Search

The microstructure of a material can be engineered to achieve unique properties not found in nature. Microstructured materials, also known as metamaterials (MMs), can exhibit properties utilizing local resonance and dynamics of their heterogeneous microstructure that are activated below the traditional Bragg limit. In this study, the linear dynamic response of a low-frequency resonant ceramic MM slab is analyzed using the Finite Element Method (FEM) in the time domain. The MM is compared to monolithic slabs and other microstructured designs in terms of stress wave mitigation, peak load retardation, and energy transfer. Simulations are conducted using various boundary conditions and domain sizes to evaluate their influence on the performance. Potential graded slab designs and material damping effects are also discussed and are both shown to reduce the energy transmitted from the impact surface to the opposing surface significantly. The results showed that the MM slabs had superior performance in reducing the peak stress wave and reducing the transfer of energy. This study demonstrates that resonant ceramic MMs are a promising material design with unique and tunable properties that can be used for stress wave mitigation and structural protection applications.

Though research into text-to-image generators (T2Is) such as Stable Diffusion has demonstrated their amplification of societal biases and potentials to cause harm, such research has primarily relied on computational methods instead of seeking information from real users who experience harm, which is a significant knowledge gap. In this paper, we conduct the largest human subjects study of Stable Diffusion, with a combination of crowdsourced data from 133 crowdworkers and 14 semi-structured interviews across diverse countries and genders. Through a mixed-methods approach of intra-set cosine similarity hierarchies (i.e., comparing multiple Stable Diffusion outputs for the same prompt with each other to examine which result is 'closest' to the prompt) and qualitative thematic analysis, we first demonstrate a large disconnect between user expectations for Stable Diffusion outputs with those generated, evidenced by a set of Stable Diffusion renditions of `a Person' providing images far away from such expectations. We then extend this finding of general dissatisfaction into highlighting representational harms caused by Stable Diffusion upon our subjects, especially those with traditionally marginalized identities, subjecting them to incorrect and often dehumanizing stereotypes about their identities. We provide recommendations for a harm-aware approach to (re)design future versions of Stable Diffusion and other T2Is., Comment: Upcoming Publication, AIES 2024

This study introduces a taxonomy of stereotype content in contemporary large language models (LLMs). We prompt ChatGPT 3.5, Llama 3, and Mixtral 8x7B, three powerful and widely used LLMs, for the characteristics associated with 87 social categories (e.g., gender, race, occupations). We identify 14 stereotype dimensions (e.g., Morality, Ability, Health, Beliefs, Emotions), accounting for ~90% of LLM stereotype associations. Warmth and Competence facets were the most frequent content, but all other dimensions were significantly prevalent. Stereotypes were more positive in LLMs (vs. humans), but there was significant variability across categories and dimensions. Finally, the taxonomy predicted the LLMs' internal evaluations of social categories (e.g., how positively/negatively the categories were represented), supporting the relevance of a multidimensional taxonomy for characterizing LLM stereotypes. Our findings suggest that high-dimensional human stereotypes are reflected in LLMs and must be considered in AI auditing and debiasing to minimize unidentified harms from reliance in low-dimensional views of bias in LLMs.

Artificial intelligence (AI) hiring tools have revolutionized resume screening, and large language models (LLMs) have the potential to do the same. However, given the biases which are embedded within LLMs, it is unclear whether they can be used in this scenario without disadvantaging groups based on their protected attributes. In this work, we investigate the possibilities of using LLMs in a resume screening setting via a document retrieval framework that simulates job candidate selection. Using that framework, we then perform a resume audit study to determine whether a selection of Massive Text Embedding (MTE) models are biased in resume screening scenarios. We simulate this for nine occupations, using a collection of over 500 publicly available resumes and 500 job descriptions. We find that the MTEs are biased, significantly favoring White-associated names in 85.1\% of cases and female-associated names in only 11.1\% of cases, with a minority of cases showing no statistically significant differences. Further analyses show that Black males are disadvantaged in up to 100\% of cases, replicating real-world patterns of bias in employment settings, and validate three hypotheses of intersectionality. We also find an impact of document length as well as the corpus frequency of names in the selection of resumes. These findings have implications for widely used AI tools that are automating employment, fairness, and tech policy., Comment: To be published in Proceedings of the 2024 AAAI/ACM Conference on AI, Ethics, and Society; code available at https://github.com/kyrawilson/Resume-Screening-Bias

