Search

Accurate estimates of causal effects play a key role in decision-making across applications such as healthcare, economics, and operations. In the absence of randomized experiments, a common approach to estimating causal effects uses \textit{covariate adjustment}. In this paper, we study covariate adjustment for discrete distributions from the PAC learning perspective, assuming knowledge of a valid adjustment set $\bZ$, which might be high-dimensional. Our first main result PAC-bounds the estimation error of covariate adjustment by a term that is exponential in the size of the adjustment set; it is known that such a dependency is unavoidable even if one only aims to minimize the mean squared error. Motivated by this result, we introduce the notion of an \emph{$\eps$-Markov blanket}, give bounds on the misspecification error of using such a set for covariate adjustment, and provide an algorithm for $\eps$-Markov blanket discovery; our second main result upper bounds the sample complexity of this algorithm. Furthermore, we provide a misspecification error bound and a constraint-based algorithm that allow us to go beyond $\eps$-Markov blankets to even smaller adjustment sets. Our third main result upper bounds the sample complexity of this algorithm, and our final result combines the first three into an overall PAC bound. Altogether, our results highlight that one does not need to perfectly recover causal structure in order to ensure accurate estimates of causal effects.

Explainability methods are often challenging to evaluate and compare. With a multitude of explainers available, practitioners must often compare and select explainers based on quantitative evaluation metrics. One particular differentiator between explainers is the diversity of explanations for a given dataset; i.e. whether all explanations are identical, unique and uniformly distributed, or somewhere between these two extremes. In this work, we define a complexity measure for explainers, globalness, which enables deeper understanding of the distribution of explanations produced by feature attribution and feature selection methods for a given dataset. We establish the axiomatic properties that any such measure should possess and prove that our proposed measure, Wasserstein Globalness, meets these criteria. We validate the utility of Wasserstein Globalness using image, tabular, and synthetic datasets, empirically showing that it both facilitates meaningful comparison between explainers and improves the selection process for explainability methods.

In clinical In-Vitro Fertilization (IVF), identifying the most viable embryo for transfer is important to increasing the likelihood of a successful pregnancy. Traditionally, this process involves embryologists manually assessing embryos' static morphological features at specific intervals using light microscopy. This manual evaluation is not only time-intensive and costly, due to the need for expert analysis, but also inherently subjective, leading to variability in the selection process. To address these challenges, we develop a multimodal model that leverages both time-lapse video data and Electronic Health Records (EHRs) to predict embryo viability. One of the primary challenges of our research is to effectively combine time-lapse video and EHR data, owing to their inherent differences in modality. We comprehensively analyze our multimodal model with various modality inputs and integration approaches. Our approach will enable fast and automated embryo viability predictions in scale for clinical IVF., Comment: Accepted to MICCAI 2024

We consider the secretary problem through the lens of learning-augmented algorithms. As it is known that the best possible expected competitive ratio is $1/e$ in the classic setting without predictions, a natural goal is to design algorithms that are 1-consistent and $1/e$-robust. Unfortunately, [FY24] provided hardness constructions showing that such a goal is not attainable when the candidates' true values are allowed to scale with $n$. Here, we provide a simple and explicit alternative hardness construction showing that such a goal is not achievable even when the candidates' true values are constants that do not scale with $n$.

