Search

We study the superconducting critical fields of two-dimensional electron systems at (111)$\mathrm{LaAlO_3/KTaO_3}$ interfaces as a function of electrostatic back-gating. Our work reveals inplane critical fields of unprecedented magnitudes at oxide interfaces. By comparing the critical fields in-plane and out-of-plane we discover an extremely anisotropic superconductor with an effective thickness below 1 nm and a 12-fold violation of the Chandrasekhar-Clogston paramagnetic limit. The analysis of magneto-transport indicates that the enhancement of the critical fields is due to an exceptionally thin superconducting layer and to a paramagnetic susceptibility suppressed by spin-orbit scattering.

Global partisan hostility and polarization has increased, and this polarization is heightened around presidential elections. Models capable of generating accurate summaries of diverse perspectives can help reduce such polarization by exposing users to alternative perspectives. In this work, we introduce a novel dataset and task for independently summarizing each political perspective in a set of passages from opinionated news articles. For this task, we propose a framework for evaluating different dimensions of perspective summary performance. We benchmark 10 models of varying sizes and architectures through both automatic and human evaluation. While recent models like GPT-4o perform well on this task, we find that all models struggle to generate summaries faithful to the intended perspective. Our analysis of summaries focuses on how extraction behavior depends on the features of the input documents.

We present the results of the "Fast Calorimeter Simulation Challenge 2022" - the CaloChallenge. We study state-of-the-art generative models on four calorimeter shower datasets of increasing dimensionality, ranging from a few hundred voxels to a few tens of thousand voxels. The 31 individual submissions span a wide range of current popular generative architectures, including Variational AutoEncoders (VAEs), Generative Adversarial Networks (GANs), Normalizing Flows, Diffusion models, and models based on Conditional Flow Matching. We compare all submissions in terms of quality of generated calorimeter showers, as well as shower generation time and model size. To assess the quality we use a broad range of different metrics including differences in 1-dimensional histograms of observables, KPD/FPD scores, AUCs of binary classifiers, and the log-posterior of a multiclass classifier. The results of the CaloChallenge provide the most complete and comprehensive survey of cutting-edge approaches to calorimeter fast simulation to date. In addition, our work provides a uniquely detailed perspective on the important problem of how to evaluate generative models. As such, the results presented here should be applicable for other domains that use generative AI and require fast and faithful generation of samples in a large phase space., Comment: 204 pages, 100+ figures, 30+ tables

Live comments, also known as Danmaku, are user-generated messages that are synchronized with video content. These comments overlay directly onto streaming videos, capturing viewer emotions and reactions in real-time. While prior work has leveraged live comments in affective analysis, its use has been limited due to the relative rarity of live comments across different video platforms. To address this, we first construct the Live Comment for Affective Analysis (LCAffect) dataset which contains live comments for English and Chinese videos spanning diverse genres that elicit a wide spectrum of emotions. Then, using this dataset, we use contrastive learning to train a video encoder to produce synthetic live comment features for enhanced multimodal affective content analysis. Through comprehensive experimentation on a wide range of affective analysis tasks (sentiment, emotion recognition, and sarcasm detection) in both English and Chinese, we demonstrate that these synthetic live comment features significantly improve performance over state-of-the-art methods.

Style representations aim to embed texts with similar writing styles closely and texts with different styles far apart, regardless of content. However, the contrastive triplets often used for training these representations may vary in both style and content, leading to potential content leakage in the representations. We introduce StyleDistance, a novel approach to training stronger content-independent style embeddings. We use a large language model to create a synthetic dataset of near-exact paraphrases with controlled style variations, and produce positive and negative examples across 40 distinct style features for precise contrastive learning. We assess the quality of our synthetic data and embeddings through human and automatic evaluations. StyleDistance enhances the content-independence of style embeddings, which generalize to real-world benchmarks and outperform leading style representations in downstream applications. Our model can be found at https://huggingface.co/StyleDistance/styledistance .

