Search

Fluid-solid interfacial free energy (IFE) is a fundamental parameter influencing wetting behaviors, which play a crucial role across a broad range of industrial applications. Obtaining reliable data for fluid-solid IFE remains challenging with experimental and semi-empirical methods, and the applicability of first-principle theoretical methods is constrained by a lack of accessible computational tools. In recent years, a variety of molecular simulation methods have been developed for determining the fluid-solid IFE. This review provides a comprehensive summary and critical evaluation of these techniques. The developments, fundamental principles, and implementations of various simulation methods are presented from mechanical routes, such as the contact angle approach, the technique using Bakker's equation, and the Wilhelmy simulation method, as well as thermodynamic routes, including the cleaving wall method, the Frenkel-Ladd technique, and the test-volume/area methods. These approaches can be applied to compute various fluid-solid interfacial properties, including IFE, relative IFE, surface stress, and superficial tension, although these properties are often used without differentiation in the literature. Additionally, selected applications of these methods are reviewed to provide insight into the behavior of fluid-solid interfacial energies in diverse systems. We also illustrate two interpretations of the fluid-solid IFE based on the theory of Navascues and Berry and Bakker's equation. It is shown that the simulation methods developed from these two interpretations are identical. This review advocates for the broader adoption of molecular simulation methods in estimating fluid-solid IFE, which is essential for advancing our understanding of wetting behaviors in various chemical systems.

Two-dimensional (2D) transition metal di-chalcogenide layers with high electrical conductivity and spin-orbit coupling (SOC) can find huge potential in spintronic devices. With limited success of 2D spin Hall material development, we demonstrate vanadium (V) substitutionally doped monolayer MoS2 (VMS) as a potential spin Hall material having tunable electrical conductivity, SOC strength, and room temperature magnetism. Systematic enhancement in the electrical conductivity is observed with the extent of V doping, where it is enhanced from ~0.3 S/m of MoS2 to ~100000 S/m upon doping to the level of 9 atomic%. Ferromagnetic resonance (FMR) based spin-pumping experiments indicate the spin transport across the junction of permalloy (Py) and VMS. Spin-torque FMR measurements demonstrate the suggesting latter's potential as a spin-orbit torque layer in 2D spintronic devices.

In recent years, the development of the Lunar Gateway and Artemis missions has renewed interest in lunar exploration, including both manned and unmanned missions. This interest necessitates accurate initial orbit determination (IOD) and orbit prediction (OP) in this domain, which faces significant challenges such as severe nonlinearity, sensitivity to initial conditions, large state-space volume, and sparse, faint, and unreliable measurements. This paper explores the capability of data-driven Koopman operator-based approximations for OP in these scenarios. Three stable periodic trajectories from distinct cislunar families are analyzed. The analysis includes theoretical justification for using a linear time-invariant system as the data-driven surrogate. This theoretical framework is supported by experimental validation. Furthermore, the accuracy is assessed by comparing the spectral content captured to period estimates derived from the fast Fourier transform (FFT) and Poincare-like sections., Comment: arXiv admin note: text overlap with arXiv:2401.13784

Computational narrative understanding studies the identification, description, and interaction of the elements of a narrative: characters, attributes, events, and relations. Narrative research has given considerable attention to defining and classifying character types. However, these character-type taxonomies do not generalize well because they are small, too simple, or specific to a domain. We require robust and reliable benchmarks to test whether narrative models truly understand the nuances of the character's development in the story. Our work addresses this by curating the Chatter dataset that labels whether a character portrays some attribute for 88148 character-attribute pairs, encompassing 2998 characters, 13324 attributes and 660 movies. We validate a subset of Chatter, called ChatterEval, using human annotations to serve as an evaluation benchmark for the character attribution task in movie scripts. ChatterEval assesses narrative understanding and the long-context modeling capacity of language models., Comment: submitted to NAACL 2025

This study explores the feasibility of using large language models (LLMs), specifically GPT-4o (ChatGPT), for automated grading of conceptual questions in an undergraduate Mechanical Engineering course. We compared the grading performance of GPT-4o with that of human teaching assistants (TAs) on ten quiz problems from the MEEN 361 course at Texas A&M University, each answered by approximately 225 students. Both the LLM and TAs followed the same instructor-provided rubric to ensure grading consistency. We evaluated performance using Spearman's rank correlation coefficient and Root Mean Square Error (RMSE) to assess the alignment between rankings and the accuracy of scores assigned by GPT-4o and TAs under zero- and few-shot grading settings. In the zero-shot setting, GPT-4o demonstrated a strong correlation with TA grading, with Spearman's rank correlation coefficient exceeding 0.6 in seven out of ten datasets and reaching a high of 0.9387. Our analysis reveals that GPT-4o performs well when grading criteria are straightforward but struggles with nuanced answers, particularly those involving synonyms not present in the rubric. The model also tends to grade more stringently in ambiguous cases compared to human TAs. Overall, ChatGPT shows promise as a tool for grading conceptual questions, offering scalability and consistency., Comment: 21 pages, 21 figures

We study the problem of learning mixtures of Gaussians with approximate differential privacy. We prove that roughly $kd^2 + k^{1.5} d^{1.75} + k^2 d$ samples suffice to learn a mixture of $k$ arbitrary $d$-dimensional Gaussians up to low total variation distance, with differential privacy. Our work improves over the previous best result [AAL24b] (which required roughly $k^2 d^4$ samples) and is provably optimal when $d$ is much larger than $k^2$. Moreover, we give the first optimal bound for privately learning mixtures of $k$ univariate (i.e., $1$-dimensional) Gaussians. Importantly, we show that the sample complexity for privately learning mixtures of univariate Gaussians is linear in the number of components $k$, whereas the previous best sample complexity [AAL21] was quadratic in $k$. Our algorithms utilize various techniques, including the inverse sensitivity mechanism [AD20b, AD20a, HKMN23], sample compression for distributions [ABDH+20], and methods for bounding volumes of sumsets., Comment: 52 pages. To appear in Neural Information Processing Systems (NeurIPS), 2024

Chronic wounds are a significant burden on individuals and the healthcare system, affecting millions of people and incurring high costs. Wound classification using deep learning techniques is a promising approach for faster diagnosis and treatment initiation. However, lack of high quality data to train the ML models is a major challenge to realize the potential of ML in wound care. In fact, data limitations are the biggest challenge in studies using medical or forensic imaging today. We study data augmentation techniques that can be used to overcome the data scarcity limitations and unlock the potential of deep learning based solutions. In our study we explore a range of data augmentation techniques from geometric transformations of wound images to advanced GANs, to enrich and expand datasets. Using the Keras, Tensorflow, and Pandas libraries, we implemented the data augmentation techniques that can generate realistic wound images. We show that geometric data augmentation can improve classification performance, F1 scores, by up to 11% on top of state-of-the-art models, across several key classes of wounds. Our experiments with GAN based augmentation prove the viability of using DE-GANs to generate wound images with richer variations. Our study and results show that data augmentation is a valuable privacy-preserving tool with huge potential to overcome the data scarcity limitations and we believe it will be part of any real-world ML-based wound care system.

