Search

This paper presents an active inference method for Extended Finite State Machines, where inputs and outputs are parametrized, and transitions can be conditioned by guards involving input parameters and internal variables called registers. The method applies to (software) systems that cannot be reset, so it learns an EFSM model of the system on a single trace., Comment: 14 pages (last page blank), 8 figures, 4 algorithms Submitted to LearnAut workshop 2024 (not published)

Multilayer moir\'e materials can exhibit topological electronic features yet are inherently quasiperiodic -- leading to wave function interference whose Anderson-localizing tendency can be mitigated by topology. We consider a quasiperiodic variant of the chiral Bistritzer-MacDonald model for twisted bilayer graphene with two incommensurate moir\'e potentials that serves as a toy model for twisted trilayer. We observe "filaments" linking magic angles with enhanced density of states and fractal wave functions that evade localization; states away from the filaments mimic fractal surface states of dirty topological superconductors. We demonstrate that topological quasiperiodicity can broadly enhance superconductivity without magic-angle fine-tuning., Comment: 6 + 22 pages, 5 + 9 figures

The aim in many sciences is to understand the mechanisms that underlie the observed distribution of variables, starting from a set of initial hypotheses. Causal discovery allows us to infer mechanisms as sets of cause and effect relationships in a generalized way -- without necessarily tailoring to a specific domain. Causal discovery algorithms search over a structured hypothesis space, defined by the set of directed acyclic graphs, to find the graph that best explains the data. For high-dimensional problems, however, this search becomes intractable and scalable algorithms for causal discovery are needed to bridge the gap. In this paper, we define a novel causal graph partition that allows for divide-and-conquer causal discovery with theoretical guarantees. We leverage the idea of a superstructure -- a set of learned or existing candidate hypotheses -- to partition the search space. We prove under certain assumptions that learning with a causal graph partition always yields the Markov Equivalence Class of the true causal graph. We show our algorithm achieves comparable accuracy and a faster time to solution for biologically-tuned synthetic networks and networks up to ${10^4}$ variables. This makes our method applicable to gene regulatory network inference and other domains with high-dimensional structured hypothesis spaces.

We report the detection of seven new galaxy candidates with redshift $z$ $>$ 11 within the JWST Advanced Deep Extragalactic Survey (JADES) GOODS-S and GOODS-N fields. These new candidates are detected through meticulous analysis of NIRCam photometry in eight filters spanning a wavelength range of 0.8-5.0 $\mu$m. Photometric redshifts of these galaxy candidates are independently measured utilizing spectral energy distribution (SED) fitting techniques using \texttt{EAZY} and \texttt{BAGPIPES} codes, followed by visual scrutiny. Two of these galaxy candidates are located in GOODS-S field, while the remaining five galaxies are located in GOODS-N field. Our analysis reveals that the stellar masses of these galaxies typically range from log $M_{\ast}$/$M_{\odot}$ = 7.75--8.75. Futhermore, these galaxies are typically young with their mass-weighted ages spanning from 80 to 240 Myr. Their specific star formation rates (sSFR), quantified as $\log (\text{sSFR}/\text{Gyr}$), are measured to vary between $\sim 0.95$ to 1.46. These new galaxy candidates offer a robust sample for probing the physical properties of galaxies within the first few hundred Myr of the history of the Universe. We also analyze the relationship between star formation rate (SFR) and stellar mass ($M_\ast$) within our sample. Using linear regression, our analysis yields a slope of $0.71 \pm 0.12$, which we then compare with results from previous studies. Continued investigation through spectroscopic analysis using JWST/NIRSpec is needed to spectroscopically confirm these high-redshift galaxy candidates and investigate further into their physical properties. We plan to follow up on these candidates with future NIRSpec observations., Comment: 12 pages, 10 figures, Submitted to MNRAS

