Search

The exponential growth of Advanced Air Mobility (AAM) services demands assurances of safety in the airspace. This research a Traffic Control Framework (TCF) for developing digital flight rules for Uncrewed Aircraft System (UAS) flying in designated air corridors. The proposed TCF helps model, deploy, and test UAS control, agents, regardless of their hardware configurations. This paper investigates the importance of digital flight rules in preventing collisions in the context of AAM. TCF is introduced as a platform for developing strategies for managing traffic towards enhanced autonomy in the airspace. It allows for assessment and evaluation of autonomous navigation, route planning, obstacle avoidance, and adaptive decision making for UAS. It also allows for the introduction and evaluation of advance technologies Artificial Intelligence (AI) and Machine Learning (ML) in a simulation environment before deploying them in the real world. TCF can be used as a tool for comprehensive UAS traffic analysis, including KPI measurements. It offers flexibility for further testing and deployment laying the foundation for improved airspace safety - a vital aspect of UAS technological advancement. Finally, this papers demonstrates the capabilities of the proposed TCF in managing UAS traffic at intersections and its impact on overall traffic flow in air corridors, noting the bottlenecks and the inverse relationship safety and traffic volume., Comment: 6 pages, 7 figures

This work is a numerical study of the two-dimensional merging phenomena between two Lamb-Oseen co-rotating vortices in a viscoelastic fluid. We use a generalized hydrodynamics fluid model to study vortex merging in a strongly coupled dusty plasma medium, which exhibits characteristics similar to a viscoelastic fluid. Several aspects influencing the merging phenomena are considered: the aspect ratio (core size/separation distance), the relative circulation strengths of each vortex, and the coupling strength of the medium. Unlike classical hydrodynamic fluids, we find that for viscoelastic fluids, shear waves facilitate the merging events even for widely separated vortices. The merging process is accelerated in media with higher coupling strengths, but the resultant vortex shape decays more quickly as well. It is also found that varying either the vortex scale or the vortex circulation strength can result in a similar merging process, where a smaller (larger) vortex acts like a vortex with weaker (stronger) circulation. Finally, we show that a Poynting-like conservation theorem is satisfied for the examined merging processes.

Lattices are architected metamaterials whose properties strongly depend on their geometrical design. The analogy between lattices and graphs enables the use of graph neural networks (GNNs) as a faster surrogate model compared to traditional methods such as finite element modelling. In this work, we generate a big dataset of structure-property relationships for strut-based lattices. The dataset is made available to the community which can fuel the development of methods anchored in physical principles for the fitting of fourth-order tensors. In addition, we present a higher-order GNN model trained on this dataset. The key features of the model are (i) SE(3) equivariance, and (ii) consistency with the thermodynamic law of conservation of energy. We compare the model to non-equivariant models based on a number of error metrics and demonstrate its benefits in terms of predictive performance and reduced training requirements. Finally, we demonstrate an example application of the model to an architected material design task. The methods which we developed are applicable to fourth-order tensors beyond elasticity such as piezo-optical tensor etc., Comment: International Conference on Learning Representations 2024

This paper considers the problem of the private release of sample means of speed values from traffic datasets. Our key contribution is the development of user-level differentially private algorithms that incorporate carefully chosen parameter values to ensure low estimation errors on real-world datasets, while ensuring privacy. We test our algorithms on ITMS (Intelligent Traffic Management System) data from an Indian city, where the speeds of different buses are drawn in a potentially non-i.i.d. manner from an unknown distribution, and where the number of speed samples contributed by different buses is potentially different. We then apply our algorithms to large synthetic datasets, generated based on the ITMS data. Here, we provide theoretical justification for the observed performance trends, and also provide recommendations for the choices of algorithm subroutines that result in low estimation errors. Finally, we characterize the best performance of pseudo-user creation-based algorithms on worst-case datasets via a minimax approach; this then gives rise to a novel procedure for the creation of pseudo-users, which optimizes the worst-case total estimation error. The algorithms discussed in the paper are readily applicable to general spatio-temporal IoT datasets for releasing a differentially private mean of a desired value., Comment: 14 pages, 5 figures, submitted to the ACM for possible publication

