Search

The Casimir-Lifshitz force arises from thermal and quantum mechanical fluctuations between classical bodies and becomes significant below the micron scale. We explore temperature-distance relations based on the concepts of Wick and Bohr arising from energy-time uncertainty relations. We show that temperature-distance relations similar to those arising from the uncertainty principle are found in various Casimir interactions, with an exact relation occurring in the low-temperature regime when the zero point energy contribution cancels the thermal radiation pressure contribution between two plates., Comment: 10 pages, 1 figure

We present Lil-Bevo, our submission to the BabyLM Challenge. We pretrained our masked language models with three ingredients: an initial pretraining with music data, training on shorter sequences before training on longer ones, and masking specific tokens to target some of the BLiMP subtasks. Overall, our baseline models performed above chance, but far below the performance levels of larger LLMs trained on more data. We found that training on short sequences performed better than training on longer sequences.Pretraining on music may help performance marginally, but, if so, the effect seems small. Our targeted Masked Language Modeling augmentation did not seem to improve model performance in general, but did seem to help on some of the specific BLiMP tasks that we were targeting (e.g., Negative Polarity Items). Training performant LLMs on small amounts of data is a difficult but potentially informative task. While some of our techniques showed some promise, more work is needed to explore whether they can improve performance more than the modest gains here. Our code is available at https://github.com/venkatasg/Lil-Bevo and out models at https://huggingface.co/collections/venkatasg/babylm-653591cdb66f4bf68922873a, Comment: Proceedings of the BabyLM Challenge

While large language models (LLMs) equipped with techniques like chain-of-thought prompting have demonstrated impressive capabilities, they still fall short in their ability to reason robustly in complex settings. However, evaluating LLM reasoning is challenging because system capabilities continue to grow while benchmark datasets for tasks like logical deduction have remained static. We introduce MuSR, a dataset for evaluating language models on multistep soft reasoning tasks specified in a natural language narrative. This dataset has two crucial features. First, it is created through a novel neurosymbolic synthetic-to-natural generation algorithm, enabling the construction of complex reasoning instances that challenge GPT-4 (e.g., murder mysteries roughly 1000 words in length) and which can be scaled further as more capable LLMs are released. Second, our dataset instances are free text narratives corresponding to real-world domains of reasoning; this makes it simultaneously much more challenging than other synthetically-crafted benchmarks while remaining realistic and tractable for human annotators to solve with high accuracy. We evaluate a range of LLMs and prompting techniques on this dataset and characterize the gaps that remain for techniques like chain-of-thought to perform robust reasoning.

Current natural language systems designed for multi-step claim validation typically operate in two phases: retrieve a set of relevant premise statements using heuristics (planning), then generate novel conclusions from those statements using a large language model (deduction). The planning step often requires expensive Transformer operations and does not scale to arbitrary numbers of premise statements. In this paper, we investigate whether an efficient planning heuristic is possible via embedding spaces compatible with deductive reasoning. Specifically, we evaluate whether embedding spaces exhibit a property we call deductive additivity: the sum of premise statement embeddings should be close to embeddings of conclusions based on those premises. We explore multiple sources of off-the-shelf dense embeddings in addition to fine-tuned embeddings from GPT3 and sparse embeddings from BM25. We study embedding models both intrinsically, evaluating whether the property of deductive additivity holds, and extrinsically, using them to assist planning in natural language proof generation. Lastly, we create a dataset, Single-Step Reasoning Contrast (SSRC), to further probe performance on various reasoning types. Our findings suggest that while standard embedding methods frequently embed conclusions near the sums of their premises, they fall short of being effective heuristics and lack the ability to model certain categories of reasoning.