Deep learning has emerged as a powerful method for extracting valuable information from large volumes of data. However, when new training data arrives continuously (i.e., is not fully available from the beginning), incremental training suffers from catastrophic forgetting (i.e., new patterns are reinforced at the expense of previously acquired knowledge). Training from scratch each time new training data becomes available would result in extremely long training times and massive data accumulation. Rehearsal-based continual learning has shown promise for addressing the catastrophic forgetting challenge, but research to date has not addressed performance and scalability. To fill this gap, we propose an approach based on a distributed rehearsal buffer that efficiently complements data-parallel training on multiple GPUs, allowing us to achieve short runtime and scalability while retaining high accuracy. It leverages a set of buffers (local to each GPU) and uses several asynchronous techniques for updating these local buffers in an embarrassingly parallel fashion, all while handling the communication overheads necessary to augment input mini-batches (groups of training samples fed to the model) using unbiased, global sampling. In this paper we explore the benefits of this approach for classification models. We run extensive experiments on up to 128 GPUs of the ThetaGPU supercomputer to compare our approach with baselines representative of training-from-scratch (the upper bound in terms of accuracy) and incremental training (the lower bound). Results show that rehearsal-based continual learning achieves a top-5 classification accuracy close to the upper bound, while simultaneously exhibiting a runtime close to the lower bound., Comment: 2024 IEEE 24th International Symposium on Cluster, Cloud and Internet Computing (CCGrid), May 2024, Philadelphia (PA), United States

We use the `Middle Ages Galaxy Properties with Integral field spectroscopy' (MAGPI) survey to investigate whether galaxies have evolved in the distribution of their stellar angular momentum in the past 3-4 Gyr, as probed by the observational proxy for spin, $\lambda_{R}$. We use 2D stellar kinematics to measure $\lambda_{R}$ along with detailed photometric models to estimate galaxy ellipticity. The combination of these measurements quantifies the kinematic classes of `fast rotators' and the rarer `slow rotators', which show no regular rotation in their line-of-sight velocity fields. We compare 51 MAGPI galaxies with $\log_{10} (M_{\star}/\mathrm{M}_\odot) > 10$ to carefully drawn samples of MaNGA galaxies in the local Universe, selected to represent possible descendants of the MAGPI progenitors. The EAGLE simulations are used to identify possible evolutionary pathways between the two samples, explicitly accounting for progenitor bias in our results and the varied evolutionary pathways a galaxy might take between the two epochs. We find that the occurrence of slow rotating galaxies is unchanged between the MAGPI ($z \sim 0.3$) and MaNGA ($z \sim 0$) samples, suggesting the massive slow rotator population was already in place $\sim 4$ Gyr ago and has not accumulated since. There is a hint of the MAGPI sample having an excess of high $\lambda_{R}$ galaxies compared to the MaNGA sample, corresponding to more ordered rotation, but statistically the samples are not significantly different. The large-scale stellar kinematics, as quantified through the $\lambda_{R}$ parameter, of galaxies at $z \sim 0.3$ have already evolved into the diversity of structures seen today in the local Universe., Comment: 10 pages, 7 figures, accepted for publication in MNRAS

The azimuthal correlation angle, $\Delta\phi$, between the scattered lepton and the leading jet in deep inelastic $e^{\pm}p$ scattering at HERA has been studied using data collected with the ZEUS detector at a centre-of-mass energy of $\sqrt{s} = 318 \;\mathrm{GeV}$, corresponding to an integrated luminosity of $326 \;\mathrm{pb}^{-1}$. A measurement of jet cross sections in the laboratory frame was made in a fiducial region corresponding to photon virtuality $10 \;\mathrm{GeV}^2 < Q^2 < 350 \;\mathrm{GeV}^2$, inelasticity $0.04 < y < 0.7$, outgoing lepton energy $E_e > 10 \;\mathrm{GeV}$, lepton polar angle $140^\circ < \theta_e < 180^\circ$, jet transverse momentum $2.5 \;\mathrm{GeV} < p_\mathrm{T,jet} < 30 \;\mathrm{GeV}$, and jet pseudorapidity $-1.5 < \eta_\mathrm{jet} < 1.8$. Jets were reconstructed using the $k_\mathrm{T}$ algorithm with the radius parameter $R = 1$. The leading jet in an event is defined as the jet that carries the highest $p_\mathrm{T,jet}$. Differential cross sections, $d\sigma/d\Delta\phi$, were measured as a function of the azimuthal correlation angle in various ranges of leading-jet transverse momentum, photon virtuality and jet multiplicity. Perturbative calculations at $\mathcal{O}(\alpha_{s}^2)$ accuracy successfully describe the data within the fiducial region, although a lower level of agreement is observed near $\Delta\phi \rightarrow \pi$ for events with high jet multiplicity, due to limitations of the perturbative approach in describing soft phenomena in QCD. The data are equally well described by Monte Carlo predictions that supplement leading-order matrix elements with parton showering.