We propose PAV, Personalized Head Avatar for the synthesis of human faces under arbitrary viewpoints and facial expressions. PAV introduces a method that learns a dynamic deformable neural radiance field (NeRF), in particular from a collection of monocular talking face videos of the same character under various appearance and shape changes. Unlike existing head NeRF methods that are limited to modeling such input videos on a per-appearance basis, our method allows for learning multi-appearance NeRFs, introducing appearance embedding for each input video via learnable latent neural features attached to the underlying geometry. Furthermore, the proposed appearance-conditioned density formulation facilitates the shape variation of the character, such as facial hair and soft tissues, in the radiance field prediction. To the best of our knowledge, our approach is the first dynamic deformable NeRF framework to model appearance and shape variations in a single unified network for multi-appearances of the same subject. We demonstrate experimentally that PAV outperforms the baseline method in terms of visual rendering quality in our quantitative and qualitative studies on various subjects., Comment: Accepted to ECCV24. Project page: https://akincaliskan3d.github.io/PAV

Our research investigates the impact of Generative Artificial Intelligence (GAI) models, specifically text-to-image generators (T2Is), on the representation of non-Western cultures, with a focus on Indian contexts. Despite the transformative potential of T2Is in content creation, concerns have arisen regarding biases that may lead to misrepresentations and marginalizations. Through a community-centered approach and grounded theory analysis of 5 focus groups from diverse Indian subcultures, we explore how T2I outputs to English prompts depict Indian culture and its subcultures, uncovering novel representational harms such as exoticism and cultural misappropriation. These findings highlight the urgent need for inclusive and culturally sensitive T2I systems. We propose design guidelines informed by a sociotechnical perspective, aiming to address these issues and contribute to the development of more equitable and representative GAI technologies globally. Our work also underscores the necessity of adopting a community-centered approach to comprehend the sociotechnical dynamics of these models, complementing existing work in this space while identifying and addressing the potential negative repercussions and harms that may arise when these models are deployed on a global scale., Comment: This is the pre-peer reviewed version, which has been accepted at the 7th AAAI ACM Conference on AI, Ethics, and Society, Oct. 21, 2024, California, USA

Existing works examining Vision-Language Models (VLMs) for social biases predominantly focus on a limited set of documented bias associations, such as gender:profession or race:crime. This narrow scope often overlooks a vast range of unexamined implicit associations, restricting the identification and, hence, mitigation of such biases. We address this gap by probing VLMs to (1) uncover hidden, implicit associations across 9 bias dimensions. We systematically explore diverse input and output modalities and (2) demonstrate how biased associations vary in their negativity, toxicity, and extremity. Our work (3) identifies subtle and extreme biases that are typically not recognized by existing methodologies. We make the Dataset of retrieved associations, (Dora), publicly available here https://github.com/chahatraj/BiasDora., Comment: Accepted to EMNLP Findings 2024

Large Language Models (LLMs) perpetuate social biases, reflecting prejudices in their training data and reinforcing societal stereotypes and inequalities. Our work explores the potential of the Contact Hypothesis, a concept from social psychology for debiasing LLMs. We simulate various forms of social contact through LLM prompting to measure their influence on the model's biases, mirroring how intergroup interactions can reduce prejudices in social contexts. We create a dataset of 108,000 prompts following a principled approach replicating social contact to measure biases in three LLMs (LLaMA 2, Tulu, and NousHermes) across 13 social bias dimensions. We propose a unique debiasing technique, Social Contact Debiasing (SCD), that instruction-tunes these models with unbiased responses to prompts. Our research demonstrates that LLM responses exhibit social biases when subject to contact probing, but more importantly, these biases can be significantly reduced by up to 40% in 1 epoch of instruction tuning LLaMA 2 following our SCD strategy. Our code and data are available at https://github.com/chahatraj/breakingbias., Comment: Under Review