Scale-Dependent Dynamic Alignment (SDDA) in Els\"asser field fluctuations is theorized to suppress nonlinearities and modulate the energy spectrum. Limited empirical evidence exists for SDDA within the solar wind turbulence's inertial range. We analyzed data from the WIND mission to assess the effects of compressibility, intermittency, and imbalance on SDDA. SDDA consistently appears at energy-containing scales, with a trend toward misalignment at inertial scales. Compressible fluctuations show no increased alignment; however, their impact on SDDA's overall behavior is minimal. The alignment angles inversely correlate with field gradient intensity, likely due to "anomalous" or "counterpropagating" wave packet interactions. This suggests that SDDA originates from mutual shearing of Els\"asser fields during imbalanced ($\delta \boldsymbol{z}^{\pm} \gg \delta \boldsymbol{z}^{\mp}$) interactions. Rigorous thresholding on field gradient intensity reveals SDDA signatures across much of the inertial range. The scaling of Els\"asser increments' alignment angle, $\Theta^{z}$, steepens with increasing global Alfv\'enic imbalance, while the angle between magnetic and velocity field increments, $\Theta^{ub}$, becomes shallower. $\Theta^{ub}$ only correlates with global Els\"asser imbalance, steepening as the imbalance increases. Furthermore, increasing alignment in $\Theta^{ub}$ persists deep into the inertial range of balanced intervals but collapses at large scales for imbalanced ones. Simplified theoretical analysis and modeling of high-frequency, low-amplitude noise in the velocity field indicate significant impacts on alignment angle measurements even at very low frequencies, with effects growing as global imbalance increases.

Bayes nets are extensively used in practice to efficiently represent joint probability distributions over a set of random variables and capture dependency relations. In a seminal paper, Chickering et al. (JMLR 2004) showed that given a distribution $\mathbb{P}$, that is defined as the marginal distribution of a Bayes net, it is $\mathsf{NP}$-hard to decide whether there is a parameter-bounded Bayes net that represents $\mathbb{P}$. They called this problem LEARN. In this work, we extend the $\mathsf{NP}$-hardness result of LEARN and prove the $\mathsf{NP}$-hardness of a promise search variant of LEARN, whereby the Bayes net in question is guaranteed to exist and one is asked to find such a Bayes net. We complement our hardness result with a positive result about the sample complexity that is sufficient to recover a parameter-bounded Bayes net that is close (in TV distance) to a given distribution $\mathbb{P}$, that is represented by some parameter-bounded Bayes net, generalizing a degree-bounded sample complexity result of Brustle et al. (EC 2020)., Comment: 15 pages, 2 figures

During Parker Solar Probe (Parker) Encounter 15 (E15), we observe an 18-hour period of near subsonic ($\mathrm{M_S \sim}$ 1) and sub-Alfv\'enic (SA), $\mathrm{M_A}$ <<< 1, slow speed solar wind from 22 to 15.6 R$_\odot$. As the most extreme SA interval measured to date and skirting the solar wind sonic point, it is the deepest Parker has probed into the formation and acceleration region of the solar wind in the corona. The stream is also measured by Wind and MMS near 1AU at times consistent with ballistic propagation of this slow stream. We investigate the stream source, properties and potential coronal heating consequences via combining these observations with coronal modeling and turbulence analysis. Through source mapping, in situ evidence and multi-point arrival time considerations of a candidate CME, we determine the stream is a steady (non-transient), long-lived and approximately Parker spiral aligned and arises from overexpanded field lines mapping back to an active region. Turbulence analysis of the Els\"{a}sser variables shows the inertial range scaling of the $\mathrm{\mathbf{z}^{+}}$ mode ($\mathrm{f \sim ^{-3/2}}$) to be dominated by the slab component. We discuss the spectral flattening and difficulties associated with measuring the $\mathrm{\mathbf{z}^{-}}$ spectra, cautioning against making definitive conclusions from the $\mathrm{\mathbf{z}^{-}}$ mode. Despite being more extreme than prior sub-Alfv\'enic intervals, its turbulent nature does not appear to be qualitatively different from previously observed streams. We conclude that this extreme low dynamic pressure solar wind interval (which has the potential for extreme space weather conditions) is a large, steady structure spanning at least to 1AU., Comment: 14 figures

We study the problem of online unweighted bipartite matching with $n$ offline vertices and $n$ online vertices where one wishes to be competitive against the optimal offline algorithm. While the classic RANKING algorithm of Karp et al. [1990] provably attains competitive ratio of $1-1/e > 1/2$, we show that no learning-augmented method can be both 1-consistent and strictly better than $1/2$-robust under the adversarial arrival model. Meanwhile, under the random arrival model, we show how one can utilize methods from distribution testing to design an algorithm that takes in external advice about the online vertices and provably achieves competitive ratio interpolating between any ratio attainable by advice-free methods and the optimal ratio of 1, depending on the advice quality., Comment: Accepted into ICML 2024