The accelerated change in our planet due to human activities has led to grand societal challenges including health crises, intensified extreme weather events, food security, environmental injustice, etc. Digital twin systems combined with emerging technologies such as artificial intelligence and edge computing provide opportunities to support planning and decision-making to address these challenges. Digital twins for Earth systems (DT4ESs) are defined as the digital representation of the complex integrated Earth system including both natural processes and human activities. They have the potential to enable a diverse range of users to explore what-if scenarios across spatial and temporal scales to improve our understanding, prediction, mitigation, and adaptation to grand societal challenges. The 4th NOAA AI Workshop convened around 100 members who are developing or interested in participating in the development of DT4ES to discuss a shared community vision and path forward on fostering a future ecosystem of interoperable DT4ES. This paper summarizes the workshop discussions around DT4ES. We first defined the foundational features of a viable digital twins for Earth system that can be used to guide the development of various use cases of DT4ES. Finally, we made practical recommendations for the community on different aspects of collaboration in order to enable a future ecosystem of interoperable DT4ES, including equity-centered use case development, community-driven investigation of interoperability for DT4ES, trust-oriented co-development, and developing a community of practice., Comment: This whitepaper is an outcome of the 4th NOAA AI Workshop

Passive Gamma Emission Tomography (PGET) has been developed by the International Atomic Energy Agency as a way to directly image the spatial distribution of individual fuel pins in a spent nuclear fuel assembly and so determine potential diversion. Constructing the analysis and interpretation of PGET measurements rely on the availability of comprehensive datasets. Experimental data are expensive, limited, and so are augmented by Monte Carlo simulations. The main issue concerning Monte Carlo simulations is the high computational cost to simulate the 360 angular views of the tomography. Similar challenges pervade numerical science. To address this challenge, we have developed a physics-aware reduced order modeling approach. It provides a framework to combine a small subset of the 360 angular views with a computationally inexpensive proxy solution, that brings the essence of the physics, to obtain a real-time high-fidelity solution at all angular views, but at a fraction of the computational cost.

A growing body of work studies how to answer a question or verify a claim by generating a natural language "proof": a chain of deductive inferences yielding the answer based on a set of premises. However, these methods can only make sound deductions when they follow from evidence that is given. We propose a new system that can handle the underspecified setting where not all premises are stated at the outset; that is, additional assumptions need to be materialized to prove a claim. By using a natural language generation model to abductively infer a premise given another premise and a conclusion, we can impute missing pieces of evidence needed for the conclusion to be true. Our system searches over two fringes in a bidirectional fashion, interleaving deductive (forward-chaining) and abductive (backward-chaining) generation steps. We sample multiple possible outputs for each step to achieve coverage of the search space, at the same time ensuring correctness by filtering low-quality generations with a round-trip validation procedure. Results on a modified version of the EntailmentBank dataset and a new dataset called Everyday Norms: Why Not? show that abductive generation with validation can recover premises across in- and out-of-domain settings., Comment: Conference of EMNLP 2022

In settings from fact-checking to question answering, we frequently want to know whether a collection of evidence (premises) entails a hypothesis. Existing methods primarily focus on the end-to-end discriminative version of this task, but less work has treated the generative version in which a model searches over the space of statements entailed by the premises to constructively derive the hypothesis. We propose a system for doing this kind of deductive reasoning in natural language by decomposing the task into separate steps coordinated by a search procedure, producing a tree of intermediate conclusions that faithfully reflects the system's reasoning process. Our experiments on the EntailmentBank dataset (Dalvi et al., 2021) demonstrate that the proposed system can successfully prove true statements while rejecting false ones. Moreover, it produces natural language explanations with a 17% absolute higher step validity than those produced by an end-to-end T5 model., Comment: Findings of EMNLP 2022

Given the growing use of Artificial Intelligence (AI) and machine learning (ML) methods across all aspects of environmental sciences, it is imperative that we initiate a discussion about the ethical and responsible use of AI. In fact, much can be learned from other domains where AI was introduced, often with the best of intentions, yet often led to unintended societal consequences, such as hard coding racial bias in the criminal justice system or increasing economic inequality through the financial system. A common misconception is that the environmental sciences are immune to such unintended consequences when AI is being used, as most data come from observations, and AI algorithms are based on mathematical formulas, which are often seen as objective. In this article, we argue the opposite can be the case. Using specific examples, we demonstrate many ways in which the use of AI can introduce similar consequences in the environmental sciences. This article will stimulate discussion and research efforts in this direction. As a community, we should avoid repeating any foreseeable mistakes made in other domains through the introduction of AI. In fact, with proper precautions, AI can be a great tool to help {\it reduce} climate and environmental injustice. We primarily focus on weather and climate examples but the conclusions apply broadly across the environmental sciences.

