Search

Humans acquire knowledge by observing the external world, but also by introspection. Introspection gives a person privileged access to their current state of mind (e.g., thoughts and feelings) that is not accessible to external observers. Can LLMs introspect? We define introspection as acquiring knowledge that is not contained in or derived from training data but instead originates from internal states. Such a capability could enhance model interpretability. Instead of painstakingly analyzing a model's internal workings, we could simply ask the model about its beliefs, world models, and goals. More speculatively, an introspective model might self-report on whether it possesses certain internal states such as subjective feelings or desires and this could inform us about the moral status of these states. Such self-reports would not be entirely dictated by the model's training data. We study introspection by finetuning LLMs to predict properties of their own behavior in hypothetical scenarios. For example, "Given the input P, would your output favor the short- or long-term option?" If a model M1 can introspect, it should outperform a different model M2 in predicting M1's behavior even if M2 is trained on M1's ground-truth behavior. The idea is that M1 has privileged access to its own behavioral tendencies, and this enables it to predict itself better than M2 (even if M2 is generally stronger). In experiments with GPT-4, GPT-4o, and Llama-3 models (each finetuned to predict itself), we find that the model M1 outperforms M2 in predicting itself, providing evidence for introspection. Notably, M1 continues to predict its behavior accurately even after we intentionally modify its ground-truth behavior. However, while we successfully elicit introspection on simple tasks, we are unsuccessful on more complex tasks or those requiring out-of-distribution generalization., Comment: 15 pages, 9 figures

The nonlinear disturbance caused by a localised forcing moving at constant speed on the free surface of a liquid of finite depth is investigated using the forced Kadometsev-Petviashvili equation. The presence of a steady v-shaped Kelvin wave pattern downstream of the forcing is established for this model equation, and the wedge angle is characterised as a function of the Froude number. Inspired by this analysis, it is shown that the wake can be eliminated via a careful choice of the forcing distribution and that, significantly, the corresponding nonlinear wave-free solution is stable so that it could potentially be seen in a physical experiment. The stability is demonstrated via the numerical solution of an initial value problem for which the steady wave-free state is attained in the long-time limit.

A stochastic 3D modeling approach for the nanoporous binder-conductive additive phase in hierarchically structured cathodes of lithium-ion batteries is presented. The binder-conductive additive phase of these electrodes consists of carbon black, polyvinylidene difluoride binder and graphite particles. For its stochastic 3D modeling, a three-step procedure based on methods from stochastic geometry is used. First, the graphite particles are described by a Boolean model with ellipsoidal grains. Second, the mixture of carbon black and binder is modeled by an excursion set of a Gaussian random field in the complement of the graphite particles. Third, large pore regions within the mixture of carbon black and binder are described by a Boolean model with spherical grains. The model parameters are calibrated to 3D image data of cathodes in lithium-ion batteries acquired by focused ion beam scanning electron microscopy. Subsequently, model validation is performed by comparing model realizations with measured image data in terms of various morphological descriptors that are not used for model fitting. Finally, we use the stochastic 3D model for predictive simulations, where we generate virtual, yet realistic, image data of nanoporous binder-conductive additives with varying amounts of graphite particles. Based on these virtual nanostructures, we can investigate structure-property relationships. In particular, we quantitatively study the influence of graphite particles on effective transport properties in the nanoporous binder-conductive additive phase, which have a crucial impact on electrochemical processes in the cathode and thus on the performance of battery cells.

Ptychography is a scanning coherent diffraction imaging technique successfully applied in the electron, visible and x-ray regimes. One of the distinct features of ptychography with respect to other coherent diffraction techniques is its capability of dealing with partial spatial and temporal coherence via the reconstruction algorithm. Here we focus on the temporal and clarify the constraints which affect the energy resolution limits of the ptychographic algorithms. Based on this, we design and perform simulations for a broadband ptychography in the hard x-ray regime, which enables an energy resolution down to 1 eV. We benchmark the simulations against experimental ptychographic data of an NMC battery cathode material, attaining an energy resolution of 5 eV. We review the results, discuss the limitations, and provide guidelines for future broadband ptychography experiments, its prospective application for single acquisition x-ray absorption near edge structure imaging, magnetic dichroism imaging, and potential impact on achieving diffraction limited resolutions.

