Tootfinder

Sources: Gemini's image generation tool had a technical fix to reduce bias in its outputs but Google didn't anticipate the overcorrection and wasn't transparent (Bloomberg)
https://www.bloomberg.com/news/articles/20

Google Left in ‘Terrible Bind’ By Pulling AI Feature After Right-Wing Backlash
Google tried using a technical fix to reduce bias in a feature that generates realistic-looking images of people. Instead, it set off a new diversity firestorm.

Interesting, some empirical research on how GPT4 /ChatGPT performs in summarizing. Still the low rate of errors can be unacceptable in some contexts. As noted "Life-critical medical decisions should remain based on full, critical, and thoughtful evaluation of the full text of research articles in context with clinical guidelines.".
http…

Quality, Accuracy, and Bias in ChatGPT-Based Summarization of Medical Abstracts
PURPOSE Worldwide clinical knowledge is expanding rapidly, but physicians have sparse time to review scientific literature. Large language models (eg, Chat Generative Pretrained Transformer [ChatGPT]), might help summarize and prioritize research articles to review. However, large language models sometimes “hallucinate” incorrect information. METHODS We evaluated ChatGPT’s ability to summarize 140 peer-reviewed abstracts from 14 journals. Physicians rated the quality, accuracy, and bias …

How A.I. Chatbots Become Political (Zvi Mowshowitz/New York Times)
https://www.nytimes.com/interactive/2024/03/28/opinion/ai-political-bias.html
http://www.memeorandum.com/240328/p32#a240328p32

"What determines the political preferences of your #AIchatbot?
Conservatives complain that many commercially available A.I. bots exhibit a persistent liberal bias. Elon Musk built Grok as an alternative language model after grumbling about ChatGPT being a “woke” A.I.
Liberals notice that A.I. output is often—in every sense—insufficiently diverse, because models learn from correlatio…

Opinion | Quiz: How Do Your Politics Stack Up Against ChatGPT’s?
We may soon rely a lot on A.I. chatbots, so keeping an eye on their political leanings is becoming more and more important.

New Study Finds NYTimes Fails to Quote Trans People in 66% of Stories About Anti-Trans Bills, by @… and
Media Matters for America https://www.readtpa.com/p/new-study…

New Study Finds NYT Fails to Quote Trans People in 66% of Stories About Anti-Trans Bills
The paper has been making a bunch of Journalism 101-level mistakes and has a major problem with anti-trans bias.

#NYTimesPitchbot
Via Matt McDermott @mattmfm
·
2h
According to the NYT, though Biden is leading his nearest opponent by 76% in the Michigan primary he “faced opposition.” Meanwhile, Trump is leading by 35% but “easily beat” his opponent.
The NYT bias in covering this election is just impossible to ignore, at this point.

Wir haben eine kleine Untersuchung zum Addition Bias auf Basis von DeReKoGram im IDS-Sprachreport veröffentlicht.
Dabei haben wir mehrere Gegenstandsbereiche (Größe, Höhe, Wichtigkeit, Wertigkeit ...) in den Blick genommen.
Paper:
https://ids-…

Great article. Great writing, thanks! https://blacktwitter.io/@CuriousScout/112174485630113991

FTBC: Forward Temporal Bias Correction for Optimizing ANN-SNN Conversion
Xiaofeng Wu, Velibor Bojkovic, Bin Gu, Kun Suo, Kai Zou
https://arxiv.org/abs/2403.18388

FTBC: Forward Temporal Bias Correction for Optimizing ANN-SNN Conversion
Spiking Neural Networks (SNNs) offer a promising avenue for energy-efficient computing compared with Artificial Neural Networks (ANNs), closely mirroring biological neural processes. However, this potential comes with inherent challenges in directly training SNNs through spatio-temporal backpropagation -- stemming from the temporal dynamics of spiking neurons and their discrete signal processing -- which necessitates alternative ways of training, most notably through ANN-SNN conversion. In this…

"The measure would require the regents to set up a department to hear grade challenges based on bias. That department... could determine there was bias and could order the faculty member to regrade the student’s assignment or reevaluate the student’s overall grade." h/t Kevin Reuning @reuning.bsky.social

Students could get grades changed over politics under Arizona Senate plan
Legislation approved by Senate would give university students a new avenue to appeal by alleging "political bias". Teachers could be ordered to change grades.

This https://arxiv.org/abs/2403.14633 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCY_…

Born With a Silver Spoon? Investigating Socioeconomic Bias in Large Language Models
Socioeconomic bias in society exacerbates disparities, influencing access to opportunities and resources based on individuals' economic and social backgrounds. This pervasive issue perpetuates systemic inequalities, hindering the pursuit of inclusive progress as a society. In this paper, we investigate the presence of socioeconomic bias, if any, in large language models. To this end, we introduce a novel dataset (SilverSpoon), consisting of 3000 samples that illustrate hypothetical scenarios th…

Beyond Self-learned Attention: Mitigating Attention Bias in Transformer-based Models Using Attention Guidance
Jiri Gesi, Iftekhar Ahmed
https://arxiv.org/abs/2402.16790

Beyond Self-learned Attention: Mitigating Attention Bias in Transformer-based Models Using Attention Guidance
Transformer-based models have demonstrated considerable potential for source code modeling tasks in software engineering. However, they are limited by their dependence solely on automatic self-attention weight learning mechanisms. Previous studies have shown that these models overemphasize delimiters added by tokenizers (e.g., [CLS], [SEP]), which may lead to overlooking essential information in the original input source code. To address this challenge, we introduce SyntaGuid, a novel approach …

"What determines the political preferences of your #AIchatbot?
Conservatives complain that many commercially available A.I. bots exhibit a persistent liberal bias. Elon Musk built Grok as an alternative language model after grumbling about ChatGPT being a “woke” A.I.
Liberals notice that A.I. output is often—in every sense—insufficiently diverse, because models learn from correlatio…

Opinion | Quiz: How Do Your Politics Stack Up Against ChatGPT’s?
We may soon rely a lot on A.I. chatbots, so keeping an eye on their political leanings is becoming more and more important.