Evaluating explanation techniques using human subjects is costly, time-consuming and can lead to subjectivity in the assessments. To evaluate the accuracy of local explanations, we require access to the true feature importance scores for a given instance. However, the prediction function of a model usually does not decompose into linear additive terms that indicate how much a feature contributes to the output. In this work, we suggest to instead focus on the log odds ratio (LOR) of the prediction function, which naturally decomposes into additive terms for logistic regression and naive Bayes. We demonstrate how we can benchmark different explanation techniques in terms of their similarity to the LOR scores based on our proposed approach. In the experiments, we compare prominent local explanation techniques and find that the performance of the techniques can depend on the underlying model, the dataset, which data point is explained, the normalization of the data and the similarity metric., Comment: Submitted to ACM FaCCT 2022 Jan 21 2022, 13 pages, 4 Figures

An interpretable system for open-domain reasoning needs to express its reasoning process in a transparent form. Natural language is an attractive representation for this purpose -- it is both highly expressive and easy for humans to understand. However, manipulating natural language statements in logically consistent ways is hard: models must cope with variation in how meaning is expressed while remaining precise. In this paper, we describe ParaPattern, a method for building models to generate deductive inferences from diverse natural language inputs without direct human supervision. We train BART-based models (Lewis et al., 2020) to generate the result of applying a particular logical operation to one or more premise statements. Crucially, we develop a largely automated pipeline for constructing suitable training examples from Wikipedia. We evaluate our models using out-of-domain sentence compositions from the QASC (Khot et al., 2020) and EntailmentBank (Dalvi et al., 2021) datasets as well as targeted perturbation sets. Our results show that our models are substantially more accurate and flexible than baseline systems. ParaPattern achieves 85% validity on examples of the 'substitution' operation from EntailmentBank without the use of any in-domain training data, matching the performance of a model fine-tuned for EntailmentBank. The full source code for our method is publicly available., Comment: Accepted to EMNLP 2021 (long paper). 9 pages (13 with references and appendix), 8 figures

The Hierarchical Vote Collective of Transformation-based Ensembles (HIVE-COTE) is a heterogeneous meta ensemble for time series classification. HIVE-COTE forms its ensemble from classifiers of multiple domains, including phase-independent shapelets, bag-of-words based dictionaries and phase-dependent intervals. Since it was first proposed in 2016, the algorithm has remained state of the art for accuracy on the UCR time series classification archive. Over time it has been incrementally updated, culminating in its current state, HIVE-COTE 1.0. During this time a number of algorithms have been proposed which match the accuracy of HIVE-COTE. We propose comprehensive changes to the HIVE-COTE algorithm which significantly improve its accuracy and usability, presenting this upgrade as HIVE-COTE 2.0. We introduce two novel classifiers, the Temporal Dictionary Ensemble (TDE) and Diverse Representation Canonical Interval Forest (DrCIF), which replace existing ensemble members. Additionally, we introduce the Arsenal, an ensemble of ROCKET classifiers as a new HIVE-COTE 2.0 constituent. We demonstrate that HIVE-COTE 2.0 is significantly more accurate than the current state of the art on 112 univariate UCR archive datasets and 26 multivariate UEA archive datasets.

The success of pretrained transformer language models (LMs) in natural language processing has led to a wide range of pretraining setups. In particular, these models employ a variety of subword tokenization methods, most notably byte-pair encoding (BPE) (Sennrich et al., 2016; Gage, 1994), the WordPiece method (Schuster and Nakajima, 2012), and unigram language modeling (Kudo, 2018), to segment text. However, to the best of our knowledge, the literature does not contain a direct evaluation of the impact of tokenization on language model pretraining. We analyze differences between BPE and unigram LM tokenization, finding that the latter method recovers subword units that align more closely with morphology and avoids problems stemming from BPE's greedy construction procedure. We then compare the fine-tuned task performance of identical transformer masked language models pretrained with these tokenizations. Across downstream tasks and two languages (English and Japanese), we find that the unigram LM tokenization method matches or outperforms BPE. We hope that developers of future pretrained LMs will consider adopting the unigram LM method over the more prevalent BPE., Comment: 5 pages, 3 figures. To be published in Findings of EMNLP 2020