How good a research scientist is ChatGPT? We systematically probed the capabilities of GPT-3.5 and GPT-4 across four central components of the scientific process: as a Research Librarian, Research Ethicist, Data Generator, and Novel Data Predictor, using psychological science as a testing field. In Study 1 (Research Librarian), unlike human researchers, GPT-3.5 and GPT-4 hallucinated, authoritatively generating fictional references 36.0% and 5.4% of the time, respectively, although GPT-4 exhibited an evolving capacity to acknowledge its fictions. In Study 2 (Research Ethicist), GPT-4 (though not GPT-3.5) proved capable of detecting violations like p-hacking in fictional research protocols, correcting 88.6% of blatantly presented issues, and 72.6% of subtly presented issues. In Study 3 (Data Generator), both models consistently replicated patterns of cultural bias previously discovered in large language corpora, indicating that ChatGPT can simulate known results, an antecedent to usefulness for both data generation and skills like hypothesis generation. Contrastingly, in Study 4 (Novel Data Predictor), neither model was successful at predicting new results absent in their training data, and neither appeared to leverage substantially new information when predicting more versus less novel outcomes. Together, these results suggest that GPT is a flawed but rapidly improving librarian, a decent research ethicist already, capable of data generation in simple domains with known characteristics but poor at predicting novel patterns of empirical data to aid future experimentation., Comment: Main article is 14 pages, 1 table. Includes SI Appendix: 26 pages, 12 tables, 2 figures. Total: 40 pages, 13 tables, 2 figures. Under revised review at PNAS

If a linear electron accelerator is installed into the SPPC (Super Proton-Proton Collider) complex, ep collision options will be available in addition to pp collisions. We consider the production of excited electrons with spin-1/2 at the future SPPC-based electron-proton colliders with center-of-mass energies of $8.44$, $11.66$, $26.68$ and $36.88$ TeV. In the $ep\rightarrow e^{\star}X\rightarrow e\gamma X$ signal process, excited electrons are produced by contact interactions and decay into the photon channel by gauge interactions. Taking into account the corresponding background process, the pseudorapidity and transverse momentum distributions of the final state particles are plotted. We reported the discovery, observation and exclusion mass limits of excited electrons by applying appropriate kinematical cuts best suited for amplify the signal of the excited electron signature. We also investigated the highest achievable values of the compositeness scale for the discovery of excited electrons at these colliders., Comment: 21 pages, 9 Figures, 5 Tables

In Coevolving Latent Space Networks with Attractors (CLSNA) models, nodes in a latent space represent social actors, and edges indicate their dynamic interactions. Attractors are added at the latent level to capture the notion of attractive and repulsive forces between nodes, borrowing from dynamical systems theory. However, CLSNA reliance on MCMC estimation makes scaling difficult, and the requirement for nodes to be present throughout the study period limit practical applications. We address these issues by (i) introducing a Stochastic gradient descent (SGD) parameter estimation method, (ii) developing a novel approach for uncertainty quantification using SGD, and (iii) extending the model to allow nodes to join and leave over time. Simulation results show that our extensions result in little loss of accuracy compared to MCMC, but can scale to much larger networks. We apply our approach to the longitudinal social networks of members of US Congress on the social media platform X. Accounting for node dynamics overcomes selection bias in the network and uncovers uniquely and increasingly repulsive forces within the Republican Party.

The lack of reliable atomic data can be a severe limitation in astrophysical modelling, in particular of events such as kilonovae that require information on all neutron-capture elements across a wide range of ionization stages. Notably, the presence of non-orthonormalities between electron orbitals representing configurations that are close in energy can introduce significant inaccuracies in computed energies and transition probabilities. Here, we propose an explicit targeted optimization method that can effectively circumvent this concern while retaining an orthonormal orbital basis set. We illustrate this method within the framework of small-scale atomic structure models of Au I, using the GRASP2018 multiconfigurational Dirac-Hartree-Fock atomic structure code. By comparing to conventional optimization schemes we show how a targeted optimization approach improves the energy level positioning and ordering. Targeted optimization also leads to better agreement with experimental data for the strongest E1 transitions. This illustrates how small-scale models can be significantly improved with minor computational costs if orbital non-orthonormalities are considered carefully. These results should prove useful to multi-element atomic structure calculations in, for example, astrophysical opacity applications involving neutron-capture elements., Comment: 14 pages, 3 figures, 6 tables; published in Journal of Physics B