We propose the convex floating body membership problem, which consists of efficiently determining when a query point $a\in\mathbb{R}^d$ belongs to the so-called $\varepsilon$-convex floating body of a given bounded convex domain $K\subset\mathbb{R}^d$. We consider this problem in an approximate setting, i.e., given a parameter $\delta>0$, the query can be answered either way if the Hilbert distance in $K$ of $a$ from the boundary of a relatively-scaled $\varepsilon$-convex floating body is less than $\delta$. We present a data structure for this problem that has storage size $O(\delta^{-d}\varepsilon^{-(d-1)/2})$ and achieves query time of $O({\delta^{-1}}\ln 1/\varepsilon)$. Our construction is motivated by a recent work of Abdelkader and Mount on APM queries, and relies on a comparison of convex floating bodies with balls in the Hilbert metric on $K$., Comment: 19 pages, 3 figures

We develop a novel two-layer approach for optimising mortgage relief products through a simulated multi-agent mortgage environment. While the approach is generic, here the environment is calibrated to the US mortgage market based on publicly available census data and regulatory guidelines. Through the simulation layer, we assess the resilience of households to exogenous income shocks, while the optimisation layer explores strategies to improve the robustness of households to these shocks by making novel mortgage assistance products available to households. Households in the simulation are adaptive, learning to make mortgage-related decisions (such as product enrolment or strategic foreclosures) that maximize their utility, balancing their available liquidity and equity. We show how this novel two-layer simulation approach can successfully design novel mortgage assistance products to improve household resilience to exogenous shocks, and balance the costs of providing such products through post-hoc analysis. Previously, such analysis could only be conducted through expensive pilot studies involving real participants, demonstrating the benefit of the approach for designing and evaluating financial products., Comment: Accepted at the 5th ACM International Conference on AI in Finance

Task-oriented dialogue systems are essential for applications ranging from customer service to personal assistants and are widely used across various industries. However, developing effective multi-domain systems remains a significant challenge due to the complexity of handling diverse user intents, entity types, and domain-specific knowledge across several domains. In this work, we propose DARD (Domain Assigned Response Delegation), a multi-agent conversational system capable of successfully handling multi-domain dialogs. DARD leverages domain-specific agents, orchestrated by a central dialog manager agent. Our extensive experiments compare and utilize various agent modeling approaches, combining the strengths of smaller fine-tuned models (Flan-T5-large & Mistral-7B) with their larger counterparts, Large Language Models (LLMs) (Claude Sonnet 3.0). We provide insights into the strengths and limitations of each approach, highlighting the benefits of our multi-agent framework in terms of flexibility and composability. We evaluate DARD using the well-established MultiWOZ benchmark, achieving state-of-the-art performance by improving the dialogue inform rate by 6.6% and the success rate by 4.1% over the best-performing existing approaches. Additionally, we discuss various annotator discrepancies and issues within the MultiWOZ dataset and its evaluation system.

Metric Temporal Logic (MTL) and Timed Propositional Temporal Logic (TPTL) extend Linear Temporal Logic (LTL) for real-time constraints, with MTL using time-bounded modalities and TPTL employing freeze quantifiers. Satisfiability for both is generally undecidable; however, MTL becomes decidable under certain non-punctual and partially-punctual restrictions. Punctuality can be restored trivially under similar non-punctual restrictions on TPTL even for one variable fragment. Our first contribution is to study more restricted notion of openness for 1-TPTL, under which punctuality can not be recovered. We show that even under such restrictions, the satisfiability checking does not get computationally easier. This implies that 1-TPTL (and hence TPTL) does not enjoy benefits of relaxing punctuality unlike MTL. As our second contribution we introduce a refined, partially adjacent restriction in 1-TPTL (PA-1-TPTL), and prove decidability for its satisfiability checking. We show that this logic is strictly more expressive than partially punctual Metric Temporal Logic, making this as one of the most expressive known boolean-closed decidable timed logic., Comment: arXiv admin note: text overlap with arXiv:1705.01501

This study evaluates the concordance between RNA sequencing (RNA-Seq) and NanoString technologies for gene expression analysis in non-human primates (NHPs) infected with Ebola virus (EBOV). We performed a detailed comparison of both platforms, demonstrating a strong correlation between them, with Spearman coefficients for 56 out of 62 samples ranging from 0.78 to 0.88, with a mean of 0.83 and a median of 0.85. Bland-Altman analysis further confirmed high consistency, with most measurements falling within 95% confidence limits. A machine learning approach, using the Supervised Magnitude-Altitude Scoring (SMAS) method trained on NanoString data, identified OAS1 as a key marker for distinguishing RT-qPCR positive from negative samples. Remarkably, when applied to RNA-Seq data, OAS1 also achieved 100% accuracy in differentiating infected from uninfected samples using logistic regression, demonstrating its robustness across platforms. Further differential expression analysis identified 12 common genes including ISG15, OAS1, IFI44, IFI27, IFIT2, IFIT3, IFI44L, MX1, MX2, OAS2, RSAD2, and OASL which demonstrated the highest levels of statistical significance and biological relevance across both platforms. Gene Ontology (GO) analysis confirmed that these genes are directly involved in key immune and viral infection pathways, reinforcing their importance in EBOV infection. In addition, RNA-Seq uniquely identified genes such as CASP5, USP18, and DDX60, which play key roles in immune regulation and antiviral defense. This finding highlights the broader detection capabilities of RNA-Seq and underscores the complementary strengths of both platforms in providing a comprehensive and accurate assessment of gene expression changes during Ebola virus infection.