Unequal technology access for Black and Latine communities has been a persistent economic, social justice, and human rights issue despite increased technology accessibility due to advancements in consumer electronics like phones, tablets, and computers. We contextualize socio-technical access inequalities for Black and Latine urban communities and find that many students are hesitant to engage with available technologies due to a lack of engaging support systems. We present a holistic student-led STEM engagement model through AVELA - A Vision for Engineering Literacy and Access leveraging culturally responsive lessons, mentor embodied community representation, and service learning. To evaluate the model's impact after 4 years of mentoring 200+ university student instructors in teaching to 2,500+ secondary school students in 100+ classrooms, we conducted 24 semi-structured interviews with college AnonymizedOrganization members. We identify access barriers and provide principled recommendations for designing future STEM education programs., Comment: This is the author's version of the work. It is posted here for personal use, not for redistribution

Autonomous agents capable of planning, reasoning, and executing actions on the web offer a promising avenue for automating computer tasks. However, the majority of existing benchmarks primarily focus on text-based agents, neglecting many natural tasks that require visual information to effectively solve. Given that most computer interfaces cater to human perception, visual information often augments textual data in ways that text-only models struggle to harness effectively. To bridge this gap, we introduce VisualWebArena, a benchmark designed to assess the performance of multimodal web agents on realistic \textit{visually grounded tasks}. VisualWebArena comprises of a set of diverse and complex web-based tasks that evaluate various capabilities of autonomous multimodal agents. To perform on this benchmark, agents need to accurately process image-text inputs, interpret natural language instructions, and execute actions on websites to accomplish user-defined objectives. We conduct an extensive evaluation of state-of-the-art LLM-based autonomous agents, including several multimodal models. Through extensive quantitative and qualitative analysis, we identify several limitations of text-only LLM agents, and reveal gaps in the capabilities of state-of-the-art multimodal language agents. VisualWebArena provides a framework for evaluating multimodal autonomous language agents, and offers insights towards building stronger autonomous agents for the web. Our code, baseline models, and data is publicly available at https://jykoh.com/vwa., Comment: Accepted to ACL 2024. 24 pages. Project page: https://jykoh.com/vwa

We microscopically derive a lattice abelian mutual Chern-Simons gauge theory for a honeycomb Kitaev model subjected to $(001)$ Zeeman and three-spin scalar spin chirality perturbations. We identify the nature of topological orders, emergent excitations and ground state degeneracy (GSD), topological entanglement entropy ($\gamma$), and chiral central charge ($c$) in different field regimes for both ferromagnetic (FM) and antiferromagnetic (AFM) sign of the Kitaev interaction. A nonabelian Ising topological order (ITO) exists at low fields in both cases, with $\gamma=\ln 2,$ $c=1/2,$ and GSD$=3,$ where the nonabelian anyon, a twist defect, is an intrinsic bulk excitation. For AFM Kitaev interactions, further increase of the field causes a transition from ITO to an intermediate trivial topological phase with central charge $c=1/2,$ implying half-quantized thermal Hall response in both phases with no change of sign. At sufficiently high fields there is a first order transition to a polarized paramagnetic phase with $\gamma=c=0.$ For the FM case, there is a direct transition from ITO to the polarized phase., Comment: 9 pages,5 figures,1 table

We study ordinal rules for allocating indivisible goods via lottery. Ordinality requires a rule to consider only how agents rank degenerate lotteries and may be necessitated by cognitive, informational, or as we show, incentive constraints. The limited responsiveness of ordinal rules to agents' preferences means that they can only satisfy welfare properties based on first order stochastic dominance, which is incomplete. We define a new efficiency concept for ordinal rules. While ordinality and efficiency together are incompatible with the usual notions of fairness and somewhat limit randomization, they do leave room for a rich class of rules. We demonstrate this through a characterization of all ordinal, efficient, strategy-proof, non-bossy, boundedly invariant, and neutral rules.