Doubly robust causal inference through penalized bias-reduced estimation: combining non-probability samples with designed surveys
Jiacong Du, Xu Shi, Donglin Zeng, Bhramar Mukherjee
https://arxiv.org/abs/2403.18039

Doubly robust causal inference through penalized bias-reduced estimation: combining non-probability samples with designed surveys
Causal inference on the average treatment effect (ATE) using non-probability samples, such as electronic health records (EHR), faces challenges from sample selection bias and high-dimensional covariates. This requires considering a selection model alongside treatment and outcome models that are typical ingredients in causal inference. This paper considers integrating large non-probability samples with external probability samples from a design survey, addressing moderately high-dimensional conf…

This https://arxiv.org/abs/2402.16458 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCL_…

D-XCB: Data-independent Debiasing for Fair and Accurate Transformer-based Cyberbullying Detection
Swear words are a common proxy to collect datasets with cyberbullying incidents. Our focus is on measuring and mitigating biases derived from spurious associations between swear words and incidents occurring as a result of such data collection strategies. After demonstrating and quantifying these biases, we introduce ID-XCB, the first data-independent debiasing technique that combines adversarial training, bias constraints and debias fine-tuning approach aimed at alleviating model attention to …

PureLottery: Fair and Bias-Resistant Leader Election with a Novel Single-Elimination Tournament Algorithm
Jonas Ballweg
https://arxiv.org/abs/2402.17459 ht…

PureLottery: Fair and Bias-Resistant Leader Election with a Novel Single-Elimination Tournament Algorithm
Leader Election (LE) is crucial in distributed systems and blockchain technology, ensuring one participant acts as the leader. Traditional LE methods often depend on distributed random number generation (RNG), facing issues like vulnerability to manipulation, lack of fairness, and the need for complex procedures such as verifiable delay functions (VDFs) and publicly-verifiable secret sharing (PVSS). This Bachelor's thesis presents a novel approach to randomized LE, leveraging a game-theoretic a…

@… @… besonders darf man bei evolution nie vergessen das dies immer auf die Masse statt dem Individuum betrachtet wird und damit auch immer survivalship bias mitschwingt. Die Menschen die gescheitert sind werden unsichtbar. Da schwingt be…

This https://arxiv.org/abs/2309.11599 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csHC_…

The Role of Inclusion, Control, and Ownership in Workplace AI-Mediated Communication
Given large language models' (LLMs) increasing integration into workplace software, it is important to examine how biases in the models may impact workers. For example, stylistic biases in the language suggested by LLMs may cause feelings of alienation and result in increased labor for individuals or groups whose style does not match. We examine how such writer-style bias impacts inclusion, control, and ownership over the work when co-writing with LLMs. In an online experiment, participants wro…

Clustering Change Sign Detection by Fusing Mixture Complexity
Kento Urano, Ryo Yuki, Kenji Yamanishi
https://arxiv.org/abs/2403.18269 https://

Clustering Change Sign Detection by Fusing Mixture Complexity
This paper proposes an early detection method for cluster structural changes. Cluster structure refers to discrete structural characteristics, such as the number of clusters, when data are represented using finite mixture models, such as Gaussian mixture models. We focused on scenarios in which the cluster structure gradually changed over time. For finite mixture models, the concept of mixture complexity (MC) measures the continuous cluster size by considering the cluster proportion bias and ov…

Human self-destruction is no-doubt part of 'intelligent design', so beloved of the religious. 😉
https://www.newscientist.com/article/mg22329820-200-understand-faulty-thinking-to-tackle-climate-change/

Understand faulty thinking to tackle climate change
The amorphous nature of climate change creates the ideal conditions for human denial and cognitive bias to come to the fore, says George Marshall

I’ll just take this news of ShotSpotter being a creepy police-state surveillance apparatus that operates outside of normal democratic oversight mechanisms as an opportunity to remind everyone that is ShotSpotter is also a mechanism for whitewashing police bias with statistics by taking a map of where police put microphones and presenting it as a supposed map of where gun violence occurs.

This https://arxiv.org/abs/2312.13963 has been replaced.
initial toot: https://mastoxiv.page/@…

Biased motility-induced phase separation: from chemotaxis to traffic jams
We propose a one-dimensional model of active particles interpolating between quorum sensing models used in the study of motility-induced phase separation (MIPS) and models of congestion of traffic flow on a single-lane highway. Particles have a target velocity with a density-dependent magnitude and a direction that flips with a finite rate that is biased toward moving right. Two key parameters are the bias and the speed relaxation time. MIPS is known to occur in such models at zero bias and zer…

Cognitively Biased Users Interacting with Algorithmically Biased Results in Whole-Session Search on Controversial Topics
Ben Wang, Jiqun Liu
https://arxiv.org/abs/2403.17286

Cognitively Biased Users Interacting with Algorithmically Biased Results in Whole-Session Search on Controversial Topics
When interacting with information retrieval (IR) systems, users, affected by confirmation biases, tend to select search results that confirm their existing beliefs on socially significant contentious issues. To understand the judgments and attitude changes of users searching online, our study examined how cognitively biased users interact with algorithmically biased search engine result pages (SERPs). We designed three-query search sessions on debated topics under various bias conditions. We re…

Points defects produced by irradiation: influence of boundary conditions on their biased elimination
Jean-louis Bocquet
https://arxiv.org/abs/2403.17569 ht…

Points defects produced by irradiation: influence of boundary conditions on their biased elimination
Most of the sinks able to absorb the point defects created under irradiation exhibit a stronger elastic interaction with the interstitial defect I than with the vacancy defect V. This bias of their elastic interaction (EB) is supposed to automatically determine the concomitant absorption bias (AB) leading to an elimination flux of I always larger than the flux of V. We show in this contribution that the link between the EB and the AB is not as rigid as thought previously. On a model system we c…

This https://arxiv.org/abs/2312.11468 has been replaced.
initial toot: https://mastoxiv.page/@arX…

Bias-Reduced Neural Networks for Parameter Estimation in Quantitative MRI
Purpose: To develop neural network (NN)-based quantitative MRI parameter estimators with minimal bias and a variance close to the Cramér-Rao bound. Theory and Methods: We generalize the mean squared error loss to control the bias and variance of the NN's estimates, which involves averaging over multiple noise realizations of the same measurements during training. Bias and variance properties of the resulting NNs are studied for two neuroimaging applications. Results: In simulations, the pr…

Saw this post on the origin on the red bullet hole diagram often used in survivorship bias discussions, and wanted to pass it along: http://www.cameronmoll.com/journal/abraham-wald-red-bullet-holes-origin-story

This https://arxiv.org/abs/2306.11574 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_nuc…

Two-particle Bose-Einstein correlations and their Lévy parameters in PbPb collisions at $\sqrt{s_\mathrm{NN}}$ = 5.02 TeV
Two-particle Bose-Einstein momentum correlation functions are studied for charged-hadron pairs in lead-lead collisions at a center-of-mass energy per nucleon pair of $\sqrt{s_\mathrm{NN}}$ = 5.02 TeV. The data sample, containing 4.27 $\times$ $10^{9}$ minimum bias events corresponding to an integrated luminosity of 0.607 nb$^{-1}$, was collected by the CMS experiment in 2018. The experimental results are discussed in terms of a Lévy-type source distribution. The parameters of this distribution…