We investigate usage of nonlinear wave chaotic amorphous silicon (a-Si) cavities as physically unclonable functions (PUF). Machine learning attacks on integrated electronic PUFs have been demonstrated to be very effective at modeling PUF behavior. Such attacks on integrated a-Si photonic PUFs are investigated through application of algorithms including linear regression, k-nearest neighbor, decision tree ensembles (random forests and gradient boosted trees), and deep neural networks (DNNs). We found that DNNs performed the best among all the algorithms studied but still failed to completely break the a-Si PUF security which we quantify through a private information metric. Furthermore, machine learning resistance of a-Si PUFs were found to be directly related to the strength of their nonlinear response.

Reinforcement learning from human feedback (RLHF) has emerged as a central tool for language model alignment. We consider online exploration in RLHF, which exploits interactive access to human or AI feedback by deliberately encouraging the model to produce diverse, maximally informative responses. By allowing RLHF to confidently stray from the pre-trained model, online exploration offers the possibility of novel, potentially super-human capabilities, but its full potential as a paradigm for language model training has yet to be realized, owing to computational and statistical bottlenecks in directly adapting existing reinforcement learning techniques. We propose a new algorithm for online exploration in RLHF, Exploratory Preference Optimization (XPO), which is simple and practical -- a one-line change to (online) Direct Preference Optimization (DPO; Rafailov et al., 2023) -- yet enjoys the strongest known provable guarantees and promising empirical performance. XPO augments the DPO objective with a novel and principled exploration bonus, empowering the algorithm to explore outside the support of the initial model and human feedback data. In theory, we show that XPO is provably sample-efficient and converges to a near-optimal language model policy under natural exploration conditions, irrespective of whether the initial model has good coverage. Our analysis, which builds on the observation that DPO implicitly performs a form of $Q^{\star}$-approximation (or, Bellman error minimization), combines previously disparate techniques from language modeling and theoretical reinforcement learning in a serendipitous fashion through the perspective of KL-regularized Markov decision processes. Empirically, we find that XPO is more sample-efficient than non-exploratory DPO variants in a preliminary evaluation.

Sample-efficiency and reliability remain major bottlenecks toward wide adoption of reinforcement learning algorithms in continuous settings with high-dimensional perceptual inputs. Toward addressing these challenges, we introduce a new theoretical framework, RichCLD (Rich-Observation RL with Continuous Latent Dynamics), in which the agent performs control based on high-dimensional observations, but the environment is governed by low-dimensional latent states and Lipschitz continuous dynamics. Our main contribution is a new algorithm for this setting that is provably statistically and computationally efficient. The core of our algorithm is a new representation learning objective; we show that prior representation learning schemes tailored to discrete dynamics do not naturally extend to the continuous setting. Our new objective is amenable to practical implementation, and empirically, we find that it compares favorably to prior schemes in a standard evaluation protocol. We further provide several insights into the statistical complexity of the RichCLD framework, in particular proving that certain notions of Lipschitzness that admit sample-efficient learning in the absence of rich observations are insufficient in the rich-observation setting., Comment: 63 pages, 4 figures, published at ICML 2024

