advisor
Dr. Ramazan Aygun
Second Advisor
Dr. Truong Tran
Third Advisor
Dr. Jiajing Huang
Abstract
Machine learning models continue to face trust issues, particularly in fields where a single erroneous prediction can result in significant costs or disastrous outcomes. WisdomNet architecture facilitates to achieve a zero error rate if certain conditions are met by rejecting data instances it is uncertain about and delegating those cases to a human expert. However, WisdomNet still faces major challenges, such as a high rejection rate and determining the appropriate point to stop fine-training.We examine the underlying causes of the high rejection rate, contingent upon having a satisfactory base model upon which WisdomNet is constructed and propose the use of two data difficulty measures, namely the polarized k-entropy measure and the dual probability difficulty measure, to identify data instances that are likely to contribute to this rejection rate in both binary and multi-labeled datasets, respectively. To determine the appropriate point to stop fine-training, we propose a novel technique called Difficulty-Driven Fine Training (DDFT), which not only determines when to stop fine-training but also minimizes the rejection rate. This technique focuses on excluding difficult data from the validation set through the use of our proposed data difficulty measures and fine-training WisdomNet with misclassified samples until the new validation set achieves a zero error rate. We conducted experiments to show the effectiveness of our technique. Our experimental results show that our method reduces the rejection rate of WisdomNet.