paper-with-me

Papers

Whose Bias?

2021-11-19 · Vasudha Jain, Mark Whitmeyer

Law enforcement acquires costly evidence with the aim of securing the conviction of a defendant, who is convicted if a decision-maker's belief exceeds a certain threshold. Either law enforcement or the decision-maker is biased and is initially overconfident that the defendant is guilty. Although an innocent defendant always prefers an unbiased decision-maker, he may prefer that law enforcement have some bias to none. Nevertheless, fixing the level of bias, an innocent defendant may prefer that the decision-maker, not law enforcement, is biased.

📄 PDF Abstract BibTeX arXiv:2111.10335

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Equity Bias: An Ethical Framework for AI Design

2026-01-28 · Mary Lockwood arxiv

Equity Bias is a philosophical and practical framework for building smarter, more equitable AI systems. Grounded in hermeneutic philosophy and epistemic injustice theory, it treats bias not as an error to eliminate but a…

Uncertainty Quantification for Evaluating Machine Translation Bias

2025-07-24 · Ieva Raminta Staliūnaitė, Julius Cheng, Andreas Vlachos arxiv

The predictive uncertainty of machine translation (MT) models is typically used as a quality estimation proxy. In this work, we posit that apart from confidently translating when a single correct translation exists, mode…

Machine Translation

Uncertainty and Inclusivity in Gender Bias Annotation: An Annotation Taxonomy and Annotated Datasets of British English Text

2022-07-01 · NAACL (GeBNLP) 2022 7 · Lucy Havens, Beatrice Alex, Benjamin Bach, Melissa Terras

Mitigating harms from gender biased language in Natural Language Processing (NLP) systems remains a challenge, and the situated nature of language means bias is inescapable in NLP data. Though efforts to mitigate gender …

Language ModelingLanguage Modelling

Adam Simplified: Bias Correction Debunked

2025-11-25 · Sam Laing, Antonio Orvieto arxiv

The Adam optimizer is a cornerstone of modern deep learning, yet the empirical necessity of each of its individual components is often taken for granted. This paper presents a focused investigation into the role of bias-…

Language Modelling

Zero-bias autoencoders and the benefits of co-adapting features

2014-02-13 · Kishore Konda, Roland Memisevic, David Krueger

Regularized training of an autoencoder typically results in hidden unit biases that take on large negative values. We show that negative biases are a natural result of using a hidden layer whose responsibility is to both…