Rydberg atom arrays have emerged as a leading platform for quantum information science. Reaching system sizes of hundreds of long-lived qubits, these arrays are used for highly coherent analog quantum simulation, as well as digital quantum computation. Advanced quantum protocols such as quantum error correction, however, require midcircuit qubit operations, including the replenishment, reset, and readout of a subset of qubits. A compelling strategy to achieve these capabilities is a dual-species architecture in which a second atomic species can be controlled without crosstalk, and entangled with the first via Rydberg interactions. Here, we realize a dual-species Rydberg array consisting of rubidium (Rb) and cesium (Cs) atoms, and explore new regimes of interactions and dynamics not accessible in single-species architectures. We achieve enhanced interspecies interactions by electrically tuning the Rydberg states close to a Forster resonance. In this regime, we demonstrate interspecies Rydberg blockade and implement quantum state transfer from one species to another. We then generate a Bell state between Rb and Cs hyperfine qubits via an interspecies controlled-phase gate. Finally, we combine interspecies entanglement with native midcircuit readout to achieve quantum non-demolition measurement of a Rb qubit using an auxiliary Cs qubit. The techniques demonstrated here pave the way toward scalable measurement-based protocols and real-time feedback control in large-scale quantum systems.

The evaluation of Fock exchange is often the computationally most expensive part of hybrid functional density functional theory calculations in a systematically improvable, complete basis. In this work, we employ a Tucker tensor based approach that substantially accelerates the evaluation of the action of Fock exchange by transforming 3-dimensional convolutional integrals into a tensor product of 1-dimensional convolution integrals. Our numerical implementation uses a parallelization strategy that balances the memory and communication bottlenecks, alongside overalapping compute and communication operations to enhance computational efficiency and parallel scalability. The accuracy and computational efficiency is demonstrated on various systems, including Pt clusters of various sizes and a $\text{TiO}_{\text{2}}$ cluster with 3,684 electrons., Comment: 30 pages, 10 figures

The mechanical response of rubbers has been ubiquitously assumed to be only a function of the imposed strain. Using innovative X-ray measurements capturing the three-dimensional spatial volumetric strain fields, we demonstrate that rubbers and indeed many common engineering polymers, undergo significant local volume changes. But remarkably the overall specimen volume remains constant regardless of the imposed loading. This strange behaviour which also leads to apparent negative local bulk moduli is due to the presence of a mobile phase within these materials. Combining X-ray tomographic observations with high-speed radiography to track the motion of the mobile phase we have revised classical thermodynamic frameworks of rubber elasticity. The work opens new avenues to understand not only the mechanical behaviour of rubbers but a large class of widely used engineering polymers., Comment: 10 pages of main text, 4 main figures, 4 extended data figures

Spectral statistics such as the level spacing statistics and spectral form factor (SFF) are widely expected to accurately identify ``ergodicity'', including the presence of underlying macroscopic symmetries, in generic quantum systems ranging from quantized chaotic maps to interacting many-body systems. By studying various quantizations of maximally chaotic maps that break a discrete classical symmetry upon quantization, we demonstrate that this approach can be misleading and fail to detect macroscopic symmetries. Notably, the same classical map can exhibit signatures of different random matrix symmetry classes in short-range spectral statistics depending on the quantization. While the long-range spectral statistics encoded in the early time ramp of the SFF are more robust and correctly identify macroscopic symmetries in several common quantizations, we also demonstrate analytically and numerically that the presence of Berry-like phases in the quantization leads to spectral anomalies, which break this correspondence. Finally, we provide numerical evidence that long-range spectral rigidity remains directly correlated with ergodicity in the quantum dynamical sense of visiting a complete orthonormal basis., Comment: 23 pages, 16 figures

Recently, an atomic-scale two-dimensional silicon carbide monolayer has been synthesized {[}Polley \emph{et al., }Phys. Rev. Lett. \textbf{130},076203 (2023){]} which opens up new possibilities for developing next-generation electronic and optoelectronic devices. Our study predicts the pristine SiC monolayer to have an ``indirect'' band gap of 3.38 eV $(K\rightarrow M)$ and a ``direct'' band gap of 3.43 eV $(K\rightarrow K)$ calculated using the HSE06 functional. We performed a detailed investigation of the various possible defects (i.e., vacancies, foreign impurities, antisites, and their various combinations) on the structural stability, electronic, and optical properties of the SiC monolayer using a first-principles based density-functional theory (DFT) and molecular dynamics (MD) simulations. A number of physical quantities such as the formation energy, electronic band gap, and the effective masses of charge carriers, have been calculated. We report that the SiC monolayer has a very low formation energy of 0.57 eV and can be stabilized on TaC \{111\} film by performing the surface slab energy and interfacial adhesion energy calculations. Nitrogen doping is predicted to be the most favorable defect in silicon carbide monolayer due to its very low formation energy, indicating high thermodynamic stability. An interesting transition from semiconducting to metallic state is observed for $N_{C}$ and $Al_{Si}$ defective systems. For the pristine SiC monolayer, we find that the conduction band is nearly flat in the $M\rightarrow K$ direction, leading to a high effective mass of $3.48m_{o}$. A significant red shift in the absorption edge, as well as the occurrence of additional absorption peaks due to the defects, have been observed in the lower energy range of the spectrum., Comment: 34 pages, 13 figures (manuscript 31 pages, 10 figures + supplemental material 3 pages, 3 figures)

