Data augmentation for fairness-aware machine learning

Pastaltzidis, Ioannis; Dimitriou, Nikolaos; Quezada‐Tavárez, Katherine; Aidinlis, Stergios; Marquenie, Thomas; Gurzawska, Agata; Tzovaras, Dimitrios

doi:10.1145/3531146.3534644

Cited by 21 publications

(11 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This improvement effectively dispels the commonly held notion that efforts to increase fairness invariably compromise performance. In fact, our findings are in line with recent research on policing and recidivism algorithms, such as Urcuqui et al (2020) and Pastaltzidis et al (2022), which assert that there is no strict trade-off between fairness and accuracy.…”

Section: Research Objective 2: Bias Mitigationsupporting

confidence: 93%

Mitigating Age-related Bias in Predictive Policing Algorithms

Almasoud

2023

Preprint

View full text Add to dashboard Cite

This study addressed algorithmic bias in predictive policing, focusing on the Chicago Police Department's Strategic Subject List (SSL) dataset. We specifically focused on identifying and mitigating age-related biases, a notably underexplored area in prior research. Our research introduced Conditional Score Recalibration as a bias mitigation strategy alongside the well-established Class Balancing technique. Conditional Score Recalibration involved reassessing and adjusting risk scores for individuals initially assigned moderately high-risk scores in the dataset. This recalibration marked such individuals as low risk if they met three conditions, namely: no prior arrests for violent offenses, no previous arrests for narcotic offenses, and having never been involved in shooting incidents. These fairness strategies were implemented on the Random Forest model, and the fairness metrics employed included Equality of Opportunity Difference, Average Odds Difference, and Demographic Parity. The results showed a significant improvement in model fairness, particularly for age biases, without compromising the model's accuracy. These findings challenged the often-assumed trade-off between fairness and accuracy, underscoring the feasibility of achieving fairness without compromising accuracy.

show abstract

Section: Research Objective 2: Bias Mitigationsupporting

confidence: 93%

Mitigating Age-related Bias in Predictive Policing Algorithms

Almasoud

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Utilization of both machine-assisted judgments (e.g., predictive policing, where AI notifies human officers of potential crime areas) and automated decision-making (e.g., the automated imposition of traffic penalties on the basis of AI surveillance systems) is encompassed in algorithmic law enforcement. Although these technologies may streamline operations and increase productivity, they also raise significant apprehensions regarding accuracy, bias, openness, and responsibility (Pastaltzidis et al, 2022). These themes thus underscore the critical necessity for ethical and legal protections and reflect the revolutionary impact of AI on our decision-making systems.…”

Section: Discussionmentioning

confidence: 99%

Mapping Ethical Artificial Intelligence Policy Landscape: A Mixed Method Analysis

Saheb,

Saheb

2024

Sci Eng Ethics

View full text Add to dashboard Cite

As more national governments adopt policies addressing the ethical implications of artificial intelligence, a comparative analysis of policy documents on these topics can provide valuable insights into emerging concerns and areas of shared importance. This study critically examines 57 policy documents pertaining to ethical AI originating from 24 distinct countries, employing a combination of computational text mining methods and qualitative content analysis. The primary objective is to methodically identify common themes throughout these policy documents and perform a comparative analysis of the ways in which various governments give priority to crucial matters. A total of nineteen topics were initially retrieved. Through an iterative coding process, six overarching themes were identified: principles, the protection of personal data, governmental roles and responsibilities, procedural guidelines, governance and monitoring mechanisms, and epistemological considerations. Furthermore, the research revealed 31 ethical dilemmas pertaining to AI that had been overlooked previously but are now emerging. These dilemmas have been referred to in different extents throughout the policy documents. This research makes a scholarly contribution to the expanding field of technology policy formulations at the national level by analyzing similarities and differences among countries. Furthermore, this analysis has practical ramifications for policymakers who are attempting to comprehend prevailing trends and potentially neglected domains that demand focus in the ever-evolving field of artificial intelligence.

show abstract

“…Bar charts were generated to visually compare each task category's baseline and improved performance metrics. This approach aids in identifying the areas where algorithmic enhancements are most impactful, contributing to a nuanced understanding of performance variations across different machine-learning tasks (Pastaltzidis I. et al, 2022). The second dimension of the research methodology focuses on the demographic analysis of contributors participating in the image categorization and content moderation tasks.…”

Section: Methodsmentioning

confidence: 99%

Ensuring ethical integrity and bias reduction in machine learning models

Gomathi,

Pawar

et al. 2024

The Scientific Temper

View full text Add to dashboard Cite

This research focused on the multifaceted realm of machine learning algorithms, focusing on the pivotal themes of ethical concerns and bias mitigation (Zeba G. et al., 2021). Employing a dual-pronged research methodology, the study first evaluates algorithmic performance across diverse tasks, such as audio transcription, content moderation, and system implementation. The research uses quantitative assessments and visual comparisons to highlight nuanced improvements in algorithmic efficiency and accuracy. The second dimension involves an in-depth analysis of demographic contributions in tasks like image categorization and content moderation. By scrutinizing the geographical distribution of contributors and demographics like age and gender, the study aims to unravel potential correlations between algorithmic effectiveness and contributor demographics. The graphical representations provide valuable visual insights, including bias distribution across categories, evolution over time, and baseline and improved performance comparisons. The findings contribute to the discourse on responsible AI development, emphasizing the need for tailored enhancements and inclusive participant recruitment strategies. Complemented by comprehensive results and discussions, this research methodology lays a robust foundation for addressing ethical concerns and advancing bias mitigation strategies in machine learning algorithms.

show abstract

Data augmentation for fairness-aware machine learning

Cited by 21 publications

References 17 publications

Mitigating Age-related Bias in Predictive Policing Algorithms

Mitigating Age-related Bias in Predictive Policing Algorithms

Mapping Ethical Artificial Intelligence Policy Landscape: A Mixed Method Analysis

Ensuring ethical integrity and bias reduction in machine learning models

Contact Info

Product

Resources

About