Random testing approaches work by generating inputs at random, or by selecting inputs randomly from some pre-defined operational profile. One long-standing question that arises in this and other testing contexts is as follows: When can we stop testing? At what point can we be certain that executing further tests in this manner will not explore previously untested (and potentially buggy) software behaviors? This is analogous to the question in Machine Learning, of how many training examples are required in order to infer an accurate model. In this paper we show how probabilistic approaches to answer this question in Machine Learning (arising from Computational Learning Theory) can be applied in our testing context. This enables us to produce an upper bound on the number of tests that are required to achieve a given level of adequacy. We are the first to enable this from only knowing the number of coverage targets (e.g. lines of code) in the source code, without needing to observe a sample test executions. We validate this bound on a large set of Java units, and an autonomous driving system., Comment: Preprint, accepted to FSE 2024

We present Causal Amortized Active Structure Learning (CAASL), an active intervention design policy that can select interventions that are adaptive, real-time and that does not require access to the likelihood. This policy, an amortized network based on the transformer, is trained with reinforcement learning on a simulator of the design environment, and a reward function that measures how close the true causal graph is to a causal graph posterior inferred from the gathered data. On synthetic data and a single-cell gene expression simulator, we demonstrate empirically that the data acquired through our policy results in a better estimate of the underlying causal graph than alternative strategies. Our design policy successfully achieves amortized intervention design on the distribution of the training environment while also generalizing well to distribution shifts in test-time design environments. Further, our policy also demonstrates excellent zero-shot generalization to design environments with dimensionality higher than that during training, and to intervention types that it has not been trained on.

This study empirically investigates claims of the increasing ubiquity of artificial intelligence (AI) within roughly 80 million research publications across 20 diverse scientific fields, by examining the change in scholarly engagement with AI from 1985 through 2022. We observe exponential growth, with AI-engaged publications increasing approximately thirteenfold (13x) across all fields, suggesting a dramatic shift from niche to mainstream. Moreover, we provide the first empirical examination of the distribution of AI-engaged publications across publication venues within individual fields, with results that reveal a broadening of AI engagement within disciplines. While this broadening engagement suggests a move toward greater disciplinary integration in every field, increased ubiquity is associated with a semantic tension between AI-engaged research and more traditional disciplinary research. Through an analysis of tens of millions of document embeddings, we observe a complex interplay between AI-engaged and non-AI-engaged research within and across fields, suggesting that increasing ubiquity is something of an oil-and-water phenomenon -- AI-engaged work is spreading out over fields, but not mixing well with non-AI-engaged work.

Effective evaluation of language models remains an open challenge in NLP. Researchers and engineers face methodological issues such as the sensitivity of models to evaluation setup, difficulty of proper comparisons across methods, and the lack of reproducibility and transparency. In this paper we draw on three years of experience in evaluating large language models to provide guidance and lessons for researchers. First, we provide an overview of common challenges faced in language model evaluation. Second, we delineate best practices for addressing or lessening the impact of these challenges on research. Third, we present the Language Model Evaluation Harness (lm-eval): an open source library for independent, reproducible, and extensible evaluation of language models that seeks to address these issues. We describe the features of the library as well as case studies in which the library has been used to alleviate these methodological concerns.

We present a study of kinematic asymmetries from the integral field spectroscopic surveys MAGPI and SAMI. By comparing the asymmetries in the ionsied gas and stars, we aim to disentangle the physical processes that contribute to kinematic disturbances. We normalise deviations from circular motion by $S_{05}$, allowing us to study kinematic asymmetries in the stars and gas, regardless of kinematic temperature. We find a similar distribution of stellar asymmetries in galaxies where we do and do not detect ionised gas, suggesting that whatever is driving the stellar asymmetries does not always lead to gas removal. In both MAGPI and SAMI, we find an anti-correlation between stellar asymmetry and stellar mass, that is absent in the gas asymmetries. After stellar mass and mean-stellar-age matching distributions, we find that at all stellar masses, MAGPI galaxies display larger stellar asymmetry compared to SAMI galaxies. In both MAGPI and SAMI galaxies, we find that star-forming galaxies with old mean-stellar-ages typically have larger asymmetries in their gas compared to their stars, whereas galaxies with young mean-stellar-ages have larger asymmetries in their stars compared to their gas. We suggest that this results from continuous, clumpy accretion of gas., Comment: 12 pages, 10 figures