The advent of generative AI models holds tremendous potential for aiding teachers in the generation of pedagogical materials. However, numerous knowledge gaps concerning the behavior of these models obfuscate the generation of research-informed guidance for their effective usage. Here we assess trends in prompt specificity, variability, and weaknesses in foreign language teacher lesson plans generated by zero-shot prompting in ChatGPT. Iterating a series of prompts that increased in complexity, we found that output lesson plans were generally high quality, though additional context and specificity to a prompt did not guarantee a concomitant increase in quality. Additionally, we observed extreme cases of variability in outputs generated by the same prompt. In many cases, this variability reflected a conflict between 20th century versus 21st century pedagogical practices. These results suggest that the training of generative AI models on classic texts concerning pedagogical practices may represent a currently underexplored topic with the potential to bias generated content towards teaching practices that have been long refuted by research. Collectively, our results offer immediate translational implications for practicing and training foreign language teachers on the use of AI tools. More broadly, these findings reveal the existence of generative AI output trends that have implications for the generation of pedagogical materials across a diversity of content areas.

Teaching robots novel skills with demonstrations via human-in-the-loop data collection techniques like kinesthetic teaching or teleoperation puts a heavy burden on human supervisors. In contrast to this paradigm, it is often significantly easier to provide raw, action-free visual data of tasks being performed. Moreover, this data can even be mined from video datasets or the web. Ideally, this data can serve to guide robot learning for new tasks in novel environments, informing both "what" to do and "how" to do it. A powerful way to encode both the "what" and the "how" is to infer a well-shaped reward function for reinforcement learning. The challenge is determining how to ground visual demonstration inputs into a well-shaped and informative reward function. We propose a technique Rank2Reward for learning behaviors from videos of tasks being performed without access to any low-level states and actions. We do so by leveraging the videos to learn a reward function that measures incremental "progress" through a task by learning how to temporally rank the video frames in a demonstration. By inferring an appropriate ranking, the reward function is able to guide reinforcement learning by indicating when task progress is being made. This ranking function can be integrated into an adversarial imitation learning scheme resulting in an algorithm that can learn behaviors without exploiting the learned reward function. We demonstrate the effectiveness of Rank2Reward at learning behaviors from raw video on a number of tabletop manipulation tasks in both simulations and on a real-world robotic arm. We also demonstrate how Rank2Reward can be easily extended to be applicable to web-scale video datasets., Comment: ICRA 2024

House allocation refers to the problem where $m$ houses are to be allocated to $n$ agents so that each agent receives one house. Since an envy-free house allocation does not always exist, we consider finding such an allocation in the presence of subsidy. We show that computing an envy-free allocation with minimum subsidy is NP-hard in general, but can be done efficiently if $m$ differs from $n$ by an additive constant or if the agents have identical utilities.

Causal graph discovery is a significant problem with applications across various disciplines. However, with observational data alone, the underlying causal graph can only be recovered up to its Markov equivalence class, and further assumptions or interventions are necessary to narrow down the true graph. This work addresses the causal discovery problem under the setting of stochastic interventions with the natural goal of minimizing the number of interventions performed. We propose the following stochastic intervention model which subsumes existing adaptive noiseless interventions in the literature while capturing scenarios such as fat-hand interventions and CRISPR gene knockouts: any intervention attempt results in an actual intervention on a random subset of vertices, drawn from a distribution dependent on attempted action. Under this model, we study the two fundamental problems in causal discovery of verification and search and provide approximation algorithms with polylogarithmic competitive ratios and provide some preliminary experimental results., Comment: Accepted into AISTATS 2024