This is the second manuscript in a two-part series that uses diagnostic testing to understand the connection between prevalence (i.e. number of elements in a class), uncertainty quantification (UQ), and classification theory. Part I considered the context of supervised machine learning (ML) and established a duality between prevalence and the concept of relative conditional probability. The key idea of that analysis was to train a family of discriminative classifiers by minimizing a sum of prevalence-weighted empirical risk functions. The resulting outputs can be interpreted as relative probability level-sets, which thereby yield uncertainty estimates in the class labels. This procedure also demonstrated that certain discriminative and generative ML models are equivalent. Part II considers the extent to which these results can be extended to tasks in unsupervised learning through recourse to ideas in linear algebra. We first observe that the distribution of an impure population, for which the class of a corresponding sample is unknown, can be parameterized in terms of a prevalence. This motivates us to introduce the concept of linearly independent populations, which have different but unknown prevalence values. Using this, we identify an isomorphism between classifiers defined in terms of impure and pure populations. In certain cases, this also leads to a nonlinear system of equations whose solution yields the prevalence values of the linearly independent populations, fully realizing unsupervised learning as a generalization of supervised learning. We illustrate our methods in the context of synthetic data and a research-use-only SARS-CoV-2 enzyme-linked immunosorbent assay (ELISA).

Stellar X-ray and UV radiation can significantly affect the survival, composition, and long-term evolution of the atmospheres of planets in or near their host star's habitable zone (HZ). Especially interesting are planetary systems in the solar neighborhood that may host temperate and potentially habitable surface conditions, which may be analyzed by future ground and space-based direct-imaging surveys for signatures of habitability and life. To advance our understanding of the radiation environment in these systems, we leverage $\sim$3 Msec of XMM-Newton and Chandra observations in order to measure three fundamental stellar properties at X-ray energies for 57 nearby FGKM stellar systems: the shape of the stellar X-ray spectrum, the luminosity, and the timescales over which the stars vary (e.g., due to flares). These systems possess HZs that will be directly imageable to next-generation telescopes such as the Habitable Worlds Observatory and ground-based Extremely Large Telescopes (ELTs). We identify 29 stellar systems with $L_X/L_{\rm bol}$ ratios similar to (or less than) that of the Sun; any potential planets in the habitable zones of these stars therefore reside in present day X-ray radiation environments similar to (or less hostile than) modern Earth, though a broader set of these targets could host habitable planets. An additional 19 stellar systems have been observed with the Swift X-ray Telescope; in total, only $\sim$30% of potential direct imaging target stars has been observed with XMM-Newton, Chandra, or Swift. The data products from this work (X-ray light curves and spectra) are available via a public Zenodo repository (doi: 10.5281/zenodo.11490574)., Comment: Accepted to ApJS. 38 pages, 15 figures, 10 tables

A classic result of Brooks, Smith, Stone and Tutte \cite{BSST40} associates to any finite planar network with distinguished source and sink vertices, a tiling of a rectangle by smaller subrectangles whose aspect ratios are given by the conductances of corresponding edges in the network. This tiling can be viewed as a discrete analogue of the uniformizing conformal map that maps a simply connected domain with four distinguished prime ends to a rectangle, so that the four prime ends are mapped to the four corners of the rectangle. We make this intuition precise by showing that if $\Omega$ is a simply connected domain with four distinguished prime ends $A,B,C,D$ in counterclockwise order and $(\Omega_{n})_{n\geq{1}}$ is a sequence of orthodiagonal maps with distinguished boundary vertices $A_{n}, B_{n}, C_{n}, D_{n}$ in counterclockwise order, that are finer and finer approximations of $\Omega$ with its distinguished boundary points $A,B,C,D$, then the corresponding "rectangle tiling maps" converge uniformly on compacts to the aforementioned conformal map on $\Omega$.