We study the density estimation problem defined as follows: given $k$ distributions $p_1, \ldots, p_k$ over a discrete domain $[n]$, as well as a collection of samples chosen from a ``query'' distribution $q$ over $[n]$, output $p_i$ that is ``close'' to $q$. Recently~\cite{aamand2023data} gave the first and only known result that achieves sublinear bounds in {\em both} the sampling complexity and the query time while preserving polynomial data structure space. However, their improvement over linear samples and time is only by subpolynomial factors. Our main result is a lower bound showing that, for a broad class of data structures, their bounds cannot be significantly improved. In particular, if an algorithm uses $O(n/\log^c k)$ samples for some constant $c>0$ and polynomial space, then the query time of the data structure must be at least $k^{1-O(1)/\log \log k}$, i.e., close to linear in the number of distributions $k$. This is a novel \emph{statistical-computational} trade-off for density estimation, demonstrating that any data structure must use close to a linear number of samples or take close to linear query time. The lower bound holds even in the realizable case where $q=p_i$ for some $i$, and when the distributions are flat (specifically, all distributions are uniform over half of the domain $[n]$). We also give a simple data structure for our lower bound instance with asymptotically matching upper bounds. Experiments show that the data structure is quite efficient in practice., Comment: To appear at NeurIPS 2024

Violation of the Bell-type inequalities is very necessary to confirm the existence of the nonlocality in the nonclassical (entangled) states. We have designed a customized operator which is made of the sum of the identity and Pauli matrices ($I$, $\sigma_x$, $\sigma_y$, and $\sigma_z$). We theoretically evaluate the Bell-type violation for the two-qubit Bell state and a four-qubit Dicke state, which gives the Bell-CHSH parameter values $2\sqrt{2}$ and $3.05$, respectively for our customized operator. For experimental implementation, IBM's 127-qubitQuantum Processing Units (QPU) were utilized, where we have applied our customized operator to evaluate Bell-type inequalities for two-qubit Bell state ($\vert\Phi^+\rangle$) and four-qubit Dicke state ($|D^{(2)}_4\rangle$). We observed, for the two-qubit Bell state, the experimental Bell violation was $2.7507\pm 0.0197$. For Dicke state, we found the violation be to $2.1239\pm0.0457$ and $2.2175\pm0.0352$ respectively for two distinct methods of state preparation. All our results show clear violation of the local realism; however, we find that the experimental violation of the Bell state ($2.75$) is close to the theoretical ($2.82$) results due to lower circuit depth in state-preparation as well as fewer measurements, while the Dicke state shows greater errors ($2.12$ and $2.21$ vs. $3.05$) from higher depth and more measurements.

The quantum numbers of monopoles in $R^3$ in the presence of massless fermions have been analyzed using a uniform flux background in $S^2\times R$ coupled to fermions. An analogous study in $T^2\times R$ is performed by studying the discrete symmetries of the Dirac Hamiltonian in the presence of a static uniform field on $T^2$ with a total flux of $Q$ in the continuum. The degenerate ground states are classified based on their transformation properties under $\frac{\pi}{2}$ rotations of $T^2$ that leave the background field invariant. We find that the lattice analysis with overlap fermions exactly reproduces the transformation properties of the single particle zero modes in the continuum. Whereas the transformation properties of the single particle negative energy states can be studied in the continuum and the lattice, we are also able to study the transformation properties and the particle number (charge) of the many-body ground state on a finite lattice, and we show that the contributions from the fully filled single-particle states cannot be ignored., Comment: 15 pages

GPT-4o is an autoregressive omni model that accepts as input any combination of text, audio, image, and video, and generates any combination of text, audio, and image outputs. It's trained end-to-end across text, vision, and audio, meaning all inputs and outputs are processed by the same neural network. GPT-4o can respond to audio inputs in as little as 232 milliseconds, with an average of 320 milliseconds, which is similar to human response time in conversation. It matches GPT-4 Turbo performance on text in English and code, with significant improvement on text in non-English languages, while also being much faster and 50\% cheaper in the API. GPT-4o is especially better at vision and audio understanding compared to existing models. In line with our commitment to building AI safely and consistent with our voluntary commitments to the White House, we are sharing the GPT-4o System Card, which includes our Preparedness Framework evaluations. In this System Card, we provide a detailed look at GPT-4o's capabilities, limitations, and safety evaluations across multiple categories, focusing on speech-to-speech while also evaluating text and image capabilities, and measures we've implemented to ensure the model is safe and aligned. We also include third-party assessments on dangerous capabilities, as well as discussion of potential societal impacts of GPT-4o's text and vision capabilities.

Prototype-based methods are intrinsically interpretable XAI methods that produce predictions and explanations by comparing input data with a set of learned prototypical examples that are representative of the training data. In this work, we discuss a series of developments in the field of prototype-based XAI that show potential for scientific learning tasks, with a focus on the geosciences. We organize the prototype-based XAI literature into three themes: the development and visualization of prototypes, types of prototypes, and the use of prototypes in various learning tasks. We discuss how the authors use prototype-based methods, their novel contributions, and any limitations or challenges that may arise when adapting these methods for geoscientific learning tasks. We highlight differences between geoscientific data sets and the standard benchmarks used to develop XAI methods, and discuss how specific geoscientific applications may benefit from using or modifying existing prototype-based XAI techniques., Comment: Accepted at AI for Science Workshop-Oral (Attention Track), Proceedings of 41st International Conference on Machine Learning (ICML) 2024

Protein inverse folding-that is, predicting an amino acid sequence that will fold into the desired 3D structure-is an important problem for structure-based protein design. Machine learning based methods for inverse folding typically use recovery of the original sequence as the optimization objective. However, inverse folding is a one-to-many problem where several sequences can fold to the same structure. Moreover, for many practical applications, it is often desirable to have multiple, diverse sequences that fold into the target structure since it allows for more candidate sequences for downstream optimizations. Here, we demonstrate that although recent inverse folding methods show increased sequence recovery, their "foldable diversity"-i.e. their ability to generate multiple non-similar sequences that fold into the structures consistent with the target-does not increase. To address this, we present RL-DIF, a categorical diffusion model for inverse folding that is pre-trained on sequence recovery and tuned via reinforcement learning on structural consistency. We find that RL-DIF achieves comparable sequence recovery and structural consistency to benchmark models but shows greater foldable diversity: experiments show RL-DIF can achieve an foldable diversity of 29% on CATH 4.2, compared to 23% from models trained on the same dataset. The PyTorch model weights and sampling code are available on GitHub.