In 2002, the UCR time series classification archive was first released with sixteen datasets. It gradually expanded, until 2015 when it increased in size from 45 datasets to 85 datasets. In October 2018 more datasets were added, bringing the total to 128. The new archive contains a wide range of problems, including variable length series, but it still only contains univariate time series classification problems. One of the motivations for introducing the archive was to encourage researchers to perform a more rigorous evaluation of newly proposed time series classification (TSC) algorithms. It has worked: most recent research into TSC uses all 85 datasets to evaluate algorithmic advances. Research into multivariate time series classification, where more than one series are associated with each class label, is in a position where univariate TSC research was a decade ago. Algorithms are evaluated using very few datasets and claims of improvement are not based on statistical comparisons. We aim to address this problem by forming the first iteration of the MTSC archive, to be hosted at the website www.timeseriesclassification.com. Like the univariate archive, this formulation was a collaborative effort between researchers at the University of East Anglia (UEA) and the University of California, Riverside (UCR). The 2018 vintage consists of 30 datasets with a wide range of cases, dimensions and series lengths. For this first iteration of the archive we format all data to be of equal length, include no series with missing data and provide train/test splits.

In short, our experiments suggest that yes, on average, rotation forest is better than the most common alternatives when all the attributes are real-valued. Rotation forest is a tree based ensemble that performs transforms on subsets of attributes prior to constructing each tree. We present an empirical comparison of classifiers for problems with only real-valued features. We evaluate classifiers from three families of algorithms: support vector machines; tree-based ensembles; and neural networks tuned with a large grid search. We compare classifiers on unseen data based on the quality of the decision rule (using classification error) the ability to rank cases (area under the receiver operating characteristic) and the probability estimates (using negative log likelihood). We conclude that, in answer to the question posed in the title, yes, rotation forest is significantly more accurate on average than competing techniques when compared on three distinct sets of datasets. Further, we assess the impact of the design features of rotation forest through an ablative study that transforms random forest into rotation forest. We identify the major limitation of rotation forest as its scalability, particularly in number of attributes. To overcome this problem we develop a model to predict the train time of the algorithm and hence propose a contract version of rotation forest where a run time cap is imposed {\em a priori}. We demonstrate that on large problems rotation forest can be made an order of magnitude faster without significant loss of accuracy. We also show that there is no real benefit (on average) from tuning rotation forest. We maintain that without any domain knowledge to indicate an algorithm preference, rotation forest should be the default algorithm of choice for problems with continuous attributes.

Shapelets are phase independent subsequences designed for time series classification. We propose three adaptations to the Shapelet Transform (ST) to capture multivariate features in multivariate time series classification. We create a unified set of data to benchmark our work on, and compare with three other algorithms. We demonstrate that multivariate shapelets are not significantly worse than other state-of-the-art algorithms.

Strong electron cooling on the neutral gas in cometary comae has been predicted for a long time, but actual measurements of low electron temperature are scarce. We present in situ measurements of plasma density, electron temperature and spacecraft potential by the Rosetta Langmuir probe instrument, LAP. Data acquired within a few hundred km from the nucleus are dominated by a warm component with electron temperature typically 5--10 eV at all heliocentric distances covered (1.25 to 3.83 AU). A cold component, with temperature no higher than about 0.1 eV, appears in the data as short (few to few tens of seconds) pulses of high probe current, indicating local enhancement of plasma density as well as a decrease in electron temperature. These pulses first appeared around 3 AU and were seen for longer periods close to perihelion. The general pattern of pulse appearance follows that of neutral gas and plasma density. We have not identified any periods with only cold electrons present. The electron flux to Rosetta was always dominated by higher energies, driving the spacecraft potential to order -10 V. The warm (5--10 eV) electron population is interpreted as electrons retaining the energy they obtained when released in the ionisation process. The sometimes observed cold populations with electron temperatures below 0.1 eV verify collisional cooling in the coma. The cold electrons were only observed together with the warm population. The general appearance of the cold population appears to be consistent with a Haser-like model, implicitly supporting also the coupling of ions to the neutral gas. The expanding cold plasma is unstable, forming filaments that we observe as pulses.