CosmoMIA: Cosmic Web-based redshift space halo distribution
Daniel Forero-S\'anchez, Francisco-Shu Kitaura, Francesco Sinigaglia, Jose Mar\'ia Coloma-Nodal, Jean-Paul Kneib
https://arxiv.org/abs/2402.17581

CosmoMIA: Cosmic Web-based redshift space halo distribution
Modern galaxy surveys demand extensive survey volumes and resolutions surpassing current dark matter-only simulations' capabilities. To address this, many methods employ effective bias models on the dark matter field to approximate object counts on a grid. However, realistic catalogs necessitate specific coordinates and velocities for a comprehensive understanding of the Universe. In this research, we explore sub-grid modeling to create accurate catalogs, beginning with coarse grid number count…

When is an Arctic bias not an Arctic bias?
I was going to blog about this cool new paper that my colleagues at DMI have produced, but John Kennedy has as always done such a good job I will just point you over there... Wondering whether a warm bias in the Arctic in ERA5 affects our estimates of global temperature change. When is an Arctic bias not an Arctic bias?

Investigating Bias in LLM-Based Bias Detection: Disparities between LLMs and Human Perception
Luyang Lin, Lingzhi Wang, Jinsong Guo, Kam-Fai Wong
https://arxiv.org/abs/2403.14896 …

Investigating Bias in LLM-Based Bias Detection: Disparities between LLMs and Human Perception
The pervasive spread of misinformation and disinformation in social media underscores the critical importance of detecting media bias. While robust Large Language Models (LLMs) have emerged as foundational tools for bias prediction, concerns about inherent biases within these models persist. In this work, we investigate the presence and nature of bias within LLMs and its consequential impact on media bias detection. Departing from conventional approaches that focus solely on bias detection in m…

Ofcom warns UK broadcasters on the importance of impartiality ahead of the general election, but stops short of outright banning using politicians as presenters (Daniel Thomas/Financial Times)
https://t.co/ksNyNSpZKP

An ensemble of data-driven weather prediction models for operational sub-seasonal forecasting
Jonathan A. Weyn, Divya Kumar, Jeremy Berman, Najeeb Kazmi, Sylwester Klocek, Pete Luferenko, Kit Thambiratnam
https://arxiv.org/abs/2403.15598

An ensemble of data-driven weather prediction models for operational sub-seasonal forecasting
We present an operations-ready multi-model ensemble weather forecasting system which uses hybrid data-driven weather prediction models coupled with the European Centre for Medium-range Weather Forecasts (ECMWF) ocean model to predict global weather at 1-degree resolution for 4 weeks of lead time. For predictions of 2-meter temperature, our ensemble on average outperforms the raw ECMWF extended-range ensemble by 4-17%, depending on the lead time. However, after applying statistical bias correcti…

Emotional Voice Messages (EMOVOME) database: emotion recognition in spontaneous voice messages
Luc\'ia G\'omez Zaragoz\'aHUMAN-tech Institute, Universitat Polit\`enica de Val\`encia, Valencia, Spain, Roc\'io del AmorHUMAN-tech Institute, Universitat Polit\`enica de Val\`encia, Valencia, Spain, Elena Parra VargasHUMAN-tech Institute, Universitat Polit\`enica de Val\`encia, Valencia, Spain, Valery NaranjoHUMAN-tech Institute, Universitat Polit\`enica de Val\`encia, Valenc…

Emotional Voice Messages (EMOVOME) database: emotion recognition in spontaneous voice messages
Emotional Voice Messages (EMOVOME) is a spontaneous speech dataset containing 999 audio messages from real conversations on a messaging app from 100 Spanish speakers, gender balanced. Voice messages were produced in-the-wild conditions before participants were recruited, avoiding any conscious bias due to laboratory environment. Audios were labeled in valence and arousal dimensions by three non-experts and two experts, which were then combined to obtain a final label per dimension. The experts …

This https://arxiv.org/abs/2402.11621 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCL_…

Decoding News Narratives: A Critical Analysis of Large Language Models in Framing Bias Detection
This work contributes to the expanding research on the applicability of LLMs in social sciences by examining the performance of GPT-3.5 Turbo, GPT-4, and Flan-T5 models in detecting framing bias in news headlines through zero-shot, few-shot, and explainable prompting methods. A key insight from our evaluation is the notable efficacy of explainable prompting in enhancing the reliability of these models, highlighting the importance of explainable settings for social science research on framing bi…

This https://arxiv.org/abs/2212.00199 has been replaced.
link: https://scholar.google.com/scholar?q=a

Evidence of $ϕ$0-Josephson junction from skewed diffraction patterns in Sn-InSb nanowires
We study Josephson junctions based on InSb nanowires with Sn shells. We observe skewed critical current diffraction patterns: the maxima in forward and reverse current bias are at different magnetic flux, with a displacement of 20-40 mT. The skew is greatest when the external field is nearly perpendicular to the nanowire, in the substrate plane. This orientation suggests that spin-orbit interaction plays a role. We develop a phenomenological model and perform tight-binding calculations, both me…

Österreichs AMS-Chatbot teurer – aber weniger Bias
Der Berufsinfomat des Arbeitsmarktservice in Österreich ist verbessert worden. Dafür kostet der Betrieb auch deutlich mehr als geplant war.
https://www.…

Quasi-Bayesian Estimation and Inference with Control Functions
Ruixuan Liu, Zhengfei Yu
https://arxiv.org/abs/2402.17374 https://arxi…

Quasi-Bayesian Estimation and Inference with Control Functions
We consider a quasi-Bayesian method that combines a frequentist estimation in the first stage and a Bayesian estimation/inference approach in the second stage. The study is motivated by structural discrete choice models that use the control function methodology to correct for endogeneity bias. In this scenario, the first stage estimates the control function using some frequentist parametric or nonparametric approach. The structural equation in the second stage, associated with certain complicat…

This https://arxiv.org/abs/2203.12572 has been replaced.
link: https://scholar.google.com/scholar?q=a

Post-selection inference for e-value based confidence intervals
Suppose that one can construct a valid $(1-δ)$-confidence interval (CI) for each of $K$ parameters of potential interest. If a data analyst uses an arbitrary data-dependent criterion to select some subset $S$ of parameters, then the aforementioned CIs for the selected parameters are no longer valid due to selection bias. We design a new method to adjust the intervals in order to control the false coverage rate (FCR). The main established method is the "BY procedure" by Benjamini and Yekutieli …

Substrate Scope Contrastive Learning: Repurposing Human Bias to Learn Atomic Representations
Wenhao Gao, Priyanka Raghavan, Ron Shprints, Connor W. Coley
https://arxiv.org/abs/2402.16882

Substrate Scope Contrastive Learning: Repurposing Human Bias to Learn Atomic Representations
Learning molecular representation is a critical step in molecular machine learning that significantly influences modeling success, particularly in data-scarce situations. The concept of broadly pre-training neural networks has advanced fields such as computer vision, natural language processing, and protein engineering. However, similar approaches for small organic molecules have not achieved comparable success. In this work, we introduce a novel pre-training strategy, substrate scope contrasti…

"Let’s pause here. What would happen if the #NYTimes suddenly hired a #Palestinian filmmaker with no journalistic background, who had recently publicly “liked” posts that called for “pushing Israeli Jews into the sea,” to co-write several of its most sensitive and contested reports?"

