Comprehensive Validation on Reweighting Samples for Bias Mitigation via AIF360
Authors:
Christina Hastings Blow,
Lijun Qian,
Camille Gibson,
Pamela Obiomon,
Xishuang Dong
Abstract:
Fairness AI aims to detect and alleviate bias across the entire AI development life cycle, encompassing data curation, modeling, evaluation, and deployment-a pivotal aspect of ethical AI implementation. Addressing data bias, particularly concerning sensitive attributes like gender and race, reweighting samples proves efficient for fairness AI. This paper contributes a systematic examination of rew…
▽ More
Fairness AI aims to detect and alleviate bias across the entire AI development life cycle, encompassing data curation, modeling, evaluation, and deployment-a pivotal aspect of ethical AI implementation. Addressing data bias, particularly concerning sensitive attributes like gender and race, reweighting samples proves efficient for fairness AI. This paper contributes a systematic examination of reweighting samples for traditional machine learning (ML) models, employing five models for binary classification on the Adult Income and COMPUS datasets with various protected attributes. The study evaluates prediction results using five fairness metrics, uncovering the nuanced and model-specific nature of reweighting sample effectiveness in achieving fairness in traditional ML models, as well as revealing the complexity of bias dynamics.
△ Less
Submitted 19 December, 2023;
originally announced December 2023.
Inverse Quantum Fourier Transform Inspired Algorithm for Unsupervised Image Segmentation
Authors:
Taoreed Akinola,
Xiangfang Li,
Richard Wilkins,
Pamela Obiomon,
Lijun Qian
Abstract:
Image segmentation is a very popular and important task in computer vision. In this paper, inverse quantum Fourier transform (IQFT) for image segmentation has been explored and a novel IQFT-inspired algorithm is proposed and implemented by leveraging the underlying mathematical structure of the IQFT. Specifically, the proposed method takes advantage of the phase information of the pixels in the im…
▽ More
Image segmentation is a very popular and important task in computer vision. In this paper, inverse quantum Fourier transform (IQFT) for image segmentation has been explored and a novel IQFT-inspired algorithm is proposed and implemented by leveraging the underlying mathematical structure of the IQFT. Specifically, the proposed method takes advantage of the phase information of the pixels in the image by encoding the pixels' intensity into qubit relative phases and applying IQFT to classify the pixels into different segments automatically and efficiently. To the best of our knowledge, this is the first attempt of using IQFT for unsupervised image segmentation. The proposed method has low computational cost comparing to the deep learning-based methods and more importantly it does not require training, thus make it suitable for real-time applications. The performance of the proposed method is compared with K-means and Otsu-thresholding. The proposed method outperforms both of them on the PASCAL VOC 2012 segmentation benchmark and the xVIEW2 challenge dataset by as much as 50% in terms of mean Intersection-Over-Union (mIOU).
△ Less
Submitted 11 January, 2023;
originally announced January 2023.