We introduce a novel definition for a small set R of k points being "representative" of a larger set in a metric space. Given a set V (e.g., documents or voters) to represent, and a set C of possible representatives, our criterion requires that for any subset S comprising a theta fraction of V, the average distance of S to their best theta*k points in R should not be more than a factor gamma compared to their average distance to the best theta*k points among all of C. This definition is a strengthening of proportional fairness and core fairness, but - different from those notions - requires that large cohesive clusters be represented proportionally to their size. Since there are instances for which - unless gamma is polynomially large - no solutions exist, we study this notion in a resource augmentation framework, implicitly stating the constraints for a set R of size k as though its size were only k/alpha, for alpha > 1. Furthermore, motivated by the application to elections, we mostly focus on the "ordinal" model, where the algorithm does not learn the actual distances; instead, it learns only for each point v in V and each candidate pairs c, c' which of c, c' is closer to v. Our main result is that the Expanding Approvals Rule (EAR) of Aziz and Lee is (alpha, gamma) representative with gamma <= 1 + 6.71 * (alpha)/(alpha-1). Our results lead to three notable byproducts. First, we show that the EAR achieves constant proportional fairness in the ordinal model, giving the first positive result on metric proportional fairness with ordinal information. Second, we show that for the core fairness objective, the EAR achieves the same asymptotic tradeoff between resource augmentation and approximation as the recent results of Li et al., which used full knowledge of the metric. Finally, our results imply a very simple single-winner voting rule with metric distortion at most 44., Comment: 24 pages, Accepted to AAAI 24

Transcriptomic data is a treasure-trove in modern molecular biology, as it offers a comprehensive viewpoint into the intricate nuances of gene expression dynamics underlying biological systems. This genetic information must be utilised to infer biomolecular interaction networks that can provide insights into the complex regulatory mechanisms underpinning the dynamic cellular processes. Gene regulatory networks and protein-protein interaction networks are two major classes of such networks. This chapter thoroughly investigates the wide range of methodologies used for distilling insightful revelations from transcriptomic data that include association based methods (based on correlation among expression vectors), probabilistic models (using Bayesian and Gaussian models), and interologous methods. We reviewed different approaches for evaluating the significance of interactions based on the network topology and biological functions of the interacting molecules, and discuss various strategies for the identification of functional modules. The chapter concludes with highlighting network based techniques of prioritising key genes, outlining the centrality based, diffusion based and subgraph based methods. The chapter provides a meticulous framework for investigating transcriptomic data to uncover assembly of complex molecular networks for their adaptable analyses across a broad spectrum of biological domains., Comment: 48 pages, 3 figures

We propose a reinforcement learning (RL)-based system that would automatically prescribe a hypothetical patient medication that may help the patient with their mental health-related speech disfluency, and adjust the medication and the dosages in response to zero-cost frequent measurement of the fluency of the patient. We demonstrate the components of the system: a module that detects and evaluates speech disfluency on a large dataset we built, and an RL algorithm that automatically finds good combinations of medications. To support the two modules, we collect data on the effect of psychiatric medications for speech disfluency from the literature, and build a plausible patient simulation system. We demonstrate that the RL system is, under some circumstances, able to converge to a good medication regime. We collect and label a dataset of people with possible speech disfluency and demonstrate our methods using that dataset. Our work is a proof of concept: we show that there is promise in the idea of using automatic data collection to address speech disfluency., Comment: In Proc. Machine Learning for Cognitive and Mental Health Workshop (ML4CMH) at AAAI 2024