Extraordinary charges of bias emerge against NYTimes reporter Anat Schwartz
New doubts are emerging about the New York Times’s coverage of sexual violence in the October 7 attack. The paper must explain why it broke its own rules by hiring a clearly biased writer who endorsed racist and violent rhetoric toward Palestinians.

This https://arxiv.org/abs/2401.07415 has been replaced.
initial toot: https://mastoxiv.page/@arX…

Alternating Bias Assisted Annealing of Amorphous Oxide Tunnel Junctions
We demonstrate a transformational technique for controllably tuning the electrical properties of fabricated thermally oxidized amorphous aluminum-oxide tunnel junctions. Using conventional test equipment to apply an alternating bias to a heated tunnel barrier, giant increases in the room temperature resistance, greater than 70%, can be achieved. The rate of resistance change is shown to be strongly temperature-dependent, and is independent of junction size in the sub-micron regime. In order to …

This https://arxiv.org/abs/2402.13391 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_sta…

De-Biasing the Bias: Methods for Improving Disparity Assessments with Noisy Group Measurements
Health care decisions are increasingly informed by clinical decision support algorithms, but these algorithms may perpetuate or increase racial and ethnic disparities in access to and quality of health care. Further complicating the problem, clinical data often have missing or poor quality racial and ethnic information, which can lead to misleading assessments of algorithmic bias. We present novel statistical methods that allow for the use of probabilities of racial/ethnic group membership in a…

Y’a reckon? 🙄 Walter Sofronoff’s ‘private and secret’ texts with Janet Albrechtsen could lead to impression of bias, court told
Bruce Lehrmann inquiry head’s 273 interactions with the Australian columnist raised in legal challenge by former ACT DPP
h…

Theory of tunneling spectroscopy in $p$-wave altermagnet-superconductor hybrid structures
Kazuki Maeda, Bo Lu, Keiji Yada, Yukio Tanaka
https://arxiv.org/abs/2403.17482

Theory of tunneling spectroscopy in $p$-wave altermagnet-superconductor hybrid structures
We theoretically study the tunneling conductance of a junction consisting of a two-dimensional $p$-wave altermagnet (AM) and a superconductor (SC) for various pairing states. The zero bias conductance peaks arising from the dispersionless surface Andreev bound states (SABSs) in $d_{xy}$-wave and $p_{x}$-wave superconductor junctions are robust against varying the altermagnetic spin-splitting strength $α_{y}$. Moreover, for chiral $p$- or chiral $d$-wave SCs, zero bias conductance shows a non-m…

Debiasing Text-to-Image Diffusion Models
Ruifei He, Chuhui Xue, Haoru Tan, Wenqing Zhang, Yingchen Yu, Song Bai, Xiaojuan Qi
https://arxiv.org/abs/2402.14577

Debiasing Text-to-Image Diffusion Models
Learning-based Text-to-Image (TTI) models like Stable Diffusion have revolutionized the way visual content is generated in various domains. However, recent research has shown that nonnegligible social bias exists in current state-of-the-art TTI systems, which raises important concerns. In this work, we target resolving the social bias in TTI diffusion models. We begin by formalizing the problem setting and use the text descriptions of bias groups to establish an unsafe direction for guiding the…

UNESCO's nvestigation into Gender Bias in Large Language Models: 'LLMs generated negative content about gay subjects in approximately 70% of instances for Llama 2 and in approximately 60% of instances for GPT-2. Finally, in generating content using prompts which intersect gender and culture with occupation, the results highlight a clear bias in AI-generated content, showing a tendency to assign more diverse and professional jobs to men (teacher, doctor, driver), while often re…

Gradient-enhanced deep Gaussian processes for multifidelity modelling
Viv Bone, Chris van der Heide, Kieran Mackle, Ingo H. J. Jahn, Peter M. Dower, Chris Manzie
https://arxiv.org/abs/2402.16059

Gradient-enhanced deep Gaussian processes for multifidelity modelling
Multifidelity models integrate data from multiple sources to produce a single approximator for the underlying process. Dense low-fidelity samples are used to reduce interpolation error, while sparse high-fidelity samples are used to compensate for bias or noise in the low-fidelity samples. Deep Gaussian processes (GPs) are attractive for multifidelity modelling as they are non-parametric, robust to overfitting, perform well for small datasets, and, critically, can capture nonlinear and input-de…

"Künstliche Intelligenz in Bibliotheken und der Hype um Large Language Models – trojanische Sau oder willkommenes Arbeitspferd": großartige Präsentationsfolien von @… für alle, die mal eben ein fundiertes Update zum Thema durchklicken wollen:

Efficient Unbiased Sparsification
Leighton Barnes, Timothy Chow, Emma Cohen, Keith Frankston, Benjamin Howard, Fred Kochman, Daniel Scheinerman, Jeffrey VanderKam
https://arxiv.org/abs/2402.14925

Efficient Unbiased Sparsification
An unbiased $m$-sparsification of a vector $p\in \mathbb{R}^n$ is a random vector $Q\in \mathbb{R}^n$ with mean $p$ that has at most $m

This https://arxiv.org/abs/2402.13787 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csSI_…

Fairness Rising from the Ranks: HITS and PageRank on Homophilic Networks
In this paper, we investigate the conditions under which link analysis algorithms prevent minority groups from reaching high ranking slots. We find that the most common link-based algorithms using centrality metrics, such as PageRank and HITS, can reproduce and even amplify bias against minority groups in networks. Yet, their behavior differs: one one hand, we empirically show that PageRank mirrors the degree distribution for most of the ranking positions and it can equalize representation of m…