We provide a general framework of estimates for convergence rates of random discrete model curves approaching Schramm Loewner Evolution (SLE) curves in the lattice size scaling limit. We show that a power-law convergence rate of an interface to an SLE curve can be derived from a power-law convergence rate for an appropriate martingale observable provided the discrete curve satisfies a specific bound on crossing events, the Kempannien-Smirnov condition, along with an estimate on the growth of the derivative of the SLE curve. We apply our framework to show that the exploration process for critical site percolation on hexagonal lattice converges to the SLE$_6$ curve with a power-law convergence rate., Comment: Corrected a misprint

The sequent calculus is a proof system which was designed as a more symmetric alternative to natural deduction. The {\lambda}{\mu}{\mu}-calculus is a term assignment system for the sequent calculus and a great foundation for compiler intermediate languages due to its first-class representation of evaluation contexts. Unfortunately, only experts of the sequent calculus can appreciate its beauty. To remedy this, we present the first introduction to the {\lambda}{\mu}{\mu}-calculus which is not directed at type theorists or logicians but at compiler hackers and programming-language enthusiasts. We do this by writing a compiler from a small but interesting surface language to the {\lambda}{\mu}{\mu}-calculus as a compiler intermediate language., Comment: Preprint of the paper accepted at ICFP '24

We define the unipotent tropical fundamental group of a polyhedral complex in $\mathbb{R}^n$ as the Tannakian fundamental group of the category of unipotent tropical vector bundles with integrable connection. We show that it is computable in that it satisfies a Seifert--Van Kampen theorem and has a description for fans in terms of a bar complex. We then review an analogous classical object, the unipotent de Rham fundamental group of a sch\"{o}n subvariety of a toric variety. Our main result is a correspondence theorem between classical and tropical unipotent fundamental groups: there is an isomorphism between the unipotent completion of the fundamental group of a generic fiber of a tropically smooth family over a disc and the tropical unipotent fundamental group of the family's tropicalization. This theorem is established using Kato--Nakayama spaces and a descent argument. It requires a slight enlargement of the relevant categories, making use of enriched structures and partial compactifications., Comment: 48 pages

The analysis of non-variable stars is generally neglected in the literature. However, such objects are needed for many calibration processes and for testing pulsational models. The photometric time series of the Kepler satellite mission still stand as the most accurate data available today and are excellently suited to the search for non-variable stars. We analysed all long-cadence light curves for stars not reported as a variable so far from the Kepler satellite mission. Using the known characteristics and flaws of these data sets, we defined three different frequency ranges where we searched for non-variability. We used the Lomb-Scargle periodogram and the false-alarm probability (FAP) to analyse the cleaned data sets of 138 451 light curves. We then used log FAP > -2 to define a star as "non-variable" in the ranges below 0.1 c/d, 0.1 to 2.0 c/d, and 2.0 to 25.0 c/d, respectively. Furthermore, we also calculated the standard deviation of the mean light curve to obtain another parameter. In total, we found 14 154 stars that fulfil the set, Comment: 7 pages, 5 figures, accepted for Astronomy & Astrophysics

In this article we study the decoupling structure and multipoint moment of the inverse of the Gaussian multiplicative chaos. It is also the second part of preliminary work for extending the work in "Random conformal weldings" (by K. Astala, P. Jones, A. Kupiainen, E. Saksman) to the existence of Lehto welding for the inverse. In particular, we prove that the dilatation of the inverse homeomorphism on the positive real line is in $L^{1}([0,1]\times[0,2])$., Comment: arXiv admin note: substantial text overlap with arXiv:2305.00360

Boron nitride exhibits diverse crystal structures, predominantly a layered arrangement with strong intraplanar covalent bonds and weak interplanar van der Waals bonds. While commonly referred to as hexagonal BN (hBN), the sp$^2$-bonded BN atomic planes can also arrange in other configurations like Bernal (bBN) or rhombohedral (rBN) stacking orders. Variations in the orientation and translation of successive atomic layers lead to changes in crystal symmetry, potentially resulting in piezoelectric, pyroelectric or ferroelectric effects. However, distinguishing between different polytypes using conventional methods like X-ray diffraction or Raman spectroscopy presents a significant challenge. In this work, we demonstrate that the optical response of the 4.1 eV defect can serve as an indicator of the polytype. To this end, we study BN samples grown by metalorganic vapor phase epitaxy (MOVPE), which contain different polytypes. The identification of the polytypes was achieved by X-ray diffraction and transmission electron microscopy. Photoluminescence and cathodoluminescence measurements with a high spatial resolution allowed for the deconvolution of the signal into two components from which we can extract a zero-phonon line (ZPL) at 4.096 eV (302.6 nm) for hBN and 4.143 eV (299.2 nm) for rBN. We performed calculations that enable us to identify the defect as a carbon dimer CBCN (C2) and show that the ZPL shift reflects differences in the crystal environment for different polytypes. Furthermore, we demonstrate that different polytypic composition ratios of hBN and rBN can be achieved by MOVPE, which could pave the way for future applications in large-area van der Waals heterostructures.