We introduce SciQAG, a novel framework for automatically generating high-quality science question-answer pairs from a large corpus of scientific literature based on large language models (LLMs). SciQAG consists of a QA generator and a QA evaluator, which work together to extract diverse and research-level questions and answers from scientific papers. Utilizing this framework, we construct a large-scale, high-quality, open-ended science QA dataset containing 188,042 QA pairs extracted from 22,743 scientific papers across 24 scientific domains. We also introduce SciQAG-24D, a new benchmark task designed to evaluate the science question-answering ability of LLMs. Extensive experiments demonstrate that fine-tuning LLMs on the SciQAG dataset significantly improves their performance on both open-ended question answering and scientific tasks. To foster research and collaboration, we make the datasets, models, and evaluation codes publicly available, contributing to the advancement of science question answering and developing more interpretable and reasoning-capable AI systems.

This paper presents a comprehensive analysis of explainable fact-checking through a series of experiments, focusing on the ability of large language models to verify public health claims and provide explanations or justifications for their veracity assessments. We examine the effectiveness of zero/few-shot prompting and parameter-efficient fine-tuning across various open and closed-source models, examining their performance in both isolated and joint tasks of veracity prediction and explanation generation. Importantly, we employ a dual evaluation approach comprising previously established automatic metrics and a novel set of criteria through human evaluation. Our automatic evaluation indicates that, within the zero-shot scenario, GPT-4 emerges as the standout performer, but in few-shot and parameter-efficient fine-tuning contexts, open-source models demonstrate their capacity to not only bridge the performance gap but, in some instances, surpass GPT-4. Human evaluation reveals yet more nuance as well as indicating potential problems with the gold explanations.

The ability to accomplish manipulation and locomotion tasks in the presence of significant time-varying external loads is a remarkable skill of humans that has yet to be replicated convincingly by humanoid robots. Such an ability will be a key requirement in the environments we envision deploying our robots: dull, dirty, and dangerous. External loads constitute a large model bias, which is typically unaccounted for. In this work, we enable our humanoid robot to engage in loco-manipulation tasks in the presence of significant model bias due to external loads. We propose an online estimation and control framework involving the combination of a physically consistent extended Kalman filter for inertial parameter estimation coupled to a whole-body controller. We showcase our results both in simulation and in hardware, where weights are mounted on Nadia's wrist links as a proxy for engaging in tasks where large external loads are applied to the robot., Comment: Submitted to the 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

Galactic cosmic ray (GCR) particles have a significant impact on the particle-induced background of X-ray observatories, and their flux exhibits substantial temporal variability, potentially influencing background levels. In this study, we present one-day binned high-energy reject rates derived from the Chandra-ACIS and XMM-Newton EPIC-pn instruments, serving as proxies for GCR particle flux. We systematically analyze the ACIS and EPIC-pn reject rates and compare them with the AMS proton flux. Our analysis initially reveals robust correlations between the AMS proton flux and the ACIS/EPIC-pn reject rates when binned over 27-day intervals. However, a closer examination reveals substantial fluctuations within each 27-day bin, indicating shorter-term variability. Upon daily binning, we observe finer. temporal structures in the datasets, demonstrating the presence of recurrent variations with periods of $\sim$ 25 days and 23 days in ACIS and EPIC-pn reject rates, respectively, spanning the years 2014 to 2018. Notably, during the 2016--2017 period, we additionally detect periodicities of $\sim$13.5 days and 9 days in the ACIS and EPIC-pn reject rates, respectively. Intriguingly, we observe a time lag of $\sim$ 6 days between the AMS proton flux and the ACIS/EPIC-pn reject rates during the second half of 2016. This time lag is not visible before 2016 and aftern2017. The underlying physical mechanisms responsible for this time lag remain a subject of ongoing investigation., Comment: 16 pages, 8 figures, accepted for publication in ApJ