"1. Enhanced structured assessments with cognitive testing"
This is basically "racism with a side of logic puzzles"
"Over the past 12 months, we have tested structured assessments for cultural alignment and technical skills, and validated these were predictive of performance at Coinbase"
Measuring performance is itself a can of worms and racism magnet. In order words: garbage in, garbage out. They're validating that they get what they measure, but if both are biased in the same way, well, probably they aren't learning what they think they are.
"We believe having a minimum bar on these assessments will drive consistency in evaluation, reduce bias, and foster a meritocracy."
Consistency? Probably. Consistency of goose turds.
Reduce bias? Certainly the opposite. They're literally creating bias via cultural alignment. That's the whole fucking point of cultural alignment: it's intentional bias.
Oh, did they mean reducing discrimination against protected classes? LOL. also no. Probably they never learned that IQ tests are racist, and what they're doing is probably worse.
"Foster a meritocracy" what is this, a parody of 2005 libertarian tech start-ups? … … oh right, griftocurrency.
Apparently they were sick the day that their grade school teacher presented why meritocracy is a myth
https://en.m.wikipedia.org/wiki/Myth_of_meritocracy

This https://arxiv.org/abs/2310.15950 has been replaced.
link: https://scholar.google.com/scholar?q=a

Representation Learning with Large Language Models for Recommendation
Recommender systems have seen significant advancements with the influence of deep learning and graph neural networks, particularly in capturing complex user-item relationships. However, these graph-based recommenders heavily depend on ID-based data, potentially disregarding valuable textual information associated with users and items, resulting in less informative learned representations. Moreover, the utilization of implicit feedback data introduces potential noise and bias, posing challenges …

My take on the Google #AI image generator controversy, and bias in AI image generation in general:
When you ask for a generative creation (including LLMs) you WANT the AI to fill in a ton of details. Some of those are fine: backgrounds, textures, expressions, etc. Some can intersect with societal issues like race or gender.
LLMs actually can be vague –a benefit of language – but image genera…

😿 A Psychologist Explains Four Reasons the Internet Feels So Broken
https://www.theringer.com/2024/4/9/24124973/psychologist-jay-van-bavel-explains-four-reasons-the-internet-feels-broken-negativ…

A Psychologist Explains Four Reasons the Internet Feels So Broken
Jay Van Bavel of New York University joins Derek to explain the group psychology dynamics behind the negativity and tribalism often seen online

We know the world is biased, the Internet even more so. Training #AI on biased data naturally leads to biased AI models.
Trying to remove these biases is laudable. But trying to remove it superficially (after all these biased steps) is hard, maybe even impossible.
We need to remove them everywhere to succeed.

Europe’s AI Act demands extensive "logs" of users - targets biometrics, bias, black boxes
https://www.thestack.technology/bias-biometrics-and-black-boxes-europes-ai-act-what-you-need-to-know/

Optimization of foreground moment deprojection for semi-blind CMB polarization reconstruction
Alessandro Carones, Mathieu Remazeilles
https://arxiv.org/abs/2402.17579

Optimization of foreground moment deprojection for semi-blind CMB polarization reconstruction
Upcoming Cosmic Microwave Background (CMB) experiments, aimed at measuring primordial CMB B-modes, require exquisite control of Galactic foreground contamination. Minimum-variance techniques, like the Needlet Internal Linear Combination (NILC), have proven effective in reconstructing the CMB polarization signal and mitigating foregrounds across diverse sky models without suffering from mismodelling errors. Still, residual contamination may bias the recovered CMB polarization at large angular sc…

Could the Book of Genesis help address gerrymandering?
Partly inspired by a Biblical method of resource allocation, a new paper co-authored by Yale political scientist Kevin DeLuca offers a novel way to reduce partisan bias in legislative districting.
https://…

This https://arxiv.org/abs/2403.16961 has been replaced.
initial toot: https://mastoxiv.page/@arXi…

Biased ensembles of pulsating active matter
We discover unexpected connections between packing configurations and rare fluctuations in dense systems of active particles subject to pulsation of size. Using large deviation theory, we examine biased ensembles which select atypical realizations of the dynamics exhibiting high synchronization in particle size. We show that the order emerging at high bias can manifest as distinct dynamical states with either a finite or a vanishing size current. Remarkably, transitions between these states ari…

And this list doesn’t even get to #AIsplaining, which is sure to be fun for everyone, probably by tomorrow afternoon.
From: @…

This https://arxiv.org/abs/2306.17667 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_…

Bias-Free Estimation of Signals on Top of Unknown Backgrounds
We present a method for obtaining unbiased signal estimates in the presence of a significant unknown background, eliminating the need for a parametric model for the background itself. Our approach is based on a minimal set of conditions for observation and background estimators, which are typically satisfied in practical scenarios. To showcase the effectiveness of our method, we apply it to simulated data from the planned dielectric axion haloscope MADMAX.

The Brains Blog || Wu, Movements of Mind. Post 4: Biased Attention as Implicit, Automatic Bias.
https://philosophyofbrains.com/2024/03/14/wu-movements-of-mind-post-4-biased-attention-as-implicit-automatic-bias.aspx

Gene Regulatory Network Inference in the Presence of Dropouts: a Causal View
Haoyue Dai, Ignavier Ng, Gongxu Luo, Peter Spirtes, Petar Stojanov, Kun Zhang
https://arxiv.org/abs/2403.15500

Gene Regulatory Network Inference in the Presence of Dropouts: a Causal View
Gene regulatory network inference (GRNI) is a challenging problem, particularly owing to the presence of zeros in single-cell RNA sequencing data: some are biological zeros representing no gene expression, while some others are technical zeros arising from the sequencing procedure (aka dropouts), which may bias GRNI by distorting the joint distribution of the measured gene expressions. Existing approaches typically handle dropout error via imputation, which may introduce spurious relations as t…

Network analysis of memristive device circuits: dynamics, stability and correlations
Frank Barrows, Forrest C. Sheldon, Francesco Caravelli
https://arxiv.org/abs/2402.16015

Network analysis of memristive device circuits: dynamics, stability and correlations
Networks with memristive devices are a potential basis for the next generation of computing devices. They are also an important model system for basic science, from modeling nanoscale conductivity to providing insight into the information-processing of neurons. The resistance in a memristive device depends on the history of the applied bias and thus displays a type of memory. The interplay of this memory with the dynamic properties of the network can give rise to new behavior, offering many fas…

This https://arxiv.org/abs/2312.15105 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_mat…

The friendship paradox for sparse random graphs
Let $G_n$ be an undirected finite graph on $n\in\mathbb{N}$ vertices labelled by $[n] = \{1,\ldots,n\}$. For $i \in [n]$, let $Δ_{i,n}$ be the friendship bias of vertex $i$, defined as the difference between the average degree of the neighbours of vertex $i$ and the degree of vertex $i$ itself when $i$ is not isolated, and zero when $i$ is isolated. Let $μ_n$ denote the friendship-bias empirical distribution, i.e., the measure that puts mass $\frac{1}{n}$ at each $Δ_{i,n}$, $i \in [n]$. The …