To improve relevance scoring on Pinterest Search, we integrate Large Language Models (LLMs) into our search relevance model, leveraging carefully designed text representations to predict the relevance of Pins effectively. Our approach uses search queries alongside content representations that include captions extracted from a generative visual language model. These are further enriched with link-based text data, historically high-quality engaged queries, user-curated boards, Pin titles and Pin descriptions, creating robust models for predicting search relevance. We use a semi-supervised learning approach to efficiently scale up the amount of training data, expanding beyond the expensive human labeled data available. By utilizing multilingual LLMs, our system extends training data to include unseen languages and domains, despite initial data and annotator expertise being confined to English. Furthermore, we distill from the LLM-based model into real-time servable model architectures and features. We provide comprehensive offline experimental validation for our proposed techniques and demonstrate the gains achieved through the final deployed system at scale., Comment: CIKM 2024 Workshop on Industrial Recommendation Systems

Objectives: ncreased prevalence of social creak particularly among female speakers has been reported in several studies. The study of social creak has been previously conducted by combining perceptual evaluation of speech with conventional acoustical parameters such as the harmonic-to-noise ratio and cepstral peak prominence. In the current study, machine learning (ML) was used to automatically distinguish speech of low amount of social creak from speech of high amount of social creak. Methods: The amount of creak in continuous speech samples produced in Finnish by 90 female speakers was first perceptually assessed by two voice specialists. Based on their assessments, the speech samples were divided into two categories (low $vs$. high amount of creak). Using the speech signals and their creak labels, seven different ML models were trained. Three spectral representations were used as feature for each model. Results: The results show that the best performance (accuracy of 71.1\%) was obtained by the following two systems: an Adaboost classifier using the mel-spectrogram feature and a decision tree classifier using the mel-frequency cepstral coefficient feature. Conclusions: The study of social creak is becoming increasingly popular in sociolinguistic and vocological research. The conventional human perceptual assessment of the amount of creak is laborious and therefore ML technology could be used to assist researchers studying social creak. The classification systems reported in this study could be considered as baselines in future ML-based studies on social creak., Comment: Accepted in Journal of Voice

The paper introduces a novel framework for safe and autonomous aerial physical interaction in industrial settings. It comprises two main components: a neural network-based target detection system enhanced with edge computing for reduced onboard computational load, and a control barrier function (CBF)-based controller for safe and precise maneuvering. The target detection system is trained on a dataset under challenging visual conditions and evaluated for accuracy across various unseen data with changing lighting conditions. Depth features are utilized for target pose estimation, with the entire detection framework offloaded into low-latency edge computing. The CBF-based controller enables the UAV to converge safely to the target for precise contact. Simulated evaluations of both the controller and target detection are presented, alongside an analysis of real-world detection performance., Comment: 8 pages,14 figures, ICUAS 2024

Listeners of long-form talk-audio content, such as podcast episodes, often find it challenging to understand the overall structure and locate relevant sections. A practical solution is to divide episodes into chapters--semantically coherent segments labeled with titles and timestamps. Since most episodes on our platform at Spotify currently lack creator-provided chapters, automating the creation of chapters is essential. Scaling the chapterization of podcast episodes presents unique challenges. First, episodes tend to be less structured than written texts, featuring spontaneous discussions with nuanced transitions. Second, the transcripts are usually lengthy, averaging about 16,000 tokens, which necessitates efficient processing that can preserve context. To address these challenges, we introduce PODTILE, a fine-tuned encoder-decoder transformer to segment conversational data. The model simultaneously generates chapter transitions and titles for the input transcript. To preserve context, each input text is augmented with global context, including the episode's title, description, and previous chapter titles. In our intrinsic evaluation, PODTILE achieved an 11% improvement in ROUGE score over the strongest baseline. Additionally, we provide insights into the practical benefits of auto-generated chapters for listeners navigating episode content. Our findings indicate that auto-generated chapters serve as a useful tool for engaging with less popular podcasts. Finally, we present empirical evidence that using chapter titles can enhance effectiveness of sparse retrieval in search tasks., Comment: 9 pages, 4 figures, CIKM industry track 2024

The widespread adoption of large language models (LLMs) and generative AI (GenAI) tools across diverse applications has amplified the importance of addressing societal biases inherent within these technologies. While the NLP community has extensively studied LLM bias, research investigating how non-expert users perceive and interact with biases from these systems remains limited. As these technologies become increasingly prevalent, understanding this question is crucial to inform model developers in their efforts to mitigate bias. To address this gap, this work presents the findings from a university-level competition, which challenged participants to design prompts for eliciting biased outputs from GenAI tools. We quantitatively and qualitatively analyze the competition submissions and identify a diverse set of biases in GenAI and strategies employed by participants to induce bias in GenAI. Our finding provides unique insights into how non-expert users perceive and interact with biases from GenAI tools.

The Hausdorff distance measures how far apart two sets are in a common metric space. By contrast, the Gromov-Hausdorff distance provides a notion of distance between two abstract metric spaces. How do these distances behave for quotients of spaces under group actions? Suppose a group $G$ acts by isometries on two metric spaces $X$ and $Y$. In this article, we study how the Hausdorff and Gromov-Hausdorff distances between $X$ and $Y$ and their quotient spaces $X/G$ and $Y/G$ are related. For the Hausdorff distance, we show that if $X$ and $Y$ are $G$-invariant subsets of a common metric space, then we have $d_{\mathrm{H}}(X,Y)=d_{\mathrm{H}}(X/G,Y/G)$. However, the Gromov-Hausdorff distance does not preserve this relationship: we show how to make the ratio $\frac{d_{\mathrm{GH}}(X/G,Y/G)}{d_{\mathrm{GH}}(X,Y)}$ both arbitrarily large and arbitrarily small, even if $X$ is an arbitrarily dense $G$-invariant subset of $Y$.

Suppose we have a memory storing $0$s and $1$s and we want to estimate the frequency of $1$s by sampling. We want to do this I/O-efficiently, exploiting that each read gives a block of $B$ bits at unit cost; not just one bit. If the input consists of uniform blocks: either all 1s or all 0s, then sampling a whole block at a time does not reduce the number of samples needed for estimation. On the other hand, if bits are randomly permuted, then getting a block of $B$ bits is as good as getting $B$ independent bit samples. However, we do not want to make any such assumptions on the input. Instead, our goal is to have an algorithm with instance-dependent performance guarantees which stops sampling blocks as soon as we know that we have a probabilistically reliable estimate. We prove our algorithms to be instance-optimal among algorithms oblivious to the order of the blocks, which we argue is the strongest form of instance optimality we can hope for. We also present similar results for I/O-efficiently estimating mean with both additive and multiplicative error, estimating histograms, quantiles, as well as the empirical cumulative distribution function. We obtain our above results on I/O-efficient sampling by reducing to corresponding problems in the so-called sequential estimation. In this setting, one samples from an unknown distribution until one can provide an estimate with some desired error probability. We then provide non-parametric instance-optimal results for several fundamental problems: mean and quantile estimation, as well as learning mixture distributions with respect to $\ell_\infty$ and the so-called Kolmogorov-Smirnov distance., Comment: To appear at FOCS 2024