Despite the success of adaptive time-stepping in ODE simulation, it has so far seen few applications for Stochastic Differential Equations (SDEs). To simulate SDEs adaptively, methods such as the Virtual Brownian Tree (VBT) have been developed, which can generate Brownian motion (BM) non-chronologically. However, in most applications, knowing only the values of Brownian motion is not enough to achieve a high order of convergence; for that, we must compute time-integrals of BM such as $\int_s^t W_r \, dr$. With the aim of using high order SDE solvers adaptively, we extend the VBT to generate these integrals of BM in addition to the Brownian increments. A JAX-based implementation of our construction is included in the popular Diffrax library (https://github.com/patrick-kidger/diffrax). Since the entire Brownian path produced by VBT is uniquely determined by a single PRNG seed, previously generated samples need not be stored, which results in a constant memory footprint and enables experiment repeatability and strong error estimation. Based on binary search, the VBT's time complexity is logarithmic in the tolerance parameter $\varepsilon$. Unlike the original VBT algorithm, which was only precise at some dyadic times, we prove that our construction exactly matches the joint distribution of the Brownian motion and its time integrals at any query times, provided they are at least $\varepsilon$ apart. We present two applications of adaptive high order solvers enabled by our new VBT. Using adaptive solvers to simulate a high-volatility CIR model, we achieve more than twice the convergence order of constant stepping. We apply an adaptive third order underdamped or kinetic Langevin solver to an MCMC problem, where our approach outperforms the No U-Turn Sampler, while using only a tenth of its function evaluations.

We report on our experiences replicating 7.3 petabytes (PB) of Earth System Grid Federation (ESGF) climate simulation data from Lawrence Livermore National Laboratory (LLNL) in California to Argonne National Laboratory (ANL) in Illinois and Oak Ridge National Laboratory (ORNL) in Tennessee. This movement of some 29 million files, twice, undertaken in order to establish new ESGF nodes at ANL and ORNL, was performed largely automatically by a simple replication tool, a script that invoked Globus to transfer large bundles of files while tracking progress in a database. Under the covers, Globus organized transfers to make efficient use of the high-speed Energy Sciences network (ESnet) and the data transfer nodes deployed at participating sites, and also addressed security, integrity checking, and recovery from a variety of transient failures. This success demonstrates the considerable benefits that can accrue from the adoption of performant data replication infrastructure.

Fully supervised models are predominant in Bayesian active learning. We argue that their neglect of the information present in unlabelled data harms not just predictive performance but also decisions about what data to acquire. Our proposed solution is a simple framework for semi-supervised Bayesian active learning. We find it produces better-performing models than either conventional Bayesian active learning or semi-supervised learning with randomly acquired data. It is also easier to scale up than the conventional approach. As well as supporting a shift towards semi-supervised models, our findings highlight the importance of studying models and acquisition methods in conjunction., Comment: Published at AISTATS 2024

The Galvanostatic Intermittent Titration Technique (GITT) is a ubiquitous method for determining the solid-state diffusivity in lithium-ion battery materials. However, it is notoriously time-consuming and relies upon assumptions whose applicability is questionable. We propose a novel methodology that allows inference of the diffusivity for a more general class of data that is simpler and faster to harvest. We infer the diffusivity (as a function of stoichiometry) by minimising the residual sum of squares between data and solutions to a spherically-symmetric diffusion model in a single representative active material particle. Using data harvested from the NMC cathode of a commercial LG M50 cell we first demonstrate that our method is able to reproduce the diffusivities inferred by the GITT, which requires ten days of galvanostatic intermittent titration data. We then demonstrate that our method reliably reconstructs diffusivity using significantly less data. Despite arising from quick-to-measure data, our method more accurately infers diffusivities. This work is a contribution towards developing faster and more reliable techniques in parameter inference for lithium-ion batteries., Comment: 17 pages, 9 figures