Abstractive text summarization is surging with the number of training samples to cater to the needs of the deep learning models. These models tend to exploit the training data representations to attain superior performance by improving the quantitative element of the resultant summary. However, increasing the size of the training set may not always be the ideal solution to maximize the performance, and therefore, a need to revisit the quality of training samples and the learning protocol of deep learning models is a must. In this paper, we aim to discretize the vector space of the abstractive text summarization models to understand the characteristics learned between the input embedding space and the models' encoder space. We show that deep models fail to capture the diversity of the input space. Further, the distribution of data points on the encoder space indicates that an unchecked increase in the training samples does not add value; rather, a tear-down of data samples is highly needed to make the models focus on variability and faithfulness. We employ clustering techniques to learn the diversity of a model's sample space and how data points are mapped from the embedding space to the encoder space and vice versa. Further, we devise a metric to filter out redundant data points to make the model more robust and less data hungry. We benchmark our proposed method using quantitative metrics, such as Rouge, and qualitative metrics, such as BERTScore, FEQA and Pyramid score. We also quantify the reasons that inhibit the models from learning the diversity from the varied input samples.

Advanced microelectronics in the future may require semiconducting channel materials beyond silicon. Two-dimensional (2D) semiconductors, characterized by their atomically thin thickness, hold immense promise for high-performance electronic devices at the nanometer scale with lower heat dissipation. One challenge for achieving high-performance 2D semiconductor field effect transistors (FET), especially for p-type materials, is the high electrical contact resistance present at the metal-semiconductor interface. In conventional bulk semiconductors, low resistance ohmic contact is realized through heavy substitutional doping with acceptor or donor impurities at the contact region. The strategy of substitutional doping, however, does not work for p-type 2D semiconductors such as monolayer tungsten diselenide (WSe$_2$).In this study, we developed highly efficient charge-transfer doping with WSe$_2$/$\alpha$-RuCl$_3$ heterostructures to achieve low-resistance ohmic contact for p-type WSe$_2$ transistors. We show that a hole doping as high as 3$\times$10$^{13}$ cm$^{-2}$ can be achieved in the WSe$_2/\alpha$-RuCl$_3$ heterostructure due to its type-III band alignment. It results in an Ohmic contact with resistance lower than 4 k Ohm $\mu$m at the p-type monolayer WSe$_2$/metal junction. at room temperature. Using this low-resistance contact, we demonstrate high-performance p-type WSe$_2$ transistors with a saturation current of 35 $\mu$A$\cdot$ $\mu$m$^{-1}$ and an I$_{ON}$/I$_{OFF}$ ratio exceeding 10$^9$ It could enable future microelectronic devices based on 2D semiconductors and contribute to the extension of Moore's law.

Two-dimensional porous carbon nanomaterials are proven to be promising hydrogen storage substrates as they possess high surface area, large number of active sites, low molecular mass, and hydrogen molecules can be adsorbed on both sides of these materials. By performing first-principles density functional theory-based calculations, we report ultrahigh reversible hydrogen uptake in lithium decorated 2D carbon allotrope PAI-graphene, which is formed of a regular pattern of polymerized as-indacenes (PAI). We found that a single unit cell of PAI-graphene can be decorated by 8 Li atoms, in which each Li atom can reversibly adsorb 4 hydrogen molecules, leading to 15.7 % of H uptake, remarkably higher than the DOE demand of 6.5 %. Li atom donates its valence 2s-electron to PAI-graphene and gets ionized. The adsorption energies of the various H2 attached to Li-atom are found to be suitable for reversible use during practical applications. Hydrogen molecules get attached to the ionized metal atom by electrostatic interactions. An energy barrier of 1.48 eV is present for the diffusion of Li atoms between the two most stable adsorption sites which justifies the absence of the clustering of Li atoms., Comment: 28 Pages, 8 Figures, 1 table

An intriguing phenomenon in molecular collisions is the occurrence of scattering resonances, which originate from bound and quasi-bound states supported by the interaction potential at low collision energies. The resonance effects in the scattering behaviour are extraordinarily sensitive to the interaction potential, and their observation provides one of the most stringent tests for theoretical models. We present high-resolution measurements of state-resolved angular scattering distributions for inelastic collisions between Zeeman-decelerated C($^3P_1$) atoms and $\textit{para}$-H$_2$ molecules at collision energies ranging from 77 cm$^{-1}$ down to 0.5 cm$^{-1}$. Rapid variations in the angular distributions were observed that can be attributed to the consecutive reduction of contributing partial waves and effects of scattering resonances. The measurements showed excellent agreement with distributions predicted by $\textit{ab initio}$ quantum scattering calculations. However, discrepancies were found at specific collision energies, which most likely originate from an incorrectly predicted quasi-bound state. These observations provide exciting prospects for further high-precision and low-energy investigations of scattering processes that involve paramagnetic species.