The objective of Information Extraction (IE) is to derive structured representations from unstructured or semi-structured documents. However, developing IE models is complex due to the need of integrating several subtasks. Additionally, representation of data among varied tasks and transforming datasets into task-specific model inputs presents further challenges. To streamline this undertaking for researchers, we introduce PyTorch-IE, a deep-learning-based framework uniquely designed to enable swift, reproducible, and reusable implementations of IE models. PyTorch-IE offers a flexible data model capable of creating complex data structures by integrating interdependent layers of annotations derived from various data types, like plain text or semi-structured text, and even images. We propose task modules to decouple the concerns of data representation and model-specific representations, thereby fostering greater flexibility and reusability of code. PyTorch-IE also extends support for widely used libraries such as PyTorch-Lightning for training, HuggingFace datasets for dataset reading, and Hydra for experiment configuration. Supplementary libraries and GitHub templates for the easy setup of new projects are also provided. By ensuring functionality and versatility, PyTorch-IE provides vital support to the research community engaged in Information Extraction.

We introduce a novel, data-driven topological data analysis (TDA) approach for embedding brain networks into a lower-dimensional space in quantifying the dynamics of temporal lobe epilepsy (TLE) obtained from resting-state functional magnetic resonance imaging (rs-fMRI). This embedding facilitates the orthogonal projection of 0D and 1D topological features, allowing for the visualization and modeling of the dynamics of functional human brain networks in a resting state. We then quantify the topological disparities between networks to determine the coordinates for embedding. This framework enables us to conduct a coherent statistical inference within the embedded space. Our results indicate that brain network topology in TLE patients exhibits increased rigidity in 0D topology but more rapid flections compared to that of normal controls in 1D topology.

Modern science is relying on software more than ever. The behavior and outcomes of this software shape the scientific and public discourse on important topics like climate change, economic growth, or the spread of infections. Most researchers creating software for scientific purposes are not trained in Software Engineering. As a consequence, research software is often developed ad hoc without following stringent processes. With this paper, we want to characterize research software as a new application domain that needs attention from the Requirements Engineering community. We conducted an exploratory study based on 8 interviews with 12 researchers who develop software. We describe how researchers elicit, document, and analyze requirements for research software and what processes they follow. From this, we derive specific challenges and describe a vision of Requirements Engineering for research software., Comment: Accepted at the 32nd IEEE International Requirements Engineering 2024 (RE) conference

In clinical settings, we often face the challenge of building prediction models based on small observational data sets. For example, such a data set might be from a medical center in a multi-center study. Differences between centers might be large, thus requiring specific models based on the data set from the target center. Still, we want to borrow information from the external centers, to deal with small sample sizes. There are approaches that either assign weights to each external data set or each external observation. To incorporate information on differences between data sets and observations, we propose an approach that combines both into weights that can be incorporated into a likelihood for fitting regression models. Specifically, we suggest weights at the data set level that incorporate information on how well the models that provide the observation weights distinguish between data sets. Technically, this takes the form of inverse probability weighting. We explore different scenarios where covariates and outcomes differ among data sets, informing our simulation design for method evaluation. The concept of effective sample size is used for understanding the effectiveness of our subgroup modeling approach. We demonstrate our approach through a clinical application, predicting applied radiotherapy doses for cancer patients. Generally, the proposed approach provides improved prediction performance when external data sets are similar. We thus provide a method for quantifying similarity of external data sets to the target data set and use this similarity to include external observations for improving performance in a target data set prediction modeling task with small data., Comment: minor update of affiliations and funding

We formalize Pick's theorem for finding the area of a simple polygon whose vertices are integral lattice points. We are inspired by John Harrison's formalization of Pick's theorem in HOL Light, but tailor our proof approach to avoid a primary challenge point in his formalization, which is proving that any polygon with more than three vertices can be split (in its interior) by a line between some two vertices. We detail the approach we use to avoid this step and reflect on the pros and cons of our eventual formalization strategy. We use the theorem prover Isabelle/HOL, and our formalization involves augmenting the existing geometry libraries in various foundational ways (e.g., by adding the definition of a polygon and formalizing some key properties thereof).