Ammonia is a key component of fertilizer and a potential clean fuel and hydrogen carrier. The Haber-Bosch process for ammonia synthesis consumes more than half of industrial hydrogen and contributes up to ~3% of global greenhouse gas emissions. Light-driven reactions via surface plasmon resonances offer a less energy-intensive pathway for ammonia production by altering reaction intermediates. Here, we report gold-ruthenium plasmonic bimetallic alloys for ammonia synthesis at room temperature and pressure, driven by visible light. We use colloidal synthesis to create AuRu$_x$ alloys (x=0.1, 0.2, 0.3) and disperse these nanoparticles on MgO supports for gas-phase ammonia synthesis. We observe a ~60 $\mu$mol/g/h reactivity and ~0.12% external quantum efficiency on a AuRu$_0$$_.$$_2$ sample under 100 mW/cm$^2$ visible light. In-situ diffuse reflective infrared Fourier transform spectroscopic measurements show that hydrogenation of nitrogen adsorbates is accelerated under light compared to thermocatalysis. Combining wavelength-dependent reactivity and spectroscopic findings with semi-classical electromagnetic modeling, we show plasmonic bimetallic alloys expedite ammonia synthesis by aiding hydrogenation of adsorbed nitrogen species via plasmon-mediated hot electrons. Quantum mechanical calculations reveal hydrogen-assisted N$_2$ splitting in the excited state is key to activating the reaction under ambient conditions. Therefore, light or H$_2$ alone cannot dissociate N$_2$ -- the key bottleneck to breaking N$_2$'s triple bond. Our findings are consistent with recent hypotheses on how nitrogenase enzymes catalyze ammonia production at mild conditions and provide insights for sustainable photochemical transformations., Comment: 21 pages, 4 figures, journal article submission soon

Background: Quantification of cardiac motion on pre-treatment CT imaging for stereotactic arrhythmia radiotherapy patients is difficult due to the presence of image artifacts caused by metal leads of implantable cardioverter-defibrillators (ICDs). New methods are needed to accurately reduce the metal artifacts in already reconstructed CTs to recover the otherwise lost anatomical information. Purpose: To develop a methodology to automatically detect metal artifacts in cardiac CT scans and inpaint the affected volume with anatomically consistent structures and values. Methods: ECG-gated 4DCT scans of 12 patients who underwent cardiac radiation therapy for treating ventricular tachycardia were collected. The metal artifacts in the images were manually contoured. A 2D U-Net deep learning (DL) model was developed to segment the metal artifacts. A dataset of synthetic CTs was prepared by adding metal artifacts from the patient images to artifact-free CTs. A 3D image inpainting DL model was trained to refill the metal artifact portion in the synthetic images with realistic values. The inpainting model was evaluated by analyzing the automated segmentation results of the four heart chambers on the synthetic dataset. Additionally, the raw cardiac patient cases were qualitatively inspected. Results: The artifact detection model produced a Dice score of 0.958 +- 0.008. The inpainting model was able to recreate images with a structural similarity index of 0.988 +- 0.012. With the chamber segmentations improved surface Dice scores from 0.684 +- 0.247 to 0.964 +- 0.067 and the Hausdorff distance reduced from 3.4 +- 3.9 mm to 0.7 +- 0.7 mm. The inpainting model's use on cardiac patient CTs was visually inspected and the artifact-inpainted images were visually plausible. Conclusion: We successfully developed two deep models to detect and inpaint metal artifacts in cardiac CT images.

Digital twins in manufacturing serve as a crucial bridge between the industrial age and the digital age, offering immense value. Current additive manufacturing processes are able to generate vast amounts of in-process data, which, when effectively ingested, can be transformed into insightful decisions. Data-driven methods from reduced order modeling and system identification are particularly promising in managing this data deluge. This study focuses on Laser Powder Directed Energy Deposition (LP-DED) equipped with in-situ process measurements to develop a compact virtual test-bed. This test-bed can accurately ingest arbitrary process inputs and report in-process observables as outputs. This virtual test-bed is derived using Dynamic Mode Decomposition with Control (DMDc) and is coupled with uncertainty quantification techniques to ensure robust predictions., Comment: 22 pages, 12 figures