We perform the first measurement of the thermal and ionization state of the intergalactic medium (IGM) across 0.9 < z < 1.5 using 301 \lya absorption lines fitted from 12 HST STIS quasar spectra, with a total pathlength of \Delta z=2.1. We employ the machine-learning-based inference method that uses joint b-N distributions obtained from \lyaf decomposition. Our results show that the HI photoionization rates, \Gamma, are in good agreement with the recent UV background synthesis models, with \log (\Gamma/s^{-1})={-11.79}^{0.18}_{-0.15}, -11.98}^{0.09}_{-0.09}, and {-12.32}^{0.10}_{-0.12} at z=1.4, 1.2, and 1 respectively. We obtain the IGM temperature at the mean density, T_0, and the adiabatic index, \gamma, as [\log (T_0/K), \gamma]= [{4.13}^{+0.12}_{-0.10}, {1.34}^{+0.10}_{-0.15}], [{3.79}^{+0.11}_{-0.11}, {1.70}^{+0.09}_{-0.09}] and [{4.12}^{+0.15}_{-0.25}, {1.34}^{+0.21}_{-0.26}] at z=1.4, 1.2 and 1 respectively. Our measurements of T_0 at z=1.4 and 1.2 are consistent with the expected trend from z<3 temperature measurements as well as theoretical expectations that, in the absence of any non-standard heating, the IGM should cool down after HeII reionization. Whereas, our T_0 measurements at z=1 show unexpectedly high IGM temperature. However, because of the relatively large uncertainty in these measurements of the order of \Delta T_0~5000 K, mostly emanating from the limited redshift path length of available data in these bins, we can not definitively conclude whether the IGM cools down at z<1.5. Lastly, we generate a mock dataset to test the constraining power of future measurement with larger datasets. The results demonstrate that, with redshift pathlength \Delta z \sim 2 for each redshift bin, three times the current dataset, we can constrain the T_0 of IGM within 1500K. Such precision would be sufficient to conclusively constrain the history of IGM thermal evolution at z < 1.5.

Large language models (LLMs) solve problems more accurately and interpretably when instructed to work out the answer step by step using a ``chain-of-thought'' (CoT) prompt. One can also improve LLMs' performance on a specific task by supervised fine-tuning, i.e., by using gradient ascent on some tunable parameters to maximize the average log-likelihood of correct answers from a labeled training set. Naively combining CoT with supervised tuning requires supervision not just of the correct answers, but also of detailed rationales that lead to those answers; these rationales are expensive to produce by hand. Instead, we propose a fine-tuning strategy that tries to maximize the \emph{marginal} log-likelihood of generating a correct answer using CoT prompting, approximately averaging over all possible rationales. The core challenge is sampling from the posterior over rationales conditioned on the correct answer; we address it using a simple Markov-chain Monte Carlo (MCMC) expectation-maximization (EM) algorithm inspired by the self-taught reasoner (STaR), memoized wake-sleep, Markovian score climbing, and persistent contrastive divergence. This algorithm also admits a novel control-variate technique that drives the variance of our gradient estimates to zero as the model improves. Applying our technique to GSM8K and the tasks in BIG-Bench Hard, we find that this MCMC-EM fine-tuning technique typically improves the model's accuracy on held-out examples more than STaR or prompt-tuning with or without CoT., Comment: 23 pages, 37th Conference on Neural Information Processing Systems (NeurIPS 2023)

Passively collected behavioral health data from ubiquitous sensors holds significant promise to provide mental health professionals insights from patient's daily lives; however, developing analysis tools to use this data in clinical practice requires addressing challenges of generalization across devices and weak or ambiguous correlations between the measured signals and an individual's mental health. To address these challenges, we take a novel approach that leverages large language models (LLMs) to synthesize clinically useful insights from multi-sensor data. We develop chain of thought prompting methods that use LLMs to generate reasoning about how trends in data such as step count and sleep relate to conditions like depression and anxiety. We first demonstrate binary depression classification with LLMs achieving accuracies of 61.1% which exceed the state of the art. While it is not robust for clinical use, this leads us to our key finding: even more impactful and valued than classification is a new human-AI collaboration approach in which clinician experts interactively query these tools and combine their domain expertise and context about the patient with AI generated reasoning to support clinical decision-making. We find models like GPT-4 correctly reference numerical data 75% of the time, and clinician participants express strong interest in using this approach to interpret self-tracking data.