Assessing COVID-19 Vaccine Effectiveness in Observational Studies via Nested Trial Emulation
Justin B. DeMonte, Bonnie E. Shook-Sa, Michael G. Hudgens
https://arxiv.org/abs/2403.18115

Assessing COVID-19 Vaccine Effectiveness in Observational Studies via Nested Trial Emulation
Observational data are often used to estimate real-world effectiveness and durability of coronavirus disease 2019 (COVID-19) vaccines. A sequence of nested trials can be emulated to draw inference from such data while minimizing selection bias, immortal time bias, and confounding. Typically, when nested trial emulation (NTE) is employed, effect estimates are pooled across trials to increase statistical efficiency. However, such pooled estimates may lack a clear interpretation when the treatment…

This is corrupt jury manipulation.
Fox covered Juror 2, Trump retweeted, and now she is OFF THE JURY.
Trump is manipulating the jury. He’s pushing off jurors who are less likely to be in the tank for him. This creates pro-Trump jury bias. /1
https://www.thegu…

“We can’t possibly accept a juror who has bias after watching the criminal do crimes for 40 years and not getting prosecuted for it”

> She worries the negative effects will spread as the technology does. "One biased human hiring manager can harm a lot of people in a year, and that's not great," she says. "But an algorithm that is maybe used in all incoming applications at a large company… that could harm hundreds of thousands of applicants."

AI hiring tools may be filtering out the best job applicants
As firms increasingly rely on artificial intelligence-driven hiring platforms, many highly qualified candidates are finding themselves on the cutting room floor.

Bill Maher: No, no criminal charges, but then he spent 300 pages calling Biden Mr. Magoo, basically. He’s, he’s, this is the words from the report. That Biden’s a well meaning elderly man with diminished faculties and advancing age. Come on. Well, I don’t want to say this thing is loaded, but the final line of the report was, I’m Donald Trump and I approve this message.

Bill Maher Calls Out Robert Hur's Bias In A Single Joke
Bill Maher called the Hur Report loaded and called out the Special Counsel report's bias in his Real Time monologue.

Gaza: language bias at the New York Times
Description of language instructions, advising against "occupation", "refugee camps" and even "Palestine".
"The analysis found that, as of November 24, the New York Times had described Israeli deaths as a “massacre” on 53 occasions and those of Palestinians just once. The ratio for the use of “slaughter” was 22 to 1, even as the documented number of Palestinians killed climbed to around 15,000."
#Gaza #language #NewYorkTimes

This https://arxiv.org/abs/2312.14461 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCR_…

Attacking Byzantine Robust Aggregation in High Dimensions
Training modern neural networks or models typically requires averaging over a sample of high-dimensional vectors. Poisoning attacks can skew or bias the average vectors used to train the model, forcing the model to learn specific patterns or avoid learning anything useful. Byzantine robust aggregation is a principled algorithmic defense against such biasing. Robust aggregators can bound the maximum bias in computing centrality statistics, such as mean, even when some fraction of inputs are arbi…

This https://arxiv.org/abs/2011.14747 has been replaced.
link: https://scholar.google.com/scholar?q=a

Asymptotic bias reduction of maximum likelihood estimates via penalized likelihoods with differential geometry
A procedure for asymptotic bias reduction of maximum likelihood estimates of generic estimands is developed. The estimator is realized as a plug-in estimator, where the parameter maximizes the penalized likelihood with a penalty function that satisfies a quasi-linear partial differential equation of the first order. The integration of the partial differential equation with the aid of differential geometry is discussed. Applications to generalized linear models, linear mixed-effects models, and …

Guided By AI: Navigating Trust, Bias, and Data Exploration in AI-Guided Visual Analytics
Sunwoo Ha, Shayan Monadjemi, Alvitta Ottley
https://arxiv.org/abs/2404.14521

Guided By AI: Navigating Trust, Bias, and Data Exploration in AI-Guided Visual Analytics
The increasing integration of artificial intelligence (AI) in visual analytics (VA) tools raises vital questions about the behavior of users, their trust, and the potential of induced biases when provided with guidance during data exploration. We present an experiment where participants engaged in a visual data exploration task while receiving intelligent suggestions supplemented with four different transparency levels. We also modulated the difficulty of the task (easy or hard) to simulate a m…

Magnetocapacitance oscillations dominated by giant Rashba spin orbit interaction in InAs/GaSb quantum wells separated by AlSb barrier
A. S. L. Ribeiro, R. Schott, C. Reichl, W. Dietsche, W. Wegscheider
https://arxiv.org/abs/2404.16419

Magnetocapacitance oscillations dominated by giant Rashba spin orbit interaction in InAs/GaSb quantum wells separated by AlSb barrier
We observed magnetocapacitance oscillations in InAs/GaSb quantum wells separated by a $20$\,nm AlSb middle barrier. By realizing independent ohmic contacts for electrons in InAs and holes in the GaSb layer, we found an out-of-plane oscillatory response in capacitance representing the density of states of this system. We were able to tune the charge carrier densities by applying a DC bias voltage, identifying the formation of beating signatures for forward bias. The coexistence of two distinguis…

Protected group bias and stereotypes in Large Language Models
Hadas Kotek, David Q. Sun, Zidi Xiu, Margit Bowler, Christopher Klein
https://arxiv.org/abs/2403.14727

Protected group bias and stereotypes in Large Language Models
As modern Large Language Models (LLMs) shatter many state-of-the-art benchmarks in a variety of domains, this paper investigates their behavior in the domains of ethics and fairness, focusing on protected group bias. We conduct a two-part study: first, we solicit sentence continuations describing the occupations of individuals from different protected groups, including gender, sexuality, religion, and race. Second, we have the model generate stories about individuals who hold different types of…

We know the world is biased, the Internet even more so. Training #AI on biased data naturally leads to biased AI models.
Trying to remove these biases is laudable. But trying to remove it superficially (after all these biased steps) is hard, maybe even impossible.
We need to remove them everywhere to succeed.