Recent state-of-the-art authorship attribution methods learn authorship representations of texts in a latent, non-interpretable space, hindering their usability in real-world applications. Our work proposes a novel approach to interpreting these learned embeddings by identifying representative points in the latent space and utilizing LLMs to generate informative natural language descriptions of the writing style of each point. We evaluate the alignment of our interpretable space with the latent one and find that it achieves the best prediction agreement compared to other baselines. Additionally, we conduct a human evaluation to assess the quality of these style descriptions, validating their utility as explanations for the latent space. Finally, we investigate whether human performance on the challenging AA task improves when aided by our system's explanations, finding an average improvement of around +20% in accuracy., Comment: 8 pages, 8 figures, under review

Given an embedded closed submanifold $\Sigma^n$ in the closed Riemannian manifold $M^{n + k}$, where $k < n + 2$, we define extrinsic global conformal invariants of $\Sigma$ by renormalizing the volume associated to the unique singular Yamabe metric with singular set $\Sigma$. In case $n$ is odd, the renormalized volume is an absolute conformal invariant, while if $n$ is even, there is a conformally invariant energy term given by the integral of a local Riemannian invariant. In particular, the volume gives a global conformal invariant of a knot embedding in the three-sphere. We extend the construction of energies for even $n$ to general codimension by considering formal solutions to the singular Yamabe problem; except that, for each fixed $n$, there are finitely many $k > n + 2$, which we identify, for which the smoothness of the formal solution is obstructed and we obtain instead a pointwise conformal invariant. We compute the new quantities in several cases., Comment: 28 pages

Despite recent advances in human pose estimation (HPE), poor generalization to out-of-distribution (OOD) data remains a difficult problem. While previous works have proposed Test-Time Adaptation (TTA) to bridge the train-test domain gap by refining network parameters at inference, the absence of ground-truth annotations makes it highly challenging and existing methods typically increase inference times by one or more orders of magnitude. We observe that 1) not every test time sample is OOD, and 2) HPE errors are significantly larger on distal keypoints (wrist, ankle). To this end, we propose ESCAPE: a lightweight correction and selective adaptation framework which applies a fast, forward-pass correction on most data while reserving costly TTA for OOD data. The free energy function is introduced to separate OOD samples from incoming data and a correction network is trained to estimate the errors of pretrained backbone HPE predictions on the distal keypoints. For OOD samples, we propose a novel self-consistency adaptation loss to update the correction network by leveraging the constraining relationship between distal keypoints and proximal keypoints (shoulders, hips), via a second ``reverse" network. ESCAPE improves the distal MPJPE of five popular HPE models by up to 7% on unseen data, achieves state-of-the-art results on two popular HPE benchmarks, and is significantly faster than existing adaptation methods., Comment: 32 pages, 8 figures

In the field of emotion analysis, much NLP research focuses on identifying a limited number of discrete emotion categories, often applied across languages. These basic sets, however, are rarely designed with textual data in mind, and culture, language, and dialect can influence how particular emotions are interpreted. In this work, we broaden our scope to a practically unbounded set of \textit{affective states}, which includes any terms that humans use to describe their experiences of feeling. We collect and publish MASIVE, a dataset of Reddit posts in English and Spanish containing over 1,000 unique affective states each. We then define the new problem of \textit{affective state identification} for language generation models framed as a masked span prediction task. On this task, we find that smaller finetuned multilingual models outperform much larger LLMs, even on region-specific Spanish affective states. Additionally, we show that pretraining on MASIVE improves model performance on existing emotion benchmarks. Finally, through machine translation experiments, we find that native speaker-written data is vital to good performance on this task., Comment: EMNLP 2024

We prove rigidity and gap theorems for self-dual and even Poincar\'e-Einstein metrics in dimension four. As a corollary, we give an obstruction to the existence of self-dual Poincar\'e-Einstein metrics in terms of conformal invariants of the boundary and the topology of the bulk. As a by-product of our proof we identify a new scalar conformal invariant of three-dimensional Riemannian manifolds.