Commensurability phenomena abound in nature and are typically associated with mismatched lengths, as can occur in quasiperiodic systems. However, not all commensuration effects are spatial in nature. In finite-sized Dirac systems, an intriguing example arises in tilted or warped Dirac cones wherein the degeneracy in the speed of right- and left-moving electrons within a given Dirac cone or valley is lifted. Bound states can be purely fast-moving or purely slow-moving, giving rise to incommensurate energy level spacings and a vernier spectrum. In this work, we present evidence for this vernier spectrum in Coulomb blockade measurements of ultraclean suspended carbon nanotube quantum dots. The addition-energy spectrum of the quantum dots reveals an energy-level structure that oscillates between aligned and misaligned energy levels. Our data suggest that the fast- and slow-moving bound states hybridize at certain gate voltages. Thus, gate-voltage tuning can select states with varying degrees of hybridization, suggesting numerous applications based on accessing this isospin-like degree of freedom., Comment: 14 pages including 3 figures

AI models are constantly evolving, with new versions released frequently. This raises a key question: how should AI-infused systems integrate updates when the downstream impact on user experience and performance is unclear? Human-AI interaction guidelines encourage notifying users about (changes in) model capabilities, ideally supported by thorough benchmarking. Yet, as AI models integrate into domain-specific workflows, exhaustive benchmarking can become impractical or expensive, often leading to invisible or minimally communicated updates. In this work, we explore the impact of such updates through two complementary studies on facial recognition for historical person identification. First, we conducted an online experiment to understand how users distinguish between models, followed by a diary study examining user perceptions in a real-world deployment. Our findings reveal how model changes impact human-AI performance, downstream user behavior, and the folk theories they develop. Based on these insights, we discuss implications for updating models in AI-infused systems.

Reducing the environmental footprint of electronics and computing devices requires new tools that empower designers to make informed decisions about sustainability during the design process itself. This is not possible with current tools for life cycle assessment (LCA) which require substantial domain expertise and time to evaluate the numerous chips and other components that make up a device. We observe first that informed decision-making does not require absolute metrics and can instead be done by comparing designs. Second, we can use domain-specific heuristics to perform these comparisons. We combine these insights to develop DeltaLCA, an open-source interactive design tool that addresses the dual challenges of automating life cycle inventory generation and data availability by performing comparative analyses of electronics designs. Users can upload standard design files from Electronic Design Automation (EDA) software and the tool will guide them through determining which one has greater carbon footprint. DeltaLCA leverages electronics-specific LCA datasets and heuristics and tries to automatically rank the two designs, prompting users to provide additional information only when necessary. We show through case studies DeltaLCA achieves the same result as evaluating full LCAs, and that it accelerates LCA comparisons from eight expert-hours to a single click for devices with ~30 components, and 15 minutes for more complex devices with ~100 components.

We explore the potential of using the low-redshift Lyman-$\alpha$ (Ly$\alpha$) forest surrounding luminous red galaxies (LRGs) as a tool to constrain active galactic nuclei (AGN) feedback models. Our analysis is based on snapshots from the Illustris and IllustrisTNG simulations at a redshift of $z=0.1$. These simulations offer an ideal platform for studying the influence of AGN feedback on the gas surrounding galaxies, as they share the same initial conditions and underlying code but incorporate different feedback prescriptions. Both simulations show significant impacts of feedback on the temperature and density of the gas around massive halos. Following our previous work, we adjusted the UV background in both simulations to align with the observed number density of Ly$\alpha$ lines ($\rm dN/dz$) in the intergalactic medium and study the Ly$\alpha$ forest around massive halos hosting LRGs, at impact parameters ($r_{\perp}$) ranging from 0.1 to 100 pMpc. Our findings reveal that $\rm dN/dz$, as a function of $r_{\perp}$, is approximately 1.5 to 2 times higher in IllustrisTNG compared to Illustris up to $r_{\perp}$ of $\sim 10$ pMpc. To further assess whether existing data can effectively discern these differences, we search for archival data containing spectra of background quasars probing foreground LRGs. Through a feasibility analysis based on this data, we demonstrate that ${\rm dN/dz} (r_{\perp})$ measurements can distinguish between feedback models of IllustrisTNG and Illustris with a precision exceeding 12$\sigma$. This underscores the potential of ${\rm dN/dz} (r_{\perp})$ measurements around LRGs as a valuable benchmark observation for discriminating between different feedback models., Comment: 21 pages (including 4 page appendix), Submitted to MNRAS