Using adaptive optics with the Multi-Unit Spectroscopic Explorer (MUSE) on the Very Large Telescope (VLT), the Middle Ages Galaxy Properties with Integral Field Spectroscopy (MAGPI) survey allows us to study the spatially resolved Universe at a crucial time of ~4 Gyr ago ($z$ ~ 0.3) when simulations predict the greatest diversity in evolutionary pathways for galaxies. We investigate the radial trends in the star formation (SF) activity and luminosity-weighted stellar ages as a function of offset from the star-forming main sequence (SFMS) for a total of 294 galaxies. Using both H$\alpha$ emission and the 4000 Angstrom break (i.e., D4000) as star formation rate (SFR) tracers, we find overall flat radial profiles for galaxies lying on and above the SFMS, suggestive of physical processes that enhance/regulate SF throughout the entire galaxy disc. However, for galaxies lying below the SFMS, we find positive gradients in SF suggestive of inside-out quenching. Placing our results in context with results from other redshift regimes suggests an evolution in radial trends at $z$ ~ 0.3 for SF galaxies above the SFMS, from uniformly enhanced SF at $z$ ~ 1 and $z$ ~ 0.3 to centrally enhanced SF at $z$ ~ 0 (when averaged over a wide range of mass). We also capture higher local SFRs for galaxies below the SFMS compared to that of $z$ ~ 0, which can be explained by a larger population of quenched satellites in the local Universe and/or different treatments of limitations set by the D4000-sSFR relation., Comment: 19 pages, 15 figures, 4 tables, accepted for publication in MNRAS

First-come first-serve scheduling can result in substantial (up to 10%) of transiently idle nodes on supercomputers. Recognizing that such unfilled nodes are well-suited for deep neural network (DNN) training, due to the flexible nature of DNN training tasks, Liu et al. proposed that the re-scaling DNN training tasks to fit gaps in schedules be formulated as a mixed-integer linear programming (MILP) problem, and demonstrated via simulation the potential benefits of the approach. Here, we introduce MalleTrain, a system that provides the first practical implementation of this approach and that furthermore generalizes it by allowing it use even for DNN training applications for which model information is unknown before runtime. Key to this latter innovation is the use of a lightweight online job profiling advisor (JPA) to collect critical scalability information for DNN jobs -- information that it then employs to optimize resource allocations dynamically, in real time. We describe the MalleTrain architecture and present the results of a detailed experimental evaluation on a supercomputer GPU cluster and several representative DNN training workloads, including neural architecture search and hyperparameter optimization. Our results not only confirm the practical feasibility of leveraging idle supercomputer nodes for DNN training but improve significantly on prior results, improving training throughput by up to 22.3\% without requiring users to provide job scalability information.

Simulators are a pervasive tool in reinforcement learning, but most existing algorithms cannot efficiently exploit simulator access -- particularly in high-dimensional domains that require general function approximation. We explore the power of simulators through online reinforcement learning with {local simulator access} (or, local planning), an RL protocol where the agent is allowed to reset to previously observed states and follow their dynamics during training. We use local simulator access to unlock new statistical guarantees that were previously out of reach: - We show that MDPs with low coverability (Xie et al. 2023) -- a general structural condition that subsumes Block MDPs and Low-Rank MDPs -- can be learned in a sample-efficient fashion with only $Q^{\star}$-realizability (realizability of the optimal state-value function); existing online RL algorithms require significantly stronger representation conditions. - As a consequence, we show that the notorious Exogenous Block MDP problem (Efroni et al. 2022) is tractable under local simulator access. The results above are achieved through a computationally inefficient algorithm. We complement them with a more computationally efficient algorithm, RVFS (Recursive Value Function Search), which achieves provable sample complexity guarantees under a strengthened statistical assumption known as pushforward coverability. RVFS can be viewed as a principled, provable counterpart to a successful empirical paradigm that combines recursive search (e.g., MCTS) with value function approximation., Comment: Fixed a small typo