Human evaluation has been the gold standard for checking faithfulness in abstractive summarization. However, with a challenging source domain like narrative, multiple annotators can agree a summary is faithful, while missing details that are obvious errors only once pointed out. We therefore introduce a new dataset, STORYSUMM, comprising LLM summaries of short stories with localized faithfulness labels and error explanations. This benchmark is for evaluation methods, testing whether a given method can detect challenging inconsistencies. Using this dataset, we first show that any one human annotation protocol is likely to miss inconsistencies, and we advocate for pursuing a range of methods when establishing ground truth for a summarization dataset. We finally test recent automatic metrics and find that none of them achieve more than 70% balanced accuracy on this task, demonstrating that it is a challenging benchmark for future work in faithfulness evaluation., Comment: EMNLP Main 2024

Prior research has enhanced the ability of Large Language Models (LLMs) to solve logic puzzles using techniques such as chain-of-thought prompting or introducing a symbolic representation. These frameworks are still usually insufficient to solve complicated logical problems, such as Zebra puzzles, due to the inherent complexity of translating natural language clues into logical statements. We introduce a multi-agent system, ZPS, that integrates LLMs with an off the shelf theorem prover. This system tackles the complex puzzle-solving task by breaking down the problem into smaller, manageable parts, generating SMT (Satisfiability Modulo Theories) code to solve them with a theorem prover, and using feedback between the agents to repeatedly improve their answers. We also introduce an automated grid puzzle grader to assess the correctness of our puzzle solutions and show that the automated grader is reliable by evaluating it in a user-study. Our approach shows improvement in all three LLMs we tested, with GPT-4 showing 166% improvement in the number of fully correct solutions.

The goal of text style transfer is to transform the style of texts while preserving their original meaning, often with only a few examples of the target style. Existing style transfer methods generally rely on the few-shot capabilities of large language models or on complex controllable text generation approaches that are inefficient and underperform on fluency metrics. We introduce TinyStyler, a lightweight but effective approach, which leverages a small language model (800M params) and pre-trained authorship embeddings to perform efficient, few-shot text style transfer. We evaluate on the challenging task of authorship style transfer and find TinyStyler outperforms strong approaches such as GPT-4. We also evaluate TinyStyler's ability to perform text attribute style transfer (formal $\leftrightarrow$ informal) with automatic and human evaluations and find that the approach outperforms recent controllable text generation methods. Our model has been made publicly available at https://huggingface.co/tinystyler/tinystyler .

Vision-language models (VLMs) can respond to queries about images in many languages. However, beyond language, culture affects how we see things. For example, individuals from Western cultures focus more on the central figure in an image while individuals from Eastern cultures attend more to scene context. In this work, we present a novel investigation that demonstrates and localizes VLMs' Western bias in image understanding. We evaluate large VLMs across subjective and objective visual tasks with culturally diverse images and annotations. We find that VLMs perform better on the Western subset than the Eastern subset of each task. Controlled experimentation tracing the source of this bias highlights the importance of a diverse language mix in text-only pre-training for building equitable VLMs, even when inference is performed in English. Moreover, while prompting in the language of a target culture can lead to reductions in bias, it is not a substitute for building AI more representative of the world's languages., Comment: 17 pages, 7 figures. Code/models: https://github.com/amith-ananthram/see-it-from-my-perspective

This Letter reports the first measurement of the oscillation amplitude and frequency of reactor antineutrinos at Daya Bay via neutron capture on hydrogen using 1958 days of data. With over 3.6 million signal candidates, an optimized candidate selection, improved treatment of backgrounds and efficiencies, refined energy calibration, and an energy response model for the capture-on-hydrogen sensitive region, the relative $\overline{\nu}_{e}$ rates and energy spectra variation among the near and far detectors gives $\mathrm{sin}^22\theta_{13} = 0.0759_{-0.0049}^{+0.0050}$ and $\Delta m^2_{32} = (2.72^{+0.14}_{-0.15})\times10^{-3}$ eV$^2$ assuming the normal neutrino mass ordering, and $\Delta m^2_{32} = (-2.83^{+0.15}_{-0.14})\times10^{-3}$ eV$^2$ for the inverted neutrino mass ordering. This estimate of $\sin^2 2\theta_{13}$ is consistent with and essentially independent from the one obtained using the capture-on-gadolinium sample at Daya Bay. The combination of these two results yields $\mathrm{sin}^22\theta_{13}= 0.0833\pm0.0022$, which represents an 8% relative improvement in precision regarding the Daya Bay full 3158-day capture-on-gadolinium result.