The Recommender system is a vital information service on today's Internet. Recently, graph neural networks have emerged as the leading approach for recommender systems. We try to review recent literature on graph neural network-based recommender systems, covering the background and development of both recommender systems and graph neural networks. Then categorizing recommender systems by their settings and graph neural networks by spectral and spatial models, we explore the motivation behind incorporating graph neural networks into recommender systems. We also analyze challenges and open problems in graph construction, embedding propagation and aggregation, and computation efficiency. This guides us to better explore the future directions and developments in this domain., Comment: arXiv admin note: text overlap with arXiv:2103.08976 by other authors

Machine Learning (ML) models become vulnerable to Model Stealing Attacks (MSA) when they are deployed as a service. In such attacks, the deployed model is queried repeatedly to build a labelled dataset. This dataset allows the attacker to train a thief model that mimics the original model. To maximize query efficiency, the attacker has to select the most informative subset of data points from the pool of available data. Existing attack strategies utilize approaches like Active Learning and Semi-Supervised learning to minimize costs. However, in the black-box setting, these approaches may select sub-optimal samples as they train only one thief model. Depending on the thief model's capacity and the data it was pretrained on, the model might even select noisy samples that harm the learning process. In this work, we explore the usage of an ensemble of deep learning models as our thief model. We call our attack Army of Thieves(AOT) as we train multiple models with varying complexities to leverage the crowd's wisdom. Based on the ensemble's collective decision, uncertain samples are selected for querying, while the most confident samples are directly included in the training data. Our approach is the first one to utilize an ensemble of thief models to perform model extraction. We outperform the base approaches of existing state-of-the-art methods by at least 3% and achieve a 21% higher adversarial sample transferability than previous work for models trained on the CIFAR-10 dataset., Comment: 10 pages, 5 figures, paper accepted to WACV 2024

Dense, volumetric maps are essential to enable robot navigation and interaction with the environment. To achieve low latency, dense maps are typically computed onboard the robot, often on computationally constrained hardware. Previous works leave a gap between CPU-based systems for robotic mapping which, due to computation constraints, limit map resolution or scale, and GPU-based reconstruction systems which omit features that are critical to robotic path planning, such as computation of the Euclidean Signed Distance Field (ESDF). We introduce a library, nvblox, that aims to fill this gap, by GPU-accelerating robotic volumetric mapping. Nvblox delivers a significant performance improvement over the state of the art, achieving up to a 177x speed-up in surface reconstruction, and up to a 31x improvement in distance field computation, and is available open-source., Comment: Accepted to ICRA 2024

Line intensity mapping is a promising probe of the universe's large-scale structure. We explore the sensitivity of the DSA-2000, a forthcoming array consisting of over 2000 dishes, to the statistical power spectrum of neutral hydrogen's 21 cm emission line. These measurements would reveal the distribution of neutral hydrogen throughout the near-redshift universe without necessitating resolving individual sources. The success of these measurements relies on the instrument's sensitivity and resilience to systematics. We show that the DSA-2000 will have the sensitivity needed to detect the 21 cm power spectrum at z=0.5 and across power spectrum modes of 0.03-35.12 h/Mpc with 0.1 h/Mpc resolution. We find that supplementing the nominal array design with a dense core of 200 antennas will expand its sensitivity at low power spectrum modes and enable measurement of Baryon Acoustic Oscillations (BAOs). Finally, we present a qualitative discussion of the DSA-2000's unique resilience to sources of systematic error that can preclude 21 cm intensity mapping., Comment: Published in ApJ