This https://arxiv.org/abs/2309.10978 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_sta…

Negative Spillover: A Potential Source of Bias in Pragmatic Clinical Trials
Pragmatic clinical trials evaluate the effectiveness of health interventions in real-world settings. Negative spillover can arise in a pragmatic trial if the study intervention affects how scarce resources are allocated between patients in the intervention and comparison groups. This can harm patients assigned to the control group and lead to overestimation of treatment effect. While this type of negative spillover is often addressed in trials of social welfare and public health interventions, …

Locating and Mitigating Gender Bias in Large Language Models
Yuchen Cai, Ding Cao, Rongxi Guo, Yaqin Wen, Guiquan Liu, Enhong Chen
https://arxiv.org/abs/2403.14409

Locating and Mitigating Gender Bias in Large Language Models
Large language models(LLM) are pre-trained on extensive corpora to learn facts and human cognition which contain human preferences. However, this process can inadvertently lead to these models acquiring biases and stereotypes prevalent in society. Prior research has typically tackled the issue of bias through a one-dimensional perspective, concentrating either on locating or mitigating it. This limited perspective has created obstacles in facilitating research on bias to synergistically complem…

IndiTag: An Online Media Bias Analysis and Annotation System Using Fine-Grained Bias Indicators
Luyang Lin, Lingzhi Wang, Jinsong Guo, Jing Li, Kam-Fai Wong
https://arxiv.org/abs/2403.13446

IndiTag: An Online Media Bias Analysis and Annotation System Using Fine-Grained Bias Indicators
In the age of information overload and polarized discourse, understanding media bias has become imperative for informed decision-making and fostering a balanced public discourse. This paper presents IndiTag, an innovative online media bias analysis and annotation system that leverages fine-grained bias indicators to dissect and annotate bias in digital content. IndiTag offers a novel approach by incorporating large language models, bias indicator, vector database to automatically detect and int…

Advancing Recommender Systems by mitigating Shilling attacks
Aditya Chichani, Juzer Golwala, Tejas Gundecha, Kiran Gawande
https://arxiv.org/abs/2404.16177

Advancing Recommender Systems by mitigating Shilling attacks
Considering the premise that the number of products offered grow in an exponential fashion and the amount of data that a user can assimilate before making a decision is relatively small, recommender systems help in categorizing content according to user preferences. Collaborative filtering is a widely used method for computing recommendations due to its good performance. But, this method makes the system vulnerable to attacks which try to bias the recommendations. These attacks, known as 'shill…

This https://arxiv.org/abs/2206.11117 has been replaced.
link: https://scholar.google.com/scholar?q=a

Causal inference in multi-cohort studies using the target trial framework to identify and minimize sources of bias
Longitudinal cohort studies, which follow a group of individuals over time, provide the opportunity to examine causal effects of complex exposures on long-term health outcomes. Utilizing data from multiple cohorts has the potential to add further benefit by improving precision of estimates through data pooling and by allowing examination of effect heterogeneity through replication of analyses across cohorts. However, the interpretation of findings can be complicated by biases that may be compou…

Magnetocapacitance oscillations dominated by giant Rashba spin orbit interaction in InAs/GaSb quantum wells separated by AlSb barrier
A. S. L. Ribeiro, R. Schott, C. Reichl, W. Dietsche, W. Wegscheider
https://arxiv.org/abs/2404.16419

Magnetocapacitance oscillations dominated by giant Rashba spin orbit interaction in InAs/GaSb quantum wells separated by AlSb barrier
We observed magnetocapacitance oscillations in InAs/GaSb quantum wells separated by a $20$\,nm AlSb middle barrier. By realizing independent ohmic contacts for electrons in InAs and holes in the GaSb layer, we found an out-of-plane oscillatory response in capacitance representing the density of states of this system. We were able to tune the charge carrier densities by applying a DC bias voltage, identifying the formation of beating signatures for forward bias. The coexistence of two distinguis…

Uri Berliner's essay about NPR reignited criticism of the outlet on social media, especially among conservatives, and vehement pushback from some staffers (New York Times)
https://www.nytimes.com/2024/04/11/business/media/npr-criticism-liberal…

NPR in Turmoil After It Is Accused of Liberal Bias
An essay from an editor at the broadcaster has generated a firestorm of criticism about the network on social media, especially among conservatives.

This https://arxiv.org/abs/2307.00567 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_sta…

A Note on Ising Network Analysis with Missing Data
The Ising model has become a popular psychometric model for analyzing item response data. The statistical inference of the Ising model is typically carried out via a pseudo-likelihood, as the standard likelihood approach suffers from a high computational cost when there are many variables (i.e., items). Unfortunately, the presence of missing values can hinder the use of pseudo-likelihood, and a listwise deletion approach for missing data treatment may introduce a substantial bias into the estim…

This https://arxiv.org/abs/2311.07650 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_…

Elucidating the impact of massive neutrinos on halo assembly bias
Massive neutrinos have non-negligible impact on the formation of large-scale structures. We investigate the impact of massive neutrinos on the halo assembly bias effect, measured by the relative halo bias $\hat{b}$ as a function of the curvature of the initial density peak $\hat{s}$, neutrino excess $ε_ν$, or halo concentration $\hat{c}$, using a large suite of $ΣM_ν{=}0.0$ eV and $0.4$ eV simulations with the same initial conditions. By tracing dark matter haloes back to their initial dens…

Understanding Optimal Feature Transfer via a Fine-Grained Bias-Variance Analysis
Yufan Li, Subhabrata Sen, Ben Adlam
https://arxiv.org/abs/2404.12481 https…

Understanding Optimal Feature Transfer via a Fine-Grained Bias-Variance Analysis
In the transfer learning paradigm models learn useful representations (or features) during a data-rich pretraining stage, and then use the pretrained representation to improve model performance on data-scarce downstream tasks. In this work, we explore transfer learning with the goal of optimizing downstream performance. We introduce a simple linear model that takes as input an arbitrary pretrained feature transform. We derive exact asymptotics of the downstream risk and its fine-grained bias-va…

Investigating Markers and Drivers of Gender Bias in Machine Translations
Peter J Barclay (Edinburgh Napier University), Ashkan Sami (Edinburgh Napier University)
https://arxiv.org/abs/2403.11896

Investigating Markers and Drivers of Gender Bias in Machine Translations
Implicit gender bias in Large Language Models (LLMs) is a well-documented problem, and implications of gender introduced into automatic translations can perpetuate real-world biases. However, some LLMs use heuristics or post-processing to mask such bias, making investigation difficult. Here, we examine bias in LLMss via back-translation, using the DeepL translation API to investigate the bias evinced when repeatedly translating a set of 56 Software Engineering tasks used in a previous study. Ea…

Uri Berliner's essay about NPR reignited criticism of the outlet on social media, especially among conservatives, and vehement pushback from some staffers (New York Times)
https://www.nytimes.com/2024/04/11/business/media/npr-criticism-liberal…

NPR in Turmoil After It Is Accused of Liberal Bias
An essay from an editor at the broadcaster has generated a firestorm of criticism about the network on social media, especially among conservatives.