This Letter presents results of a search for the mixing of a sub-eV sterile neutrino with three active neutrinos based on the full data sample of the Daya Bay Reactor Neutrino Experiment, collected during 3158 days of detector operation, which contains $5.55 \times 10^{6}$ reactor \anue candidates identified as inverse beta-decay interactions followed by neutron-capture on gadolinium. The analysis benefits from a doubling of the statistics of our previous result and from improvements of several important systematic uncertainties. No significant oscillation due to mixing of a sub-eV sterile neutrino with active neutrinos was found. Exclusion limits are set by both Feldman-Cousins and CLs methods. Light sterile neutrino mixing with $\sin^2 2\theta_{14} \gtrsim 0.01$ can be excluded at 95\% confidence level in the region of $0.01$ eV$^2 \lesssim |\Delta m^{2}_{41}| \lesssim 0.1 $ eV$^2$. This result represents the world-leading constraints in the region of $2 \times 10^{-4}$ eV$^2 \lesssim |\Delta m^{2}_{41}| \lesssim 0.2 $ eV$^2$., Comment: 7 pages, 4 figures, 1 table

We evaluate recent Large Language Models (LLMs) on the challenging task of summarizing short stories, which can be lengthy, and include nuanced subtext or scrambled timelines. Importantly, we work directly with authors to ensure that the stories have not been shared online (and therefore are unseen by the models), and to obtain informed evaluations of summary quality using judgments from the authors themselves. Through quantitative and qualitative analysis grounded in narrative theory, we compare GPT-4, Claude-2.1, and LLama-2-70B. We find that all three models make faithfulness mistakes in over 50% of summaries and struggle with specificity and interpretation of difficult subtext. We additionally demonstrate that LLM ratings and other automatic metrics for summary quality do not correlate well with the quality ratings from the writers., Comment: pre-MIT Press publication version

We present NewsQs (news-cues), a dataset that provides question-answer pairs for multiple news documents. To create NewsQs, we augment a traditional multi-document summarization dataset with questions automatically generated by a T5-Large model fine-tuned on FAQ-style news articles from the News On the Web corpus. We show that fine-tuning a model with control codes produces questions that are judged acceptable more often than the same model without them as measured through human evaluation. We use a QNLI model with high correlation with human annotations to filter our data. We release our final dataset of high-quality questions, answers, and document clusters as a resource for future work in query-based multi-document summarization., Comment: minor wording change

There are many settings where it is useful to predict and explain the success or failure of a dialogue. Circumplex theory from psychology models the social orientations (e.g., Warm-Agreeable, Arrogant-Calculating) of conversation participants and can be used to predict and explain the outcome of social interactions. Our work is novel in its systematic application of social orientation tags to modeling conversation outcomes. In this paper, we introduce a new data set of dialogue utterances machine-labeled with social orientation tags. We show that social orientation tags improve task performance, especially in low-resource settings, on both English and Chinese language benchmarks. We also demonstrate how social orientation tags help explain the outcomes of social interactions when used in neural models. Based on these results showing the utility of social orientation tags for dialogue outcome prediction tasks, we release our data sets, code, and models that are fine-tuned to predict social orientation tags on dialogue utterances., Comment: Accepted to LREC-COLING 2024

Humor is a fundamental facet of human cognition and interaction. Yet, despite recent advances in natural language processing, humor detection remains a challenging task that is complicated by the scarcity of datasets that pair humorous texts with similar non-humorous counterparts. In our work, we investigate whether large language models (LLMs), can generate synthetic data for humor detection via editing texts. We benchmark LLMs on an existing human dataset and show that current LLMs display an impressive ability to 'unfun' jokes, as judged by humans and as measured on the downstream task of humor detection. We extend our approach to a code-mixed English-Hindi humor dataset, where we find that GPT-4's synthetic data is highly rated by bilingual annotators and provides challenging adversarial examples for humor classifiers.