There are now a broad range of time series classification (TSC) algorithms designed to exploit different representations of the data. These have been evaluated on a range of problems hosted at the UCR-UEA TSC Archive (www.timeseriesclassification.com), and there have been extensive comparative studies. However, our understanding of why one algorithm outperforms another is still anecdotal at best. This series of experiments is meant to help provide insights into what sort of discriminatory features in the data lead one set of algorithms that exploit a particular representation to be better than other algorithms. We categorise five different feature spaces exploited by TSC algorithms then design data simulators to generate randomised data from each representation. We describe what results we expected from each class of algorithm and data representation, then observe whether these prior beliefs are supported by the experimental evidence. We provide an open source implementation of all the simulators to allow for the controlled testing of hypotheses relating to classifier performance on different data representations. We identify many surprising results that confounded our expectations, and use these results to highlight how an over simplified view of classifier structure can often lead to erroneous prior beliefs. We believe ensembling can often overcome prior bias, and our results support the belief by showing that the ensemble approach adopted by the Hierarchical Collective of Transform based Ensembles (HIVE-COTE) is significantly better than the alternatives when the data representation is unknown, and is significantly better than, or not significantly significantly better than, or not significantly worse than, the best other approach on three out of five of the individual simulators.

We perform the first reflection study of the soft X-ray transient and Type 1 burst source XTE J1709-267 using NuSTAR observations during its 2016 June outburst. There was an increase in flux near the end of the observations, which corresponds to an increase from $\sim$0.04 L$_{\mathrm{Edd}}$ to $\sim$0.06 L$_{\mathrm{Edd}}$ assuming a distance of 8.5 kpc. We have separately examined spectra from the low and high flux intervals, which were soft and show evidence of a broad Fe K line. Fits to these intervals with relativistic disk reflection models have revealed an inner disk radius of $13.8_{-1.8}^{+3.0}\ R_{g}$ (where $R_{g} = GM/c^{2}$) for the low flux spectrum and $23.4_{-5.4}^{+15.6}\ R_{g}$ for the high flux spectrum at the 90\% confidence level. The disk is likely truncated by a boundary layer surrounding the neutron star or the magnetosphere. Based on the measured luminosity and using the accretion efficiency for a disk around a neutron star, we estimate that the theoretically expected size for the boundary layer would be $\sim0.9-1.1 \ R_{g}$ from the neutron star's surface, which can be increased by spin or viscosity effects. Another plausible scenario is that the disk could be truncated by the magnetosphere. We place a conservative upper limit on the strength of the magnetic field at the poles, assuming $a_{*}=0$ and $M_{NS}=1.4\ M_{\odot}$, of $B\leq0.75-3.70\times10^{9}$ G, though X-ray pulsations have not been detected from this source., Comment: Accepted for publication in ApJ, 5 pages, 4 figures, 1 table. arXiv admin note: text overlap with arXiv:1701.01774

We present $\emph{NuSTAR}$ observations of neutron star (NS) low-mass X-ray binaries: 4U 1636-53, GX 17+2, and 4U 1705-44. We observed 4U 1636-53 in the hard state, with an Eddington fraction, $F_{\mathrm{Edd}}$, of 0.01; GX 17+2 and 4U 1705-44 were in the soft state with fractions of 0.57 and 0.10, respectively. Each spectrum shows evidence for a relativistically broadened Fe K$_{\alpha}$ line. Through accretion disk reflection modeling, we constrain the radius of the inner disk in 4U 1636-53 to be $R_{in}=1.03\pm0.03$ ISCO (innermost stable circular orbit) assuming a dimensionless spin parameter $a_{*}=cJ/GM^{2}=0.0$, and $R_{in}=1.08\pm0.06$ ISCO for $a_{*}=0.3$ (errors quoted at 1 $\sigma$). This value proves to be model independent. For $a_{*}=0.3$ and $M=1.4\ M_{\odot}$, for example, $1.08\pm0.06$ ISCO translates to a physical radius of $R=10.8\pm0.6$ km, and the neutron star would have to be smaller than this radius (other outcomes are possible for allowed spin parameters and masses). For GX 17+2, $R_{in}=1.00-1.04$ ISCO for $a_{*}=0.0$ and $R_{in}=1.03-1.30$ ISCO for $a_{*}=0.3$. For $a_{*}=0.3$ and $M=1.4\ M_{\odot}$, $R_{in}=1.03-1.30$ ISCO translates to $R=10.3-13.0$ km. The inner accretion disk in 4U 1705-44 may be truncated just above the stellar surface, perhaps by a boundary layer or magnetosphere; reflection models give a radius of 1.46-1.64 ISCO for $a_{*}=0.0$ and 1.69-1.93 ISCO for $a_{*}=0.3$. We discuss the implications that our results may have on the equation of state of ultradense, cold matter and our understanding of the innermost accretion flow onto neutron stars with low surface magnetic fields, and systematic errors related to the reflection models and spacetime metric around less idealized neutron stars., Comment: Accepted for publication in ApJ, 12 pages, 5 Figures, 6 Tables