We present chemical abundances of the very bright metal-poor star HD~1936 based on high-resolution and high SNR spectra from AUKR. We obtain the abundances of 29 atomic species with atomic numbers between 3 and 63. In this context, the derived lithium abundance of 1.01 is consistent with the thin Li plateau observed in lower red giant branch stars. The star is a carbon-normal with the ratio of -0.31, just like other low-luminosity red giants on the thin Li plateau. We find the ratios of [Eu/Fe]=0.43 and [Ba/Eu]=-0.64, indicating very little s-process contamination. These ratios allow us to classify the star as a moderately r-process-enhanced (r-I) metal-poor star for the first time. It is worth mentioning that the star has a metallicity of -1.74, a [Cu/Fe] of -0.74, a [Zn/Fe] of 0.04, and a [Mg/C] of 0.69. The results suggest that it may be a second-generation star formed in a multi-enriched environment, rather than being a descendant of very massive first-generation stars. A last point worth mentioning is the possibility that HD~1936 may host a sub-stellar component with a mass of 18.35$M_{\rm J}$. Although our study does not confirm or deny this, we briefly discuss the possibility of the star hosting a planet.

We report results of optical identification and multi-wavelength study of a new polar-type magnetic cataclysmic variable (MCV), SRGA J213151.5+491400, discovered by Spectrum Roentgen-Gamma ($SRG$) observatory in the course of the all-sky survey. We present optical data from telescopes in Turkey (RTT-150 and T100 at the T\"UBITAK National Observatory), and in Russia (6-m and 1-m at SAO RAS), together with the X-ray data obtained with $ART-XC$ and $eROSITA$ telescopes aboard $SRG$ and the $NICER$ observatory. We detect SRGA J213151.5+491400 in a high state in 2020 (17.9 mag) that decreases about 3 mag into a low state (21 mag) in 2021. We find only one significant period using optical photometric time series analysis which reveals the white dwarf spin/orbital period to be 0.059710(1) days (85.982 min). The long slit spectroscopy in the high state yields a power law continuum increasing towards the blue with a prominent He II line along with the Balmer line emissions with no cyclotron humps; consistent with MCV nature. Doppler Tomography confirms the polar nature revealing ballistic stream accretion along with magnetic stream during the high state. These characteristics show that the new source is a polar-type MCV. $SRG$ $ART-XC$ detections yield an X-ray flux of (4.0-7.0)$\times$10$^{-12}$ erg cm$^2$ s$^{-1}$ in the high state. $eROSITA$ detects a dominating hot plasma component (kT$_{\rm{max}}$ $>$ 21 keV in the high state) declining to (4.0-6.0)$\times$10$^{-13}$ erg cm$^2$ s$^{-1}$ in 2021 (low state). The $NICER$ data obtained in the low state reveal a two-pole accretor showing a soft X-ray component at (6-7)$\sigma$ significance with a blackbody temperature of 15-18 eV. A soft X-ray component has never been detected for a polar in the low state before., Comment: 16 pages, 4 Tables and 15 Figures. Accepted for publication in A&A as it stands

Astrometry, the precise measurement of star motions, offers an alternative avenue to investigate low-frequency gravitational waves through the spatial deflection of photons, complementing pulsar timing arrays reliant on timing residuals. Upcoming data from Gaia and Roman can not only cross-check pulsar timing array findings but also explore the uncharted frequency range bridging pulsar timing arrays and LISA. We present an analytical framework to evaluate the feasibility of detecting a gravitational wave background, considering measurement noise and the intrinsic variability of the stochastic background. Furthermore, we highlight astrometry's crucial role in uncovering key properties of the gravitational wave background, such as spectral index and chirality, employing information-matrix analysis. Finally, we simulate the emergence of quadrupolar correlations, commonly referred to as the generalized Hellings-Downs curves., Comment: 34 pages, 8 figures. Published version in JCAP. v2: minor corrections