Single document news summarization has seen substantial progress on faithfulness in recent years, driven by research on the evaluation of factual consistency, or hallucinations. We ask whether these advances carry over to other text summarization domains. We propose a new evaluation benchmark on topic-focused dialogue summarization, generated by LLMs of varying sizes. We provide binary sentence-level human annotations of the factual consistency of these summaries along with detailed explanations of factually inconsistent sentences. Our analysis shows that existing LLMs hallucinate significant amounts of factual errors in the dialogue domain, regardless of the model's size. On the other hand, when LLMs, including GPT-4, serve as binary factual evaluators, they perform poorly and can be outperformed by prevailing state-of-the-art specialized factuality evaluation metrics. Finally, we conducted an analysis of hallucination types with a curated error taxonomy. We find that there are diverse errors and error distributions in model-generated summaries and that non-LLM based metrics can capture all error types better than LLM-based evaluators., Comment: NAACL 2024; Linguistic annotations available at https://github.com/amazon-science/tofueval

Pre-trained language models (LMs) are capable of in-context learning (ICL): they can adapt to a task with only a few examples given in the prompt without any parameter update. However, it is unclear where this capability comes from as there is a stark distribution shift between pre-training text and ICL prompts. In this work, we study what patterns of the pre-training data contribute to ICL. We find that LMs' ICL ability depends on $\textit{parallel structures}$ in the pre-training data -- pairs of phrases following similar templates in the same context window. Specifically, we detect parallel structures by checking whether training on one phrase improves prediction of the other, and conduct ablation experiments to study their effect on ICL. We show that removing parallel structures in the pre-training data reduces LMs' ICL accuracy by 51% (vs 2% from random ablation). This drop persists even when excluding common patterns such as n-gram repetitions and long-range dependency, showing the diversity and generality of parallel structures. A closer look at the detected parallel structures indicates that they cover diverse linguistic tasks and span long distances in the data.

Confinement of monolayers into quasi-one-dimensional atomically-thin nanoribbons could lead to novel quantum phenomena beyond those achieved in their bulk and monolayer counterparts. However, current experimental availability of nanoribbon species beyond graphene has been limited to bottom-up synthesis or top-down patterning. In this study, we introduce a versatile and direct lithography-free approach to exfoliate a variety of bulk van der Waals (vdW) crystals into nanoribbons. Akin to the Scotch tape exfoliation in producing monolayers, this technique provides convenient access to a wide range of nanoribbons derived from their corresponding bulk crystals, including MoS2, WS2, MoSe2, WSe2, MoTe2, WTe2, ReS2, and hBN. The nanoribbons are single-crystalline, parallel-aligned, flat, and have high aspect ratio. We demonstrated the electrical, magnetic, and optical properties from the confinement, strain, and edge configurations of these nanoribbons. This versatile preparation technique will pave the way for future experimental investigation and broad applications in optoelectronic, sensing, electronic and quantum devices., Comment: 18 pages, 5 figures

Daya Bay presents the first measurement of cosmogenic $^8$He isotope production in liquid scintillator, using an innovative method for identifying cascade decays of $^8$He and its child isotope, $^8$Li. We also measure the production yield of $^9$Li isotopes using well-established methodology. The results, in units of 10$^{-8}\mu^{-1}$g$^{-1}$cm$^{2}$, are 0.307$\pm$0.042, 0.341$\pm$0.040, and 0.546$\pm$0.076 for $^8$He, and 6.73$\pm$0.73, 6.75$\pm$0.70, and 13.74$\pm$0.82 for $^9$Li at average muon energies of 63.9~GeV, 64.7~GeV, and 143.0~GeV, respectively. The measured production rate of $^8$He isotopes is more than an order of magnitude lower than any other measurement of cosmogenic isotope production. It replaces the results of previous attempts to determine the ratio of $^8$He to $^9$Li production that yielded a wide range of limits from 0 to 30\%. The results provide future liquid-scintillator-based experiments with improved ability to predict cosmogenic backgrounds.