$ $The classical theory of statistical estimation aims to estimate a parameter of interest under data generated from a fixed design ("offline estimation"), while the contemporary theory of online learning provides algorithms for estimation under adaptively chosen covariates ("online estimation"). Motivated by connections between estimation and interactive decision making, we ask: is it possible to convert offline estimation algorithms into online estimation algorithms in a black-box fashion? We investigate this question from an information-theoretic perspective by introducing a new framework, Oracle-Efficient Online Estimation (OEOE), where the learner can only interact with the data stream indirectly through a sequence of offline estimators produced by a black-box algorithm operating on the stream. Our main results settle the statistical and computational complexity of online estimation in this framework. $\bullet$ Statistical complexity. We show that information-theoretically, there exist algorithms that achieve near-optimal online estimation error via black-box offline estimation oracles, and give a nearly-tight characterization for minimax rates in the OEOE framework. $\bullet$ Computational complexity. We show that the guarantees above cannot be achieved in a computationally efficient fashion in general, but give a refined characterization for the special case of conditional density estimation: computationally efficient online estimation via black-box offline estimation is possible whenever it is possible via unrestricted algorithms. Finally, we apply our results to give offline oracle-efficient algorithms for interactive decision making.

We develop new data structures and algorithms for checking verification queries in NetKAT, a domain-specific language for specifying the behavior of network data planes. Our results extend the techniques obtained in prior work on symbolic automata and provide a framework for building efficient and scalable verification tools. We present KATch, an implementation of these ideas in Scala, featuring an extended set of NetKAT operators that are useful for expressing network-wide specifications, and a verification engine that constructs a bisimulation or generates a counter-example showing that none exists. We evaluate the performance of our implementation on real-world and synthetic benchmarks, verifying properties such as reachability and slice isolation, typically returning a result in well under a second, which is orders of magnitude faster than previous approaches. Our advancements underscore NetKAT's potential as a practical, declarative language for network specification and verification.

Since the turn of the century, astronomers have been exploiting the rich information afforded by combining stellar kinematic maps and imaging in an attempt to recover the intrinsic, three-dimensional (3D) shape of a galaxy. A common intrinsic shape recovery method relies on an expected monotonic relationship between the intrinsic misalignment of the kinematic and morphological axes and the triaxiality parameter. Recent studies have, however, cast doubt about underlying assumptions relating shape and intrinsic kinematic misalignment. In this work, we aim to recover the 3D shape of individual galaxies using their projected stellar kinematic and flux distributions using a supervised machine learning approach with mixture density network (MDN). Using a mock dataset of the EAGLE hydrodynamical cosmological simulation, we train the MDN model for a carefully selected set of common kinematic and photometric parameters. Compared to previous methods, we demonstrate potential improvements achieved with the MDN model to retrieve the 3D galaxy shape along with the uncertainties, especially for prolate and triaxial systems. We make specific recommendations for recovering galaxy intrinsic shapes relevant for current and future integral field spectroscopic galaxy surveys., Comment: Accepted for publication in PASA. 18 pages, 12 figures, 2 tables

Rotational spectroscopy is the most accurate method for determining structures of molecules in the gas phase. It is often assumed that a rotational spectrum is a unique "fingerprint" of a molecule. The availability of large molecular databases and the development of artificial intelligence methods for spectroscopy makes the testing of this assumption timely. In this paper, we pose the determination of molecular structures from rotational spectra as an inverse problem. Within this framework, we adopt a funnel-based approach to search for molecular twins, which are two or more molecules, which have similar rotational spectra but distinctly different molecular structures. We demonstrate that there are twins within standard levels of computational accuracy by generating rotational constants for many molecules from several large molecular databases, indicating the inverse problem is ill-posed. However, some twins can be distinguished by increasing the accuracy of the theoretical methods or by performing additional experiments.