Children's Overtrust and Shifting Perspectives of Generative AI
Jaemarie Solyst, Ellia Yang, Shixian Xie, Jessica Hammer, Amy Ogan, Motahhare Eslami
https://arxiv.org/abs/2404.14511

Children's Overtrust and Shifting Perspectives of Generative AI
The capabilities of generative AI (genAI) have dramatically increased in recent times, and there are opportunities for children to leverage new features for personal and school-related endeavors. However, while the future of genAI is taking form, there remain potentially harmful limitations, such as generation of outputs with misinformation and bias. We ran a workshop study focused on ChatGPT to explore middle school girls' (N = 26) attitudes and reasoning about how genAI works. We focused on g…

Understanding Optimal Feature Transfer via a Fine-Grained Bias-Variance Analysis
Yufan Li, Subhabrata Sen, Ben Adlam
https://arxiv.org/abs/2404.12481 https…

Understanding Optimal Feature Transfer via a Fine-Grained Bias-Variance Analysis
In the transfer learning paradigm models learn useful representations (or features) during a data-rich pretraining stage, and then use the pretrained representation to improve model performance on data-scarce downstream tasks. In this work, we explore transfer learning with the goal of optimizing downstream performance. We introduce a simple linear model that takes as input an arbitrary pretrained feature transform. We derive exact asymptotics of the downstream risk and its fine-grained bias-va…

Adjusting for Ascertainment Bias in Meta-Analysis of Penetrance for Cancer Risk
Thanthirige Lakshika M. Ruberu, Danielle Braun, Giovanni Parmigiani, Swati Biswas
https://arxiv.org/abs/2402.15030

Adjusting for Ascertainment Bias in Meta-Analysis of Penetrance for Cancer Risk
Multi-gene panel testing allows efficient detection of pathogenic variants in cancer susceptibility genes including moderate-risk genes such as ATM and PALB2. A growing number of studies examine the risk of breast cancer (BC) conferred by pathogenic variants of such genes. A meta-analysis combining the reported risk estimates can provide an overall age-specific risk of developing BC, i.e., penetrance for a gene. However, estimates reported by case-control studies often suffer from ascertainment…

This https://arxiv.org/abs/2401.01989 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCL_…

Revisiting Zero-Shot Abstractive Summarization in the Era of Large Language Models from the Perspective of Position Bias
We characterize and study zero-shot abstractive summarization in Large Language Models (LLMs) by measuring position bias, which we propose as a general formulation of the more restrictive lead bias phenomenon studied previously in the literature. Position bias captures the tendency of a model unfairly prioritizing information from certain parts of the input text over others, leading to undesirable behavior. Through numerous experiments on four diverse real-world datasets, we study position bias…

How Do Recommendation Models Amplify Popularity Bias? An Analysis from the Spectral Perspective
Siyi Lin, Chongming Gao, Jiawei Chen, Sheng Zhou, Binbin Hu, Can Wang
https://arxiv.org/abs/2404.12008

How Do Recommendation Models Amplify Popularity Bias? An Analysis from the Spectral Perspective
Recommendation Systems (RS) are often plagued by popularity bias. Specifically,when recommendation models are trained on long-tailed datasets, they not only inherit this bias but often exacerbate it. This effect undermines both the precision and fairness of RS and catalyzes the so-called Matthew Effect. Despite the widely recognition of this issue, the fundamental causes remain largely elusive. In our research, we delve deeply into popularity bias amplification. Our comprehensive theoretical an…

This https://arxiv.org/abs/2403.07910 has been replaced.
link: https://scholar.google.com/scholar?q=a

MAGPIE: Multi-Task Media-Bias Analysis Generalization for Pre-Trained Identification of Expressions
Media bias detection poses a complex, multifaceted problem traditionally tackled using single-task models and small in-domain datasets, consequently lacking generalizability. To address this, we introduce MAGPIE, the first large-scale multi-task pre-training approach explicitly tailored for media bias detection. To enable pre-training at scale, we present Large Bias Mixture (LBM), a compilation of 59 bias-related tasks. MAGPIE outperforms previous approaches in media bias detection on the Bias …

Demographic Bias of Expert-Level Vision-Language Foundation Models in Medical Imaging
Yuzhe Yang, Yujia Liu, Xin Liu, Avanti Gulhane, Domenico Mastrodicasa, Wei Wu, Edward J Wang, Dushyant W Sahani, Shwetak Patel
https://arxiv.org/abs/2402.14815

Demographic Bias of Expert-Level Vision-Language Foundation Models in Medical Imaging
Advances in artificial intelligence (AI) have achieved expert-level performance in medical imaging applications. Notably, self-supervised vision-language foundation models can detect a broad spectrum of pathologies without relying on explicit training annotations. However, it is crucial to ensure that these AI models do not mirror or amplify human biases, thereby disadvantaging historically marginalized groups such as females or Black patients. The manifestation of such biases could systematica…

This https://arxiv.org/abs/2403.01150 has been replaced.
link: https://scholar.google.com/scholar?q=a

Sequential Rotations and Error Analysis of a Simple Quaternion Estimator
Reference [1] introduces a novel closed-form quaternion estimator from two vector observations. The simplicity of the estimator sometimes yields singular expressions, the current annotation provides the simple rotation schemes for four singular cases. The estimator enables clear physical insights and a closed-form expression for the bias as a function of the quaternion error covariance matrix. The latter could be approximated up to second order with respect to the underlying measurement noise a…

This https://arxiv.org/abs/2304.02339 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_sta…

Combining experimental and observational data through a power likelihood
Randomized controlled trials are the gold standard for causal inference and play a pivotal role in modern evidence-based medicine. However, the sample sizes they use are often too limited to draw significant causal conclusions for subgroups that are less prevalent in the population. In contrast, observational data are becoming increasingly accessible in large volumes but can be subject to bias as a result of hidden confounding. Given these complementary features, we propose a power likelihood a…

AI Procurement Checklists: Revisiting Implementation in the Age of AI Governance
Tom Zick, Mason Kortz, David Eaves, Finale Doshi-Velez
https://arxiv.org/abs/2404.14660

AI Procurement Checklists: Revisiting Implementation in the Age of AI Governance
Public sector use of AI has been quietly on the rise for the past decade, but only recently have efforts to regulate it entered the cultural zeitgeist. While simple to articulate, promoting ethical and effective roll outs of AI systems in government is a notoriously elusive task. On the one hand there are hard-to-address pitfalls associated with AI-based tools, including concerns about bias towards marginalized communities, safety, and gameability. On the other, there is pressure not to make it…

This https://arxiv.org/abs/2312.15496 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_sta…

A Simple Bias Reduction for Chatterjee's Correlation
Chatterjee's rank correlation coefficient $ξ_n$ is an empirical index for detecting functional dependencies between two variables $X$ and $Y$. It is an estimator for a theoretical quantity $ξ$ that is zero for independence and one if $Y$ is a measurable function of $X$. Based on an equivalent characterization of sorted numbers, we derive an upper bound for $ξ_n$ and suggest a simple normalization aimed at reducing its bias for small sample size $n$. In Monte Carlo simulations of various case…