This study explores the impact of a turbulent scattering mechanism, akin to those influencing solar and galactic cosmic rays propagating in the interplanetary medium, on the population of suprathermal electrons in the solar wind. We employ a Fokker-Planck equation to model the radial evolution of electron pitch angle distributions under the action of magnetic focusing, which moves the electrons away from isotropy, and of a diffusion process that tends to bring them back to it. We compare the steady-state solutions of this Fokker-Planck equation with data obtained from the Solar Orbiter and Parker Solar Probe missions and find a remarkable agreement, varying the turbulent mean free path as the sole free parameter in our model. The obtained mean free paths are of the order of the astronomical unit, and display weak dependence on electron energy within the $100$ eV to $1$ keV range. This value is notably lower than Coulomb collision estimates but aligns well with observed mean free paths of low-rigidity solar energetic particles events. The strong agreement between our model and observations leads us to conclude that the hypothesis of turbulent scattering at work on electrons at all heliospheric distances is justified. We discuss several implications, notably the existence of a low Knudsen number region at large distances from the Sun, which offers a natural explanation for the presence of an isotropic ``halo'' component at all distances from the Sun -- electrons being isotropized in this distant region before travelling back into the inner part of the interplanetary medium., Comment: 23 pages, 13 figures, 1 table

Several recent methods for interpretability model feature interactions by looking at the Hessian of a neural network. This poses a challenge for ReLU networks, which are piecewise-linear and thus have a zero Hessian almost everywhere. We propose SmoothHess, a method of estimating second-order interactions through Stein's Lemma. In particular, we estimate the Hessian of the network convolved with a Gaussian through an efficient sampling algorithm, requiring only network gradient calls. SmoothHess is applied post-hoc, requires no modifications to the ReLU network architecture, and the extent of smoothing can be controlled explicitly. We provide a non-asymptotic bound on the sample complexity of our estimation procedure. We validate the superior ability of SmoothHess to capture interactions on benchmark datasets and a real-world medical spirometry dataset., Comment: Accepted to NeurIPS 2023 as a conference paper

Ion scale wave events or "wave storms" in the solar wind are characterised by enhancements in magnetic field fluctuations as well as coherent magnetic field polarisation signatures at or around the local ion cyclotron frequencies. In this paper we study in detail one such wave event from Parker Solar Probe's (PSP) fourth encounter, consisting of an initial period of left-handed (LH) polarisation abruptly transitioning to a strong period of right-handed (RH) polarisation, accompanied by clear core-beam structure in both the alpha and proton velocity distribution functions. A linear stability analysis shows that the LH polarised waves are anti-Sunward propagating Alfv\'en/ion cyclotron (A/IC) waves primarily driven by a proton cyclotron instability in the proton core population, and the RH polarised waves are anti-Sunward propagating fast magnetosonic/whistler (FM/W) waves driven by a firehose-like instability in the secondary alpha beam population. The abrupt transition from LH to RH is caused by a drop in the proton core temperature anisotropy. We find very good agreement between the frequencies and polarisations of the unstable wave modes as predicted by linear theory and those observed in the magnetic field spectra. Given the ubiquity of ion scale wave signatures observed by PSP, this work gives insight into which exact instabilities may be active and mediating energy transfer in wave-particle interactions in the inner heliosphere, as well as highlighting the role a secondary alpha population may play as a rarely considered source of free energy available for producing wave activity.

Switchbacks are sudden and large deflections in the magnetic field that Parker Solar Probe frequently observes in the inner heliosphere. Their ubiquitous occurrence has prompted numerous studies to determine their nature and origin. Our goal is to describe the boundary of these switchbacks using a series of events detected during the spacecraft's first encounter with the Sun. Using FIELDS and SWEAP data, we investigate different methods for determining the boundary normal. The observed boundaries are arc-polarized structures with a rotation that is always contained in a plane. Classical minimum variance analysis (MVA) gives misleading results and overestimates the number of rotational discontinuities. We propose a robust geometric method to identify the nature of these discontinuities, which involves determining whether or not the plane that contains them also includes the origin ($\textbf{B}=0$). Most boundaries appear to have the same characteristics as tangential discontinuities in the context of switchbacks, with little evidence for having rotational discontinuities. We find no effect of the size of the Parker spiral deviation. Furthermore, the thickness of the boundary is within MHD scales. We conclude that most of the switchback boundaries observed by Parker Solar Probe are likely to be closed, in contrast to previous studies. Our results suggest that their erosion may be much slower than expected., Comment: 11 pages, 6 figures, Animations available at https://doi.org/10.5281/zenodo.8424748