In this study, we investigate the under-explored intervention planning aimed at disseminating accurate information within dynamic opinion networks by leveraging learning strategies. Intervention planning involves identifying key nodes (search) and exerting control (e.g., disseminating accurate or official information through the nodes) to mitigate the influence of misinformation. However, as the network size increases, the problem becomes computationally intractable. To address this, we first introduce a ranking algorithm to identify key nodes for disseminating accurate information, which facilitates the training of neural network classifiers that provide generalized solutions for the search and planning problems. Second, we mitigate the complexity of label generation, which becomes challenging as the network grows, by developing a reinforcement learning-based centralized dynamic planning framework. We analyze these NN-based planners for opinion networks governed by two dynamic propagation models. Each model incorporates both binary and continuous opinion and trust representations. Our experimental results demonstrate that the ranking algorithm-based classifiers provide plans that enhance infection rate control, especially with increased action budgets for small networks. Further, we observe that the reward strategies focusing on key metrics, such as the number of susceptible nodes and infection rates, outperform those prioritizing faster blocking strategies. Additionally, our findings reveal that graph convolutional network-based planners facilitate scalable centralized plans that achieve lower infection rates (higher control) across various network configurations, including Watts-Strogatz topology, varying action budgets, varying initial infected nodes, and varying degrees of infected nodes., Comment: Accepted at NeurIPS 2024

In-context Learning (ICL) has become the primary method for performing natural language tasks with Large Language Models (LLMs). The knowledge acquired during pre-training is crucial for this few-shot capability, providing the model with task priors. However, recent studies have shown that ICL predominantly relies on retrieving task priors rather than "learning" to perform tasks. This limitation is particularly evident in complex subjective domains such as emotion and morality, where priors significantly influence posterior predictions. In this work, we examine whether this is the result of the aggregation used in corresponding datasets, where trying to combine low-agreement, disparate annotations might lead to annotation artifacts that create detrimental noise in the prompt. Moreover, we evaluate the posterior bias towards certain annotators by grounding our study in appropriate, quantitative measures of LLM priors. Our results indicate that aggregation is a confounding factor in the modeling of subjective tasks, and advocate focusing on modeling individuals instead. However, aggregation does not explain the entire gap between ICL and the state of the art, meaning other factors in such tasks also account for the observed phenomena. Finally, by rigorously studying annotator-level labels, we find that it is possible for minority annotators to both better align with LLMs and have their perspectives further amplified., Comment: 12 pages, 7 figures, 2 tables

Wearable sensors have become ubiquitous thanks to a variety of health tracking features. The resulting continuous and longitudinal measurements from everyday life generate large volumes of data; however, making sense of these observations for scientific and actionable insights is non-trivial. Inspired by the empirical success of generative modeling, where large neural networks learn powerful representations from vast amounts of text, image, video, or audio data, we investigate the scaling properties of sensor foundation models across compute, data, and model size. Using a dataset of up to 40 million hours of in-situ heart rate, heart rate variability, electrodermal activity, accelerometer, skin temperature, and altimeter per-minute data from over 165,000 people, we create LSM, a multimodal foundation model built on the largest wearable-signals dataset with the most extensive range of sensor modalities to date. Our results establish the scaling laws of LSM for tasks such as imputation, interpolation and extrapolation, both across time and sensor modalities. Moreover, we highlight how LSM enables sample-efficient downstream learning for tasks like exercise and activity recognition.

Given two nonincreasing $n$-tuples of real numbers $\lambda_n$, $\mu_n$, the Horn problem asks for a description of all nonincreasing $n$-tuples of real numbers $\nu_n$ such that there exist Hermitian matrices $X_n$, $Y_n$ and $Z_n$ respectively with these spectra such that $X_n + Y_n = Z_n$. There is also a randomized version of this problem where $X_n$ and $Y_n$ are sampled uniformly at random from orbits of Hermitian matrices arising from the conjugacy action by elements of the unitary group. One then asks for a description of the probability measure of the spectrum of the sum $Z_n$. Both the original Horn problem and its randomized version have solutions using the hives introduced by Knutson and Tao. In an asymptotic sense, as $n \rightarrow \infty$, large deviations for the randomized Horn problem were given by Narayanan and Sheffield in terms of the surface tension of hives. In this paper, we provide upper and lower bounds on this surface tension function. We also obtain a closed-form expression for the total entropy of a surface tension minimizing continuum hive with boundary conditions arising from GUE eigenspectra. Finally, we give several empirical results for random hives and lozenge tilings arising from an application of the octahedron recurrence for large $n$ and a numerical approximation of the surface tension function.

Large Language Model (LLM)-based applications are graduating from research prototypes to products serving millions of users, influencing how people write and consume information. A prominent example is the appearance of Answer Engines: LLM-based generative search engines supplanting traditional search engines. Answer engines not only retrieve relevant sources to a user query but synthesize answer summaries that cite the sources. To understand these systems' limitations, we first conducted a study with 21 participants, evaluating interactions with answer vs. traditional search engines and identifying 16 answer engine limitations. From these insights, we propose 16 answer engine design recommendations, linked to 8 metrics. An automated evaluation implementing our metrics on three popular engines (You.com, Perplexity.ai, BingChat) quantifies common limitations (e.g., frequent hallucination, inaccurate citation) and unique features (e.g., variation in answer confidence), with results mirroring user study insights. We release our Answer Engine Evaluation benchmark (AEE) to facilitate transparent evaluation of LLM-based applications.

The dearth of high quality spectroscopy of dusty star-forming galaxies (DSFGs) -- the main drivers of the assembly of dust and stellar mass at the peak of activity in the Universe -- greatly hinders our ability to interpret their physical processes and evolutionary pathways. We present JWST/NIRSpec observations from RUBIES of four submillimeter-selected, ALMA-detected DSFGs at cosmic noon, $z\sim2.3-2.7$. While photometry uniformly suggests vigorous ongoing star formation for the entire sample in line with canonical DSFGs, the spectra differ: one source has spectroscopic evidence of an evolved stellar population, indicating a recent transition to a post-starburst phase, while the remainder show strong spectroscopic signatures of ongoing starbursts. All four galaxies are infrared-luminous (log$_{10}$$L_{\rm{IR}}$/L$_{\rm \odot}$ $>12.4$), massive (log$_{10}\,M_\star$/M$_{\rm \odot}$ $>11$), and very dust-obscured ($A_V\sim3-4$ ABmag). Leveraging detections of multiple Balmer and Paschen lines, we derive an optical attenuation curve consistent with Calzetti overall, yet an optical extinction ratio $R_V\sim2.5$, potentially indicating smaller dust grains or differences in star-dust geometry. This case study provides some of the first detailed spectroscopic evidence that the DSFGs encompass a heterogeneous sample spanning a range of star formation properties and evolutionary stages, and illustrates the advantages of synergistic JWST and ALMA analysis of DSFGs., Comment: 20 pages, 4 figures; submitted to ApJ