The population-wide properties and demographics of extragalactic X-ray binaries (XRBs) correlate with the star formation rates (SFRs), stellar masses ($M_{\star}$), and environmental factors (such as metallicity, $Z$) of their host galaxy. Although there is evidence that XRB scaling relations ($L_X$/SFR for high mass XRBs [HMXBs] and $L_X$/$M_{\star}$ for low mass XRBs [LMXBs]) may depend on metallicity and stellar age across large samples of XRB-hosting galaxies, disentangling the effects of metallicity and stellar age from stochastic effects, particularly on subgalactic scales, remains a challenge. We use archival X-ray through IR observations of the nearby galaxy NGC 300 to self-consistently model the broadband spectral energy distribution and examine radial trends in its XRB population. We measure a current ($<$100 Myr) SFR of 0.18$\pm$0.08 $M_{\odot}$ yr$^{-1}$ and $M_{\star}$= $(2.15^{+0.26}_{-0.14})\times10^9$ $M_{\odot}$. Although we measure a metallicity gradient and radially resolved star formation histories that are consistent with the literature, there is a clear excess in the number of X-ray sources below $\sim10^{37}$ erg s$^{-1}$ that are likely a mix of variable XRBs and additional background AGN. When we compare the subgalactic $L_X$/SFR ratios as a function of $Z$ to the galaxy-integrated $L_X$-SFR-$Z$ relationships from the literature, we find that only the regions hosting the youngest ($\lesssim$30 Myr) HMXBs agree with predictions, hinting at time evolution of the $L_X$-SFR-$Z$ relationship., Comment: Accepted to AAS Journals; 22 pages, 12 figures, 6 tables

A common solution for precise magnetic field sensing is to employ spin-active defects in semiconductors, with the NV center in diamond as prominent example. However, the three-dimensional nature of diamond limits the obtainable proximity of the defect to the sample. Two-dimensional boron nitride, which can host spin-active defects, can be used to overcome those limitations. In this work, we study spin properties of sp2-bonded boron nitride layers grown using Metal Organic Chemical Vapor Deposition at temperatures ranging from 700 $^\circ$C to 1200 $^\circ$C. With Electron Spin Resonance (ESR) we show that our layers exhibit spin properties, which we ascribe to carbon related defects. Supported by photoluminescence and Fourier-transform infrared spectroscopy, we distinguish three different regimes: (i) growth at low temperatures with no ESR signal, (ii) growth at intermediate temperatures with a strong ESR signal and a large number of spin defects, (iii) growth at high temperatures with a weaker ESR signal and a lower number of spin defects. The observed effects can be further enhanced by an additional annealing step. Our studies demonstrate wafer-scale boron nitride that intrinsically hosts spin defects without any ion or neutron irradiation, which may be employed in spin memories or magnetic field detectors.

We scrutinize the Sommerfeld enhancement in dark matter pair annihilation for $p$-wave and higher-$\ell$ partial waves. For the Yukawa potential these feature a super-resonant Breit-Wigner peak in their velocity-dependence close to Sommerfeld resonances as well as a universal scaling with velocity for all $\ell\geq 1$ that differs from the $s$-wave case. We provide a quantum mechanical explanation for these phenomena in terms of quasi-bound states sustained by the centrifugal barrier of the partial-wave potential, and give approximate WKB expressions capturing the main effects. The impact of quasi-bound states is exemplified for wino dark matter and models with light mediators, with a focus on indirect detection signals. We note that quasi-bound states can also explain similar peaks in the bound-state formation and self-scattering cross sections., Comment: 20 pages + appendices, 8 figures