We establish finite-sample guarantees for efficient proper learning of bounded-degree polytrees, a rich class of high-dimensional probability distributions and a subclass of Bayesian networks, a widely-studied type of graphical model. Recently, Bhattacharyya et al. (2021) obtained finite-sample guarantees for recovering tree-structured Bayesian networks, i.e., 1-polytrees. We extend their results by providing an efficient algorithm which learns $d$-polytrees in polynomial time and sample complexity for any bounded $d$ when the underlying undirected graph (skeleton) is known. We complement our algorithm with an information-theoretic sample complexity lower bound, showing that the dependence on the dimension and target accuracy parameters are nearly tight., Comment: Fixed some typos. Added some discussions. Accepted to ALT 2024

We present the results of processing the effects of the powerful Gamma Ray Burst GRB221009A captured by the charged particle detectors (electrostatic analyzers and solid-state detectors) onboard spacecraft at different points in the heliosphere on October 9, 2022. To follow the GRB221009A propagation through the heliosphere we used the electron and proton flux measurements from solar missions Solar Orbiter and STEREO-A; Earth magnetosphere and the solar wind missions THEMIS and Wind; meteorological satellites POES15, POES19, MetOp3; and MAVEN - a NASA mission orbiting Mars. GRB221009A had a structure of four bursts: less intense Pulse 1 - the triggering impulse - was detected by gamma-ray observatories at 131659 UT (near the Earth); the most intense Pulses 2 and 3 were detected on board all the spacecraft from the list, and Pulse 4 detected in more than 500 s after Pulse 1. Due to their different scientific objectives, the spacecraft, which data was used in this study, were separated by more than 1 AU (Solar Orbiter and MAVEN). This enabled tracking GRB221009A as it was propagating across the heliosphere. STEREO-A was the first to register Pulse 2 and 3 of the GRB, almost 100 seconds before their detection by spacecraft in the vicinity of Earth. MAVEN detected GRB221009A Pulses 2, 3, and 4 at the orbit of Mars about 237 seconds after their detection near Earth. By processing the time delays observed we show that the source location of the GRB221009A was at RA 288.5 degrees, Dec 18.5 degrees (J2000) with an error cone of 2 degrees

Automated Program Repair (APR) techniques typically rely on a given test-suite to guide the repair process. Apart from the need to provide test oracles, this makes the produced patches prone to test data over-fitting. In this work, instead of relying on test cases, we show how to automatically repair memory safety issues, by leveraging static analysis (specifically Incorrectness Separation Logic) to guide repair. Our proposed approach learns what a desirable patch is by inspecting how close a patch is to fixing the bug based on the feedback from incorrectness separation logic based static analysis (specifically the Pulse analyser), and turning this information into a distribution of probabilities over context free grammars. Furthermore, instead of focusing on heuristics for reducing the search space of patches, we make repair scalable by creating classes of equivalent patches according to the effect they have on the symbolic heap, and then invoking the validation oracle only once per class of patch equivalence. This allows us to efficiently discover repairs even in the presence of a large pool of patch candidates offered by our generic patch synthesis mechanism. Experimental evaluation of our approach was conducted by repairing real world memory errors in OpenSSL, swoole and other subjects. The evaluation results show the scalability and efficacy of our approach in automatically producing high quality patches.