We study the phases and phase transitions of a disordered one-dimensional quantum $q$-state clock Hamiltonian using large-scale Monte Carlo simulations. Making contact with earlier results, we confirm that the clean, translational invariant version of the model, for $q=6$, hosts an intermediate emergent quasi-long-range ordered (QLRO) phase between the symmetry-broken true long-range ordered (TLRO) phase and the disordered (paramagnetic) phase. With increasing disorder strength, the quasi-long-range ordered phase shrinks and finally vanishes at a multi-critical point, beyond which there is a direct transition from the TLRO phase to the paramagnetic phase. After establishing the phase diagram, we characterize the critical behaviors of the various quantum phase transitions in the model. We find that weak disorder is an irrelevant perturbation of the Berezinskii-Kosterlitz-Thouless transitions that separate the QLRO phase from the TLRO and paramagnetic phases. For stronger disorder, some of the critical exponents become disorder-dependent already before the system reaches the multicritical point. We also show that beyond the multicritical point, the direct transition from the TLRO phase to the paramagnetic phase is governed by an infinite-randomness critical point in line with strong-disorder renormalization group predictions. While our numerical results are for $q=6$, we expect the qualitative features of the behavior to hold for all $q>4$., Comment: 17 pages, 18 figures

Recent advances in text-to-speech (TTS) systems, particularly those with voice cloning capabilities, have made voice impersonation readily accessible, raising ethical and legal concerns due to potential misuse for malicious activities like misinformation campaigns and fraud. While synthetic speech detectors (SSDs) exist to combat this, they are vulnerable to ``test domain shift", exhibiting decreased performance when audio is altered through transcoding, playback, or background noise. This vulnerability is further exacerbated by deliberate manipulation of synthetic speech aimed at deceiving detectors. This work presents the first systematic study of such active malicious attacks against state-of-the-art open-source SSDs. White-box attacks, black-box attacks, and their transferability are studied from both attack effectiveness and stealthiness, using both hardcoded metrics and human ratings. The results highlight the urgent need for more robust detection methods in the face of evolving adversarial threats.

The first very long baseline interferometry (VLBI) detections at 870$\mu$m wavelength (345$\,$GHz frequency) are reported, achieving the highest diffraction-limited angular resolution yet obtained from the surface of the Earth, and the highest-frequency example of the VLBI technique to date. These include strong detections for multiple sources observed on inter-continental baselines between telescopes in Chile, Hawaii, and Spain, obtained during observations in October 2018. The longest-baseline detections approach 11$\,$G$\lambda$ corresponding to an angular resolution, or fringe spacing, of 19$\mu$as. The Allan deviation of the visibility phase at 870$\mu$m is comparable to that at 1.3$\,$mm on the relevant integration time scales between 2 and 100$\,$s. The detections confirm that the sensitivity and signal chain stability of stations in the Event Horizon Telescope (EHT) array are suitable for VLBI observations at 870$\mu$m. Operation at this short wavelength, combined with anticipated enhancements of the EHT, will lead to a unique high angular resolution instrument for black hole studies, capable of resolving the event horizons of supermassive black holes in both space and time., Comment: Corresponding author: S. Doeleman

Incompressible Navier-Stokes equations in the spherical coordinates are solved using a pseudo-spectral method to simulate the problem of spherical Couette flow. The flow is investigated for a narrow gap ratio with only the inner sphere rotating. We find that the flow is sensitive to the initial conditions and have used various initial conditions to obtain di!erent branches of the bifurcation curve of the flow. We have identified three di!erent branches dominated respectively by axisymmetric flow, traveling wave instability, and equatorial instability. The axisymmetric branch shows unsteadiness at large Reynolds numbers. The traveling wave instability branch shows spiral instability and is prominent near poles. The traveling wave instability branch further exhibits a reversal in the propagation direction of the spiral instability as the Reynolds number is increased. This branch also exhibits a multi-mode equatorial instability at larger Reynolds numbers. The equatorial instability branch exhibits twin jet streams on either side of the equator, which becomes unstable at larger Reynolds numbers. The flow topology on the three branches are also investigated in their phase space and the found to exhibit a chaotic behavior at large Reynolds numbers on the traveling wave instability branch., Comment: 22 pages

Sensory earables have evolved from basic audio enhancement devices into sophisticated platforms for clinical-grade health monitoring and wellbeing management. This paper introduces OmniBuds, an advanced sensory earable platform integrating multiple biosensors and onboard computation powered by a machine learning accelerator, all within a real-time operating system (RTOS). The platform's dual-ear symmetric design, equipped with precisely positioned kinetic, acoustic, optical, and thermal sensors, enables highly accurate and real-time physiological assessments. Unlike conventional earables that rely on external data processing, OmniBuds leverage real-time onboard computation to significantly enhance system efficiency, reduce latency, and safeguard privacy by processing data locally. This capability includes executing complex machine learning models directly on the device. We provide a comprehensive analysis of OmniBuds' design, hardware and software architecture demonstrating its capacity for multi-functional applications, accurate and robust tracking of physiological parameters, and advanced human-computer interaction.

We consider coverless steganography where a Large Language Model (LLM) drives an arithmetic coding decoder to generate stego-texts. An efficient method should embed secret message bits in as few language tokens as possible, while still keeping the stego-text natural and fluent. We show that on the individual token level, this problem is mathematically equivalent to maximizing the entropy of a replacement probability distribution of the next token generation, subject to a constraint on the KL divergence between the chosen probability distribution and the original distribution given by the LLM. A closed-form solution is provided for the optimization problem, which can be computed efficiently. Several important practical issues are also tackled: 1) An often-overlooked tokenization mismatch issue is resolved with a simple prompt selection approach, 2) The combination of the optimized distribution and the vocabulary truncation technique is considered, and 3) The combination of the optimized distribution with other sequence-level selection heuristics to further enhance the efficiency and reliability is studied., Comment: 9 figures