In the last five years there have been a large number of new time series classification algorithms proposed in the literature. These algorithms have been evaluated on subsets of the 47 data sets in the University of California, Riverside time series classification archive. The archive has recently been expanded to 85 data sets, over half of which have been donated by researchers at the University of East Anglia. Aspects of previous evaluations have made comparisons between algorithms difficult. For example, several different programming languages have been used, experiments involved a single train/test split and some used normalised data whilst others did not. The relaunch of the archive provides a timely opportunity to thoroughly evaluate algorithms on a larger number of datasets. We have implemented 18 recently proposed algorithms in a common Java framework and compared them against two standard benchmark classifiers (and each other) by performing 100 resampling experiments on each of the 85 datasets. We use these results to test several hypotheses relating to whether the algorithms are significantly more accurate than the benchmarks and each other. Our results indicate that only 9 of these algorithms are significantly more accurate than both benchmarks and that one classifier, the Collective of Transformation Ensembles, is significantly more accurate than all of the others. All of our experiments and results are reproducible: we release all of our code, results and experimental details and we hope these experiments form the basis for more rigorous testing of new algorithms in the future.

We analyzed near-infrared data of the nearby galaxy IC5063 taken with the Very Large Telescope SINFONI instrument. IC5063 is an elliptical galaxy that has a radio jet nearly aligned with the major axis of a gas disk in its center. The data reveal multiple signatures of molecular and atomic gas that has been kinematically distorted by the passage of the jet plasma or cocoon within an area of ~1 kpc^2. Concrete evidence that the interaction of the jet with the gas causes the gas to accelerate comes from the detection of outflows in four different regions along the jet trail: near the two radio lobes, between the radio emission tip and the optical narrow-line-region cone, and at a region with diffuse 17.8 GHz emission midway between the nucleus and the north radio lobe. The outflow in the latter region is biconical, centered 240 pc away from the nucleus, and oriented perpendicularly to the jet trail. The diffuse emission that is observed as a result of the gas entrainment or scattering unfolds around the trail and away from the nucleus with increasing velocity. It overall extends for >700 pc parallel and perpendicular to the trail. Near the outflow starting points, the gas has a velocity excess of 600 km/s to 1200 km/s with respect to ordered motions, as seen in [FeII], Pa alpha, or H2 lines. High H2 (1-0) S(3)/S(1) flux ratios indicate non-thermal excitation of gas in the diffuse outflow., Comment: Accepted for publication in ApJ

We present the first high signal-to-noise XMM-Newton observations of the broad-line radio galaxy 3C 411. After fitting various spectral models, an absorbed double power-law continuum and a blurred relativistic disk reflection model (kdblur) are found to be equally plausible descriptions of the data. While the softer power-law component ($\Gamma$=2.11) of the double power-law model is entirely consistent with that found in Seyfert galaxies (and hence likely originates from a disk corona), the additional power law component is very hard ($\Gamma$=1.05); amongst the AGN zoo, only flat-spectrum radio quasars have such hard spectra. Together with the very flat radio-spectrum displayed by this source, we suggest that it should instead be classified as a FSRQ. This leads to potential discrepancies regarding the jet inclination angle, with the radio morphology suggesting a large jet inclination but the FSRQ classification suggesting small inclinations. The kdblur model predicts an inner disk radius of at most 20 r$_g$ and relativistic reflection.