The full data set of the Daya Bay reactor neutrino experiment is used to probe the effect of the charged current non-standard interactions (CC-NSI) on neutrino oscillation experiments. Two different approaches are applied and constraints on the corresponding CC-NSI parameters are obtained with the neutrino flux taken from the Huber-Mueller model with a $5\%$ uncertainty. For the quantum mechanics-based approach (QM-NSI), the constraints on the CC-NSI parameters $\epsilon_{e\alpha}$ and $\epsilon_{e\alpha}^{s}$ are extracted with and without the assumption that the effects of the new physics are the same in the production and detection processes, respectively. The approach based on the weak effective field theory (WEFT-NSI) deals with four types of CC-NSI represented by the parameters $[\varepsilon_{X}]_{e\alpha}$. For both approaches, the results for the CC-NSI parameters are shown for cases with various fixed values of the CC-NSI and the Dirac CP-violating phases, and when they are allowed to vary freely. We find that constraints on the QM-NSI parameters $\epsilon_{e\alpha}$ and $\epsilon_{e\alpha}^{s}$ from the Daya Bay experiment alone can reach the order $\mathcal{O}(0.01)$ for the former and $\mathcal{O}(0.1)$ for the latter, while for WEFT-NSI parameters $[\varepsilon_{X}]_{e\alpha}$, we obtain $\mathcal{O}(0.1)$ for both cases., Comment: 25 pages, 16 figures, 6 tables; 36 pages, format changed, references added

We show that the half-ball in $\mathbb{R}^4$ can be conformally changed so that the only contribution to the Gauss--Bonnet formula is a constant term at the corner. This may be seen as a fourth-order Cherrier--Escobar-type problem on the half-ball., Comment: 28 pages

This project reviews data collection practices regarding race, ethnicity and ancestry (REA) in post-secondary institutions (PSIs) in Canada, as well as in other relevant sectors (e.g., health, K-12 education, government agencies). The goal of the project was to identify promising practices and to develop recommendations to guide REA data collection initiatives both at the institutional and system level in BC post-secondary education. The information was collected using systematic literature and documentary reviews of existing practices on data collection regarding REA, interviews with representatives from PSIs and other organizations, and focus groups with students and community representatives. The framework for the collection and use of student demographic data, which was developed through the project work, offers a comprehensive approach to demographic data both at the institutional and system levels, considering the tools to be used, as well as appropriate processes to meet the objectives of the data collection. The report contains a discussion of key terms and definitions as well as examples of data collection questions used by other PSIs and organizations. [This report was prepared by the Planning and Institutional Research Office University of British Columbia in collaboration with DPM Research for EducationPlannerBC (EPBC).]

People from different social and demographic groups express diverse perspectives and conflicting opinions on a broad set of topics such as product reviews, healthcare, law, and politics. A fair summary should provide a comprehensive coverage of diverse perspectives without underrepresenting certain groups. However, current work in summarization metrics and Large Language Models (LLMs) evaluation has not explored fair abstractive summarization. In this paper, we systematically investigate fair abstractive summarization for user-generated data. We first formally define fairness in abstractive summarization as not underrepresenting perspectives of any groups of people, and we propose four reference-free automatic metrics by measuring the differences between target and source perspectives. We evaluate nine LLMs, including three GPT models, four LLaMA models, PaLM 2, and Claude, on six datasets collected from social media, online reviews, and recorded transcripts. Experiments show that both the model-generated and the human-written reference summaries suffer from low fairness. We conduct a comprehensive analysis of the common factors influencing fairness and propose three simple but effective methods to alleviate unfair summarization. Our dataset and code are available at https://github.com/psunlpgroup/FairSumm., Comment: NAACL 2024