Decoherence of qubits during gate operations in the error correction phase of fault-tolerant computation is significant since, in practice, gate operations are neither instantaneous nor are they sufficiently parallelized. Thus, in practice, the overhead of a fault-tolerant scheme should depend on the level of parallelization, the speed of gate operations, and the rate of decoherence. In this paper, we obtain a lower bound on the required space overhead for any fault-tolerant technique when the decoherence process is Markov and the errors are either erasure or depolarization. An implication of this bound is that the noise or decoherence threshold, i.e., the noise beyond which no reliable memory (circuit) can be realized, decreases with a decrease in the level of parallelization in gate operations. Also, the noise threshold approaches zero as the level of parallelization approaches zero, implying that any fault-tolerant implementations must strive for sufficient parallelization in gate operations., Comment: 8 pages, 0 figures

Deploying large language models (LLMs) locally on mobile devices is advantageous in scenarios where transmitting data to remote cloud servers is either undesirable due to privacy concerns or impractical due to network connection. Recent advancements (MLC, 2023a; Gerganov, 2023) have facilitated the local deployment of LLMs. However, local deployment also presents challenges, particularly in balancing quality (generative performance), latency, and throughput within the hardware constraints of mobile devices. In this paper, we introduce our lightweight, all-in-one automated benchmarking framework that allows users to evaluate LLMs on mobile devices. We provide a comprehensive benchmark of various popular LLMs with different quantization configurations (both weights and activations) across multiple mobile platforms with varying hardware capabilities. Unlike traditional benchmarks that assess full-scale models on high-end GPU clusters, we focus on evaluating resource efficiency (memory and power consumption) and harmful output for compressed models on mobile devices. Our key observations include i) differences in energy efficiency and throughput across mobile platforms; ii) the impact of quantization on memory usage, GPU execution time, and power consumption; and iii) accuracy and performance degradation of quantized models compared to their non-quantized counterparts; and iv) the frequency of hallucinations and toxic content generated by compressed LLMs on mobile devices., Comment: 10 pages

Large ASR models can inadvertently leak sensitive information, which can be mitigated by formal privacy measures like differential privacy (DP). However, traditional DP training is computationally expensive, and can hurt model performance. Our study explores DP parameter-efficient fine-tuning as a way to mitigate privacy risks with smaller computation and performance costs for ASR models. Through extensive experimentation and progressive optimization, we achieve 4.6%/8.1% word error rate on LibriSpeech clean/other test-sets, setting a new performance benchmark while maintaining (10, 3.52e-6)-DP in fine-tuning a large ASR model with over 600M parameters.

The optimization of expensive black-box simulators arises in a myriad of modern scientific and engineering applications. Bayesian optimization provides an appealing solution, by leveraging a fitted surrogate model to guide the selection of subsequent simulator evaluations. In practice, however, the objective is often not to obtain a single good solution, but rather a ''basket'' of good solutions from which users can choose for downstream decision-making. This need arises in our motivating application for real-time control of internal combustion engines for flight propulsion, where a diverse set of control strategies is essential for stable flight control. There has been little work on this front for Bayesian optimization. We thus propose a new Diverse Expected Improvement (DEI) method that searches for diverse ''$\epsilon$-optimal'' solutions: locally-optimal solutions within a tolerance level $\epsilon > 0$ from a global optimum. We show that DEI yields a closed-form acquisition function under a Gaussian process surrogate model, which facilitates efficient sequential queries via automatic differentiation. This closed form further reveals a novel exploration-exploitation-diversity trade-off, which incorporates the desired diversity property within the well-known exploration-exploitation trade-off. We demonstrate the improvement of DEI over existing methods in a suite of numerical experiments, then explore the DEI in two applications on rover trajectory optimization and engine control for flight propulsion.

Hundreds of exoplanets between 1-1.8 times the size of the Earth have been discovered on close in orbits. However, these planets show such a diversity in densities that some appear to be made entirely of iron, while others appear to host gaseous envelopes. To test this diversity in composition, we update the masses of 5 rocky exoplanets (HD 93963 A b, Kepler-10 b, Kepler-100 b, Kepler-407 b, and TOI-1444 b) and present the confirmation of a new planet (TOI-1011) using 187 high precision RVs from Gemini/MAROON-X and Keck/KPF. Our updated planet masses suggest compositions closer to that of the Earth than previous literature values for all planets in our sample. In particular, we report that two previously identified ``super-Mercuries'' (Kepler-100 b and HD 93963 A b) have lower masses that suggest less iron-rich compositions. We then compare the ratio of iron to rock-building species to the abundance ratios of those elements in their host stars. These updated planet compositions do not suggest a steep relationship between planet and host star compositions, contradictory to previous results, and suggest that planets and host stars have similar abundance ratios., Comment: Submitted to AJ 09/30/2024

We present the properties of a massive, large, dusty, metal-rich, star-forming galaxy at z_spec=6.73. GOODSN-100182 was observed with JWST/NIRSpec as part of the AURORA survey, and is also covered by public multi-wavelength HST and JWST imaging. While the large mass of GOODSN-100182 (~10^10 M_sun) was indicated prior to JWST, NIRCam rest-optical imaging now reveals the presence of an extended disk (r_eff~1.5 kpc). In addition, the NIRSpec R~1000 spectrum of GOODSN-100182 includes the detection of a large suite of rest-optical nebular emission lines ranging in wavelength from [OII]3727 up to [NII]6583. The ratios of Balmer lines suggest significant dust attenuation (E(B-V)_gas=0.40+0.10/-0.09), consistent with the red rest-UV slope inferred for GOODSN-100182 (beta=-0.50+/-0.09). The star-formation rate based on dust-corrected H-alpha emission is log(SFR(H-alpha)/ M_sun/yr)=2.02+0.13/-0.14, well above the z~7 star-forming main sequence in terms of specific SFR. Strikingly, the ratio of [NII]6583/H-alpha emission suggests almost solar metallicity, as does the ratio ([OIII]5007/H-beta)/([NII]6583/H-alpha) and the detection of the faint [FeII]4360 emission feature, whereas the [OIII]5007/[OII]3727 ratio suggests roughly 50% solar metallicity. Overall, the excitation and ionization properties of GOODSN-100182 more closely resemble those of typical star-forming galaxies at z~2-3 rather than z~7. Based on public spectroscopy of the GOODS-N field, we find that GOODSN-100182 resides within a significant galaxy overdensity, and is accompanied by a spectroscopically-confirmed neighbor galaxy. GOODSN-100182 demonstrates the existence of mature, chemically-enriched galaxies within the first billion years of cosmic time, whose properties must be explained by galaxy formation models., Comment: 16 pages, 13 figures, submitted to ApJ