Past observations show that solar wind (SW) acceleration occurs inside the sub-Alfvenic region, reaching the local Alfven speed at typical distances ~ 10 - 20 Rs (solar radii). Recently, Parker Solar Probe (PSP) traversed regions of sub-Alfvenic SW near perihelia in encounters E8-E12 for the first time providing data in these regions. It became evident that properties of the magnetically dominated SW are considerably different from the super-Alfvenic wind. For example, there are changes in relative abundances and drift of alpha particles with respect to protons, as well as in the magnitude of magnetic fluctuations. We use data of the magnetic field from the FIELDS instrument, and construct ion velocity distribution functions (VDFs) from the sub-Alfvenic regions using Solar Probe Analyzer Ions (SPAN-I) data, and run 2.5D and 3D hybrid models of proton-alpha sub-Alfvenic SW plasma. We investigate the nonlinear evolution of the ion kinetic instabilities in several case studies, and quantify the transfer of energy between the protons, alpha particles, and the kinetic waves. The models provide the 3D ion VDFs at the various stages of the instability evolution in the SW frame. By combining observational analysis with the modeling results, we gain insights on the evolution of the ion instabilities, the heating and the acceleration processes of the sub-Alfvenic SW plasma and quantify the exchange of energy between the magnetic and kinetic components. The modeling results suggest that the ion kinetic instabilities are produced locally in the SW, resulting in anisotropic heating of the ions, as observed by PSP., Comment: Accepted for publication in The Astrophysical Journal

Causal discovery from interventional data is an important problem, where the task is to design an interventional strategy that learns the hidden ground truth causal graph $G(V,E)$ on $|V| = n$ nodes while minimizing the number of performed interventions. Most prior interventional strategies broadly fall into two categories: non-adaptive and adaptive. Non-adaptive strategies decide on a single fixed set of interventions to be performed while adaptive strategies can decide on which nodes to intervene on sequentially based on past interventions. While adaptive algorithms may use exponentially fewer interventions than their non-adaptive counterparts, there are practical concerns that constrain the amount of adaptivity allowed. Motivated by this trade-off, we study the problem of $r$-adaptivity, where the algorithm designer recovers the causal graph under a total of $r$ sequential rounds whilst trying to minimize the total number of interventions. For this problem, we provide a $r$-adaptive algorithm that achieves $O(\min\{r,\log n\} \cdot n^{1/\min\{r,\log n\}})$ approximation with respect to the verification number, a well-known lower bound for adaptive algorithms. Furthermore, for every $r$, we show that our approximation is tight. Our definition of $r$-adaptivity interpolates nicely between the non-adaptive ($r=1$) and fully adaptive ($r=n$) settings where our approximation simplifies to $O(n)$ and $O(\log n)$ respectively, matching the best-known approximation guarantees for both extremes. Our results also extend naturally to the bounded size interventions., Comment: Accepted into UAI 2023

Parker Solar Probe (PSP) observes unexpectedly prevalent switchbacks, which are rapid magnetic field reversals that last from seconds to hours, in the inner heliosphere, posing new challenges to understanding their nature, origin, and evolution. In this work, we investigate the thermal states, electron pitch angle distributions, and pressure signatures of both inside and outside switchbacks, separating a switchback into spike, transition region (TR), and quiet period (QP). Based on our analysis, we find that the proton temperature anisotropies in TRs seem to show an intermediate state between spike and QP plasmas. The proton temperatures are more enhanced in spike than in TR and QP, but the alpha temperatures and alpha-to-proton temperature ratios show the opposite trends, implying that the preferential heating mechanisms of protons and alphas are competing in different regions of switchbacks. Moreover, our results suggest that the electron integrated intensities are almost the same across the switchbacks but the electron pitch angle distributions are more isotropic inside than outside switchbacks, implying switchbacks are intact structures but strong scattering of electrons happens inside switchbacks. In addition, the examination of pressures reveals that the total pressures are comparable through an individual switchback, confirming switchbacks are pressure-balanced structures. These characteristics could further our understanding of ion heating, electron scattering, and the structure of switchbacks., Comment: published in ApJ