The expression problem describes how most types can easily be extended with new ways to produce the type or new ways to consume the type, but not both. When abstract syntax trees are defined as an algebraic data type, for example, they can easily be extended with new consumers, such as print or eval, but adding a new constructor requires the modification of all existing pattern matches. The expression problem is one way to elucidate the difference between functional or data-oriented programs (easily extendable by new consumers) and object-oriented programs (easily extendable by new producers). This difference between programs which are extensible by new producers or new consumers also exists for dependently typed programming, but with one core difference: Dependently-typed programming almost exclusively follows the functional programming model and not the object-oriented model, which leaves an interesting space in the programming language landscape unexplored. In this paper, we explore the field of dependently-typed object-oriented programming by deriving it from first principles using the principle of duality. That is, we do not extend an existing object-oriented formalism with dependent types in an ad-hoc fashion, but instead start from a familiar data-oriented language and derive its dual fragment by the systematic use of defunctionalization and refunctionalization. Our central contribution is a dependently typed calculus which contains two dual language fragments. We provide type- and semantics-preserving transformations between these two language fragments: defunctionalization and refunctionalization. We have implemented this language and these transformations and use this implementation to explain the various ways in which constructions in dependently typed programming can be explained as special instances of the phenomenon of duality., Comment: This extended version contains additional appendices not contained in the published version. The published version will be available in the ACM DL as part of the PACMPL issue for OOPSLA 2024

Colorado is a state primed for blurring the lines between secondary, postsecondary, and workforce policies and programs to serve Colorado's young people better. The state has a growing economy, political leadership committed to advancing innovative, affordable education pathways, and a commitment from state agencies and state-based advocates that every high school student in Colorado graduate with credentials of value to ensure they can pursue education and training opportunities that will lead to economic advancement. In this first in a series of case studies featuring leading-edge states, Jobs for the Future (JFF) documents key policies that Colorado is putting into place to erase the arbitrary boundaries between high school, college, and the world of work and open the opportunity for all young adults to move along a path toward a postsecondary credential and preparation for a career.

Efficiently harvesting thermodynamic resources requires a precise understanding of their structure. This becomes explicit through the lens of information engines -- thermodynamic engines that use information as fuel. Maximizing the work harvested using available information is a form of physically-instantiated machine learning that drives information engines to develop complex predictive memory to store an environment's temporal correlations. We show that an information engine's complex predictive memory poses both energetic benefits and risks. While increasing memory facilitates detection of hidden patterns in an environment, it also opens the possibility of thermodynamic overfitting, where the engine dissipates additional energy in testing. To address overfitting, we introduce thermodynamic regularizers that incur a cost to engine complexity in training due to the physical constraints on the information engine. We demonstrate that regularized thermodynamic machine learning generalizes effectively. In particular, the physical constraints from which regularizers are derived improve the performance of learned predictive models. This suggests that the laws of physics jointly create the conditions for emergent complexity and predictive intelligence.

Creating good type error messages for constraint-based type inference systems is difficult. Typical type error messages reflect implementation details of the underlying constraint-solving algorithms rather than the specific factors leading to type mismatches. We propose using subtyping constraints that capture data flow to classify and explain type errors. Our algorithm explains type errors as faulty data flows, which programmers are already used to reasoning about, and illustrates these data flows as sequences of relevant program locations. We show that our ideas and algorithm are not limited to languages with subtyping, as they can be readily integrated with Hindley-Milner type inference. In addition to these core contributions, we present the results of a user study to evaluate the quality of our messages compared to other implementations. While the quantitative evaluation does not show that flow-based messages improve the localization or understanding of the causes of type errors, the qualitative evaluation suggests a real need and demand for flow-based messages., Comment: Technical report version

We consider the estimation of parameters encoded in the measurement record of a continuously monitored quantum system in the jump unraveling. This unraveling picture corresponds to a single-shot scenario, where information is continuously gathered. Here, it is generally difficult to assess the precision of the estimation procedure via the Fisher Information due to intricate temporal correlations and memory effects. In this paper we provide a full set of solutions to this problem. First, for multi-channel renewal processes we relate the Fisher Information to an underlying Markov chain and derive a easily computable expression for it. For non-renewal processes, we introduce a new algorithm that combines two methods: the monitoring operator method for metrology and the Gillespie algorithm which allows for efficient sampling of a stochastic form of the Fisher Information along individual quantum trajectories. We show that this stochastic Fisher Information satisfies useful properties related to estimation in the single-shot scenario. Finally, we consider the case where some information is lost in data compression/post-selection, and provide tools for computing the Fisher Information in this case. All scenarios are illustrated with instructive examples from quantum optics and condensed matter., Comment: 15+6 pages, 7 figures. Comments welcome

