Search

We present a novel inverse rendering-based framework to estimate the 3D shape (per-pixel surface normals and depth) of objects and scenes from single-view polarization images, the problem popularly known as Shape from Polarization (SfP). The existing physics-based and learning-based methods for SfP perform under certain restrictions, i.e., (a) purely diffuse or purely specular reflections, which are seldom in the real surfaces, (b) availability of the ground truth surface normals for direct supervision that are hard to acquire and are limited by the scanner's resolution, and (c) known refractive index. To overcome these restrictions, we start by learning to separate the partially-polarized diffuse and specular reflection components, which we call reflectance cues, based on a modified polarization reflection model and then estimate shape under mixed polarization through an inverse-rendering based self-supervised deep learning framework called SS-SfP, guided by the polarization data and estimated reflectance cues. Furthermore, we also obtain the refractive index as a non-linear least squares solution. Through extensive quantitative and qualitative evaluation, we establish the efficacy of the proposed framework over simple single-object scenes from DeepSfP dataset and complex in-the-wild scenes from SPW dataset in an entirely self-supervised setting. To the best of our knowledge, this is the first learning-based approach to address SfP under mixed polarization in a completely self-supervised framework., Comment: Published in Pacific Graphics 2023

We argue that Content-based filtering (CBF) and Graph-based methods (GB) complement one another in Academic Search recommendations. The scientific literature can be viewed as a conversation between authors and the audience. CBF uses abstracts to infer authors' positions, and GB uses citations to infer responses from the audience. In this paper, we describe nine differences between CBF and GB, as well as synergistic opportunities for hybrid combinations. Two embeddings will be used to illustrate these opportunities: (1) Specter, a CBF method based on BERT-like deepnet encodings of abstracts, and (2) ProNE, a GB method based on spectral clustering of more than 200M papers and 2B citations from Semantic Scholar.

We investigate positivity properties in quantum field theory (QFT). We find that planar Feynman integrals in QFT, as well as many related quantities, satisfy an infinite number of positivity conditions: the functions, as well as all their signed derivatives, are non-negative in a specified kinematic region. Such functions are known as completely monotonic (CM) in the mathematics literature. A powerful way to certify complete monotonicity is via integral representations. We thus show that it applies to non-planar integrals possessing a Euclidean region, to cosmological correlators, as well as to certain stringy integrals. Motivated by Positive Geometry, we investigate positivity properties in planar maximally supersymmetric Yang-Mills theory. We present evidence, based on known analytic multi-loop results, that the CM property extends to several physical quantities in this theory. This includes the (suitably normalized) finite remainder function of the six-particle maximally-helicity violating (MHV) amplitude, four-point scattering amplitudes on the Coulomb branch, four-point correlation functions, as well as the angle-dependent cusp anomalous dimension. Our findings are however not limited to supersymmetric theories. It is shown that the CM property holds for the QCD and QED cusp anomalous dimensions, to three and four loops, respectively. We comment on open questions, and on possible numerical applications of complete monotonicity., Comment: 6 pages, 2 figures

We compare different local-global principles for torsors under a reductive group G defined over a semiglobal field F. In particular if the F-group G s a retract rational F-variety, we prove that the local global principle holds for the completions with respect to divisorial valuations of F.

UV wind line variability in OB stars appears to be universal. To quantify this variation and to estimate its effect on a mass loss rate determined from a single observation, we use the IUE archive to identify non-peculiar OB stars with well developed but unsaturated Si IV 1400 doublets and at least 10 independent observations. This resulted in 1699 spectra of 25 stars. We use SEI modelling to translate the profile variations into optical depth variations and, hence, variations in measured mass loss rates. These variations quantify the intrinsic error inherent in any mass loss rate derived from a single observation. The derived rates have an overall RMS variation of about 22%, but this differs with effective temperature, being as small at 8% for the hottest stars and up to 45% for the cooler ones. Furthermore, any single determination can differ from the mean by a factor of 2 or more. Our results also imply that mass loss rates determined from non-simultaneous observations (such as UV and ground based data) need not agree. We also use our results to examine the nature of the wind structures responsible for the variability. Our findings suggest that the optical depth variations result from optically very thick structures occulting more or less of the line of sight to the stellar disk. Further, the smaller optical depth variations in the hottest stars suggests that the responsible structures are disrupted in their more powerful winds., Comment: 23 pages, 8 figures, Accepted by ApJ

Numerical simulations of detonation-containing flows have emerged as crucial tools for designing next-generation power and propulsion devices. As these tools mature, it is important for the combustion community to properly understand and isolate grid resolution effects when simulating detonations. To this end, this work provides a comprehensive analysis of the numerical convergence of unsteady detonation simulations, with focus on isolating the impacts of chemical timescale modifications on convergence characteristics in the context of operator splitting. With the aid of an adaptive mesh refinement based flow solver, the convergence analysis is conducted using two kinetics configurations: (1) a simplified three-step model mechanism, in which chemical timescales in the detonation are modified by adjusting activation energies, and (2) a detailed hydrogen mechanism, in which chemical timescales are adjusted through ambient pressure modifications. The convergence of unsteady self-sustained detonations in one-dimensional channels is then analyzed with reference to steady-state theoretical baseline solutions using these mechanisms. The goal of the analysis is to provide a detailed comparison of the effects of grid resolution on both macroscopic (peak pressures and detonation wave speeds) and microscopic (detonation wave structure) quantities of interest, drawing connections between the deviations from steady-state baselines and minimum chemical timescales. This work uncovers resolution-dependent unsteady detonation regimes, and highlights the important role played by not only the chemical timescales, but also the ratio between chemical and induction timescales in the detonation wave structure on simulation convergence properties.

Restless multi-armed bandits (RMAB) extend multi-armed bandits so pulling an arm impacts future states. Despite the success of RMABs, a key limiting assumption is the separability of rewards into a sum across arms. We address this deficiency by proposing restless-multi-armed bandit with global rewards (RMAB-G), a generalization of RMABs to global non-separable rewards. To solve RMAB-G, we develop the Linear- and Shