Results 61 to 70 of about 16,825 (201)

Debiasing knowledge graph embeddings [PDF]

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), 2020
It has been shown that knowledge graph embeddings encode potentially harmful social biases, such as the information that women are more likely to be nurses, and men more likely to be bankers. As graph embeddings begin to be used more widely in NLP pipelines, there is a need to develop training methods which remove such biases.
Joseph Fisher, Arpit Mittal, Dave Palfrey, Christos Christodoulopoulos   +3 more
openaire   +1 more source

Accountability as a Debiasing Strategy: Testing the Effect of Racial Diversity in Employment Committees [PDF]

, 2018
Congress passed Title VII of the Civil Rights Act of 1964 with the primary goal of integrating the workforce and eliminating arbitrary bias against minorities and other groups who had been historically excluded.
Bowman Williams, Jamillah
core   +1 more source

Evaluating the Underlying Gender Bias in Contextualized Word Embeddings [PDF]

, 2019
Gender bias is highly impacting natural language processing applications. Word embeddings have clearly been proven both to keep and amplify gender biases that are present in current data sources.
Basta, Christine, Casas, Noe, Costa-jussà, Marta R.   +2 more
core   +2 more sources

Forecast verification using information and noise

Quarterly Journal of the Royal Meteorological Society, EarlyView.
Verification of weather forecasts is usually expressed in terms of total error metrics. This is useful for end users of the forecasts but does not allow evaluation of the intrinsic information content of the forecasts. To overcome this limitation, we propose a new total error decomposition into information and noise error measures, connect it to ...
Massimo Bonavita, Alan J. Geer
wiley   +1 more source

Debiasing omission neglect

Journal of Business Research, 2006
Abstract Two experiments investigated the effectiveness of two new procedures for improving judgment by increasing sensitivity to missing information. When consumers are insensitive to important missing information, overly extreme product evaluations are formed.
Kardes, Frank, Posavac, Steven, Silvera, David, Cronley, Maria, Sanbonmatsu, David, Schertzer, Susan, Miller, Felicia, Herr, Paul, Chandrashekaran, Murali   +8 more
openaire   +2 more sources

Lifecycle‐Based Governance to Build Reliable Ethical AI Systems

Systems Research and Behavioral Science, EarlyView.
ABSTRACT Artificial intelligence (AI) systems represent a paradigm shift in technological capabilities, offering transformative potential across industries while introducing novel governance and implementation challenges. This paper presents a comprehensive framework for understanding AI systems through three critical dimensions: trustworthiness ...
Maikel Leon
wiley   +1 more source

The Behavioral Paradox: Why Investor Irrationality Calls for Lighter and Simpler Financial Regulation [PDF]

, 2012
It is widely believed that behavioral economics justifies more intrusive regulation of financial markets, because people are not fully rational and need to be protected from their quirks. This Article challenges that belief.
Juurikkala, Oskari
core   +3 more sources

Internal delensing of cosmic microwave background polarization B-Modes with the POLARBEAR experiment [PDF]

, 2019
International audienceUsing only cosmic microwave background polarization data from the polarbear experiment, we measure B-mode polarization delensing on subdegree scales at more than 5σ significance. We achieve a 14% B-mode power variance reduction, the
Adachi, S, Aguilar Faúndez, MAO, Akiba, Y, Ali, A, Arnold, K, Baccigalupi, C, Barron, D, Beck, D, Bianchini, F, Borrill, J, Carron, J, Cheung, K, Chinone, Y, Crowley, K, El Bouhargani, H, Elleflot, T, Errard, J, Fabbian, G, Feng, C, Fujino, T, Goeckner-Wald, N, Hasegawa, M, Hazumi, M, Hill, CA, Howe, L, Katayama, N, Keating, B, Kikuchi, S, Kusaka, A, Lee, AT, Leon, D, Linder, E, Lowry, LN, Matsuda, F, Matsumura, T, Minami, Y, Namikawa, T, Navaroli, M, Nishino, H, Peloton, J, Pham, ATP, POLARBEAR Collaboration, Poletti, D, Puglisi, G, Reichardt, CL, Segawa, Y, Sherwin, BD, Silva-Feaver, M, Siritanasak, P, Stompor, R, Tajima, O, Takatori, S, Tanabe, D, Teply, GP, Vergès, C   +54 more
core   +6 more sources

Generating Compressed Counterfactual Hard Negative Samples for Graph Contrastive Learning

CAAI Transactions on Intelligence Technology, EarlyView.
ABSTRACT Graph contrastive learning (GCL) relies on acquiring high‐quality positive and negative samples to learn the structural semantics of the input graph. Previous approaches typically sampled negative samples from the same training batch or an irrelevant external graph.
Haoran Yang, Hongxu Chen, Xiangyu Zhao, Sixiao Zhang, Xiangguo Sun, Qian Li, Hongzhi Yin, Guandong Xu   +7 more
wiley   +1 more source

Debiasing Strategies and Judicial Decision-Making: Exploring a Duty to Improve Judges’ Capabilities

Athena
Judicial decision-making carries profound consequences for individuals and society, with expectations that judges act with objectivity and neutrality.
Giovana Peluso Lopes
doaj   +1 more source