Pulsar-timing collaborations have recently reported evidence for the detection of an isotropic stochastic gravitational-wave background consistent with one sourced by a population of inspiralling supermassive black hole binaries. However, a certain degree of anisotropy and polarization may be present. Thus, the characterization of the energy density and polarization of the background at different angular scales is important. In this paper, we describe the signatures of linear polarization in the stochastic gravitational-wave background on the timing residuals obtained with pulsar-timing arrays. We expand the linear polarization map in terms of spin-weighted spherical harmonics and recast it into the $E$-mode (parity even) and $B$-mode (parity odd) basis. We provide expressions for the minimum-variance estimators for the coefficients of that expansion and evaluate the smallest detectable signal as a function of the signal-to-noise ratio with which the isotropic GW signal is detected and the number of pulsars in the survey. We evaluate the covariance between the estimators for the spherical-harmonic coefficients of the linear polarization $E$-modes and those for the intensity anisotropy. We also show that there is no covariance between the spherical-harmonic coefficients for the $B$-modes of the linear polarization and those for the circular polarization, even though both have the same parity. Our approach results in simple, elegant, and easily evaluated expressions for the overlap reduction functions for linear polarization., Comment: 15 pages, 4 figures. All comments are welcome

The intergalactic helium became fully ionized by the end of cosmic noon ($z\sim2$). Similarly to the reionization of hydrogen, helium reionization is expected to be patchy, driven by luminous quasars that ionize the intergalactic gas in their surrounding environment. Probing the morphology of ionized electrons during this epoch can provide crucial information about early structure formation, including the clustering and luminosities of quasars, the accretion rates, variability, and lifetimes of active galactic nuclei, as well as the growth and evolution of supermassive black holes. In this study, we present how measurements of the cosmic microwave background (CMB) can be used to reconstruct the optical-depth fluctuations resulting from patchy helium reionization. As helium reionization occurred at lower redshifts, upcoming probes of large-scale structure surveys will present a significant opportunity to enhance the prospects of probing this epoch by their combined analysis with the CMB. Using a joint information-matrix analysis of hydrogen and helium reionization, we show that near-future galaxy and CMB surveys will have enough statistical power to detect optical-depth fluctuations due to doubly-ionized helium, providing a way of measuring the redshift and duration of helium reionization to high significance. We also show that modeling uncertainties in helium reionization can impact the measurement precision of parameters characterizing hydrogen reionization., Comment: 34 pages, 20 figures, and 3 tables. All comments are welcome

We study stereotypes embedded within one of the most popular text-to-image generators: Stable Diffusion. We examine what stereotypes of gender and nationality/continental identity does Stable Diffusion display in the absence of such information i.e. what gender and nationality/continental identity is assigned to `a person', or to `a person from Asia'. Using vision-language model CLIP's cosine similarity to compare images generated by CLIP-based Stable Diffusion v2.1 verified by manual examination, we chronicle results from 136 prompts (50 results/prompt) of front-facing images of persons from 6 different continents, 27 nationalities and 3 genders. We observe how Stable Diffusion outputs of `a person' without any additional gender/nationality information correspond closest to images of men and least with persons of nonbinary gender, and to persons from Europe/North America over Africa/Asia, pointing towards Stable Diffusion having a concerning representation of personhood to be a European/North American man. We also show continental stereotypes and resultant harms e.g. a person from Oceania is deemed to be Australian/New Zealander over Papua New Guinean, pointing to the erasure of Indigenous Oceanic peoples, who form a majority over descendants of colonizers both in Papua New Guinea and in Oceania overall. Finally, we unexpectedly observe a pattern of oversexualization of women, specifically Latin American, Mexican, Indian and Egyptian women relative to other nationalities, measured through an NSFW detector. This demonstrates how Stable Diffusion perpetuates Western fetishization of women of color through objectification in media, which if left unchecked will amplify this stereotypical representation. Image datasets are made publicly available., Comment: Upcoming publication, Findings of EMNLP 2023

Previous work has established that a person's demographics and speech style affect how well speech processing models perform for them. But where does this bias come from? In this work, we present the Speech Embedding Association Test (SpEAT), a method for detecting bias in one type of model used for many speech tasks: pre-trained models. The SpEAT is inspired by word embedding association tests in natural language processing, which quantify intrinsic bias in a model's representations of different concepts, such as race or valence (something's pleasantness or unpleasantness) and capture the extent to which a model trained on large-scale socio-cultural data has learned human-like biases. Using the SpEAT, we test for six types of bias in 16 English speech models (including 4 models also trained on multilingual data), which come from the wav2vec 2.0, HuBERT, WavLM, and Whisper model families. We find that 14 or more models reveal positive valence (pleasantness) associations with abled people over disabled people, with European-Americans over African-Americans, with females over males, with U.S. accented speakers over non-U.S. accented speakers, and with younger people over older people. Beyond establishing that pre-trained speech models contain these biases, we also show that they can have real world effects. We compare biases found in pre-trained models to biases in downstream models adapted to the task of Speech Emotion Recognition (SER) and find that in 66 of the 96 tests performed (69%), the group that is more associated with positive valence as indicated by the SpEAT also tends to be predicted as speaking with higher valence by the downstream model. Our work provides evidence that, like text and image-based models, pre-trained speech based-models frequently learn human-like biases. Our work also shows that bias found in pre-trained models can propagate to the downstream task of SER.