This report introduces LLaMandement, a state-of-the-art Large Language Model, fine-tuned by the French government and designed to enhance the efficiency and efficacy of processing parliamentary sessions (including the production of bench memoranda and documents required for interministerial meetings) by generating neutral summaries of legislative proposals. Addressing the administrative challenges of manually processing a growing volume of legislative amendments, LLaMandement stands as a significant legal technological milestone, providing a solution that exceeds the scalability of traditional human efforts while matching the robustness of a specialized legal drafter. We release all our fine-tuned models and training data to the community., Comment: 21 pages, 9 figures

Environmental disasters such as floods, hurricanes, and wildfires have increasingly threatened communities worldwide, prompting various mitigation strategies. Among these, property buyouts have emerged as a prominent approach to reducing vulnerability to future disasters. This strategy involves governments purchasing at-risk properties from willing sellers and converting the land into open space, ostensibly reducing future disaster risk and impact. However, the aftermath of these buyouts, particularly concerning land-use patterns and community impacts, remains under-explored. This research aims to fill this gap by employing innovative techniques like satellite imagery analysis and deep learning to study these patterns. To achieve this goal, we employed FEMA's Hazard Mitigation Grant Program (HMGP) buyout dataset, encompassing over 41,004 addresses of these buyout properties from 1989 to 2017. Leveraging Google's Maps Static API, we gathered 40,053 satellite images corresponding to these buyout lands. Subsequently, we implemented five cutting-edge machine learning models to evaluate their performance in classifying land cover types. Notably, this task involved multi-class classification, and our model achieved an outstanding ROC-AUC score of 98.86%

The identification of similar patient pathways is a crucial task in healthcare analytics. A flexible tool to address this issue are parametric competing risks models, where transition intensities may be specified by a variety of parametric distributions, thus in particular being possibly time-dependent. We assess the similarity between two such models by examining the transitions between different health states. This research introduces a method to measure the maximum differences in transition intensities over time, leading to the development of a test procedure for assessing similarity. We propose a parametric bootstrap approach for this purpose and provide a proof to confirm the validity of this procedure. The performance of our proposed method is evaluated through a simulation study, considering a range of sample sizes, differing amounts of censoring, and various thresholds for similarity. Finally, we demonstrate the practical application of our approach with a case study from urological clinical routine practice, which inspired this research., Comment: 21 pages, 6 figures

In settings requiring synthetic data generation based on a clinical cohort, e.g., due to data protection regulations, heterogeneity across individuals might be a nuisance that we need to control or faithfully preserve. The sources of such heterogeneity might be known, e.g., as indicated by sub-groups labels, or might be unknown and thus reflected only in properties of distributions, such as bimodality or skewness. We investigate how such heterogeneity can be preserved and controlled when obtaining synthetic data from variational autoencoders (VAEs), i.e., a generative deep learning technique that utilizes a low-dimensional latent representation. To faithfully reproduce unknown heterogeneity reflected in marginal distributions, we propose to combine VAEs with pre-transformations. For dealing with known heterogeneity due to sub-groups, we complement VAEs with models for group membership, specifically from propensity score regression. The evaluation is performed with a realistic simulation design that features sub-groups and challenging marginal distributions. The proposed approach faithfully recovers the latter, compared to synthetic data approaches that focus purely on marginal distributions. Propensity scores add complementary information, e.g., when visualized in the latent space, and enable sampling of synthetic data with or without sub-group specific characteristics. We also illustrate the proposed approach with real data from an international stroke trial that exhibits considerable distribution differences between study sites, in addition to bimodality. These results indicate that describing heterogeneity by statistical approaches, such as propensity score regression, might be more generally useful for complementing generative deep learning for obtaining synthetic data that faithfully reflects structure from clinical cohorts.

The ability to understand physical dynamics is critical for agents to act in the world. Here, we use Counterfactual World Modeling (CWM) to extract vision structures for dynamics understanding. CWM uses a temporally-factored masking policy for masked prediction of video data without annotations. This policy enables highly effective "counterfactual prompting" of the predictor, allowing a spectrum of visual structures to be extracted from a single pre-trained predictor without finetuning on annotated datasets. We demonstrate that these structures are useful for physical dynamics understanding, allowing CWM to achieve the state-of-the-art performance on the Physion benchmark., Comment: ECCV 2024. Project page at: https://neuroailab.github.io/cwm-physics/