Large Language Models (LLMs) are widely used in both industry and academia for various tasks, yet evaluating the consistency of generated text responses continues to be a challenge. Traditional metrics like ROUGE and BLEU show a weak correlation with human judgment. More sophisticated metrics using Natural Language Inference (NLI) have shown improved correlations but are complex to implement, require domain-specific training due to poor cross-domain generalization, and lack explainability. More recently, prompt-based metrics using LLMs as evaluators have emerged; while they are easier to implement, they still lack explainability and depend on task-specific prompts, which limits their generalizability. This work introduces Automated eXplainable Consistency Evaluation using LLMs (AXCEL), a prompt-based consistency metric which offers explanations for the consistency scores by providing detailed reasoning and pinpointing inconsistent text spans. AXCEL is also a generalizable metric which can be adopted to multiple tasks without changing the prompt. AXCEL outperforms both non-prompt and prompt-based state-of-the-art (SOTA) metrics in detecting inconsistencies across summarization by 8.7%, free text generation by 6.2%, and data-to-text conversion tasks by 29.4%. We also evaluate the influence of underlying LLMs on prompt based metric performance and recalibrate the SOTA prompt-based metrics with the latest LLMs for fair comparison. Further, we show that AXCEL demonstrates strong performance using open source LLMs.

Deep Learning Recommendation Models (DLRMs) have become increasingly popular and prevalent in today's datacenters, consuming most of the AI inference cycles. The performance of DLRMs is heavily influenced by available bandwidth due to their large vector sizes in embedding tables and concurrent accesses. To achieve substantial improvements over existing solutions, novel approaches towards DLRM optimization are needed, especially, in the context of emerging interconnect technologies like CXL. This study delves into exploring CXL-enabled systems, implementing a process-in-fabric-switch (PIFS) solution to accelerate DLRMs while optimizing their memory and bandwidth scalability. We present an in-depth characterization of industry-scale DLRM workloads running on CXL-ready systems, identifying the predominant bottlenecks in existing CXL systems. We, therefore, propose PIFS-Rec, a PIFS-based scheme that implements near-data processing through downstream ports of the fabric switch. PIFS-Rec achieves a latency that is 3.89x lower than Pond, an industry-standard CXL-based system, and also outperforms BEACON, a state-of-the-art scheme, by 2.03x.

The ability to reliably transcribe child-adult conversations in a clinical setting is valuable for diagnosis and understanding of numerous developmental disorders such as Autism Spectrum Disorder. Recent advances in deep learning architectures and availability of large scale transcribed data has led to development of speech foundation models that have shown dramatic improvements in ASR performance. However, the ability of these models to translate well to conversational child-adult interactions is under studied. In this work, we provide a comprehensive evaluation of ASR performance on a dataset containing child-adult interactions from autism diagnostic sessions, using Whisper, Wav2Vec2, HuBERT, and WavLM. We find that speech foundation models show a noticeable performance drop (15-20% absolute WER) for child speech compared to adult speech in the conversational setting. Then, we employ LoRA on the best performing zero shot model (whisper-large) to probe the effectiveness of fine-tuning in a low resource setting, resulting in ~8% absolute WER improvement for child speech and ~13% absolute WER improvement for adult speech., Comment: 5 pages, 3 figures, 4 tables

Understanding speech production both visually and kinematically can inform second language learning system designs, as well as the creation of speaking characters in video games and animations. In this work, we introduce a data-driven method to visually represent articulator motion in Magnetic Resonance Imaging (MRI) videos of the human vocal tract during speech based on arbitrary audio or speech input. We leverage large pre-trained speech models, which are embedded with prior knowledge, to generalize the visual domain to unseen data using a speech-to-video diffusion model. Our findings demonstrate that the visual generation significantly benefits from the pre-trained speech representations. We also observed that evaluating phonemes in isolation is challenging but becomes more straightforward when assessed within the context of spoken words. Limitations of the current results include the presence of unsmooth tongue motion and video distortion when the tongue contacts the palate., Comment: 4 pages

Large Language Models (LLMs) show promise in generating code comments for novice programmers, but their educational effectiveness remains under-evaluated. This study assesses the instructional quality of code comments produced by GPT-4, GPT-3.5-Turbo, and Llama2, compared to expert-developed comments, focusing on their suitability for novices. Analyzing a dataset of ``easy'' level Java solutions from LeetCode, we find that GPT-4 exhibits comparable quality to expert comments in aspects critical for beginners, such as clarity, beginner-friendliness, concept elucidation, and step-by-step guidance. GPT-4 outperforms Llama2 in discussing complexity (chi-square = 11.40, p = 0.001) and is perceived as significantly more supportive for beginners than GPT-3.5 and Llama2 with Mann-Whitney U-statistics = 300.5 and 322.5, p = 0.0017 and 0.0003). This study highlights the potential of LLMs for generating code comments tailored to novice programmers.

Self-supervised learning (SSL) methods for large speech models have proven to be highly effective at ASR. With the interest in public deployment of large pre-trained models, there is a rising concern for unintended memorization and leakage of sensitive data points from the training data. In this paper, we apply differentially private (DP) pre-training to a SOTA Conformer-based encoder, and study its performance on a downstream ASR task assuming the fine-tuning data is public. This paper is the first to apply DP to SSL for ASR, investigating the DP noise tolerance of the BEST-RQ pre-training method. Notably, we introduce a novel variant of model pruning called gradient-based layer freezing that provides strong improvements in privacy-utility-compute trade-offs. Our approach yields a LibriSpeech test-clean/other WER (%) of 3.78/ 8.41 with ($10$, 1e^-9)-DP for extrapolation towards low dataset scales, and 2.81/ 5.89 with (10, 7.9e^-11)-DP for extrapolation towards high scales., Comment: In proceedings of the IEEE Spoken Language Technologies Workshop, 2024

A Machine Learning (ML) network based on transfer learning and transformer networks is applied to wave propagation models for complex indoor settings. This network is designed to predict signal propagation in environments with a variety of objects, effectively simulating the diverse range of furniture typically found in indoor spaces. We propose Attention U-Net with Efficient Networks as the backbone, to process images encoded with the essential information of the indoor environment. The indoor environment is defined by its fundamental structure, such as the arrangement of walls, windows, and doorways, alongside varying configurations of furniture placement. An innovative algorithm is introduced to generate a 3D environment from a 2D floorplan, which is crucial for efficient collection of data for training. The model is evaluated by comparing the predicted signal coverage map with ray tracing (RT) simulations. The prediction results show a root mean square error of less than 6 dB across all tested scenarios, with significant improvements observed when using a Double U-Net structure compared to a single U-Net model.