Our interdisciplinary study investigates how effectively U.S. laws confront the challenges posed by Generative AI to human values. Through an analysis of diverse hypothetical scenarios crafted during an expert workshop, we have identified notable gaps and uncertainties within the existing legal framework regarding the protection of fundamental values, such as privacy, autonomy, dignity, diversity, equity, and physical/mental well-being. Constitutional and civil rights, it appears, may not provide sufficient protection against AI-generated discriminatory outputs. Furthermore, even if we exclude the liability shield provided by Section 230, proving causation for defamation and product liability claims is a challenging endeavor due to the intricate and opaque nature of AI systems. To address the unique and unforeseeable threats posed by Generative AI, we advocate for legal frameworks that evolve to recognize new threats and provide proactive, auditable guidelines to industry stakeholders. Addressing these issues requires deep interdisciplinary collaborations to identify harms, values, and mitigation strategies., Comment: 25 pages, 7 figures

The Laser Interferometer Space Antenna (LISA) will detect gravitational waves (GWs) emitted by massive black hole binaries (MBHBs) in the low-frequency ($\sim$mHz) band. Low-mass lenses, such as low-mass dark matter halos or subhalos, have sizes comparable to the wavelength of these GWs. Encounters with these lenses produce wave-optics (WO) effects that alter waveform phase and amplitude. Thus, a single event with observable WO effects can be used to probe the lens properties. In this paper, we first compute the probability of observing WO effects in a model-agnostic way. We perform information-matrix analyses over approximately 1000 MBHBs with total mass, mass ratio, and redshift spanning the ranges relevant to LISA. We then calculate lensing rates using three semi-analytical models of MBHB populations. In both cases, we use a waveform model that includes merger, ringdown, and higher-order modes. We use two lens population models: the theory-based Press-Schechter halo mass function and an observation-based model derived from Sloan Digital Sky Survey. We find that the probability of detecting WO effects can be as large as $\sim 3\%$, $\sim1.5\%$, and $\sim 1 \%$ at $1\sigma$, $3\sigma$, and $5\sigma$ confidence levels, respectively. The most optimistic MBHB population model yields $\sim 8$, $\sim 4$, and $\sim 3$ events with detectable WO effects at the same confidence levels, while the rates drop to $\sim 0.01$ in the more pessimistic scenarios. The most likely lens masses probed by LISA are in the range $(10^3, 10^8)\, M_{\odot}$, and the most probable redshifts are in the range $(0.3, 1.7)$. Therefore, LISA observations of WO effects can probe low-mass DM halos, complementing strong lensing and other observations., Comment: 25 pages, 15 figures; matches the version published in Phys. Rev. D 108, 123543 (2023)

Language models are trained on large-scale corpora that embed implicit biases documented in psychology. Valence associations (pleasantness/unpleasantness) of social groups determine the biased attitudes towards groups and concepts in social cognition. Building on this established literature, we quantify how social groups are valenced in English language models using a sentence template that provides an intersectional context. We study biases related to age, education, gender, height, intelligence, literacy, race, religion, sex, sexual orientation, social class, and weight. We present a concept projection approach to capture the valence subspace through contextualized word embeddings of language models. Adapting the projection-based approach to embedding association tests that quantify bias, we find that language models exhibit the most biased attitudes against gender identity, social class, and sexual orientation signals in language. We find that the largest and better-performing model that we study is also more biased as it effectively captures bias embedded in sociocultural data. We validate the bias evaluation method by overperforming on an intrinsic valence evaluation task. The approach enables us to measure complex intersectional biases as they are known to manifest in the outputs and applications of language models that perpetuate historical biases. Moreover, our approach contributes to design justice as it studies the associations of groups underrepresented in language such as transgender and homosexual individuals., Comment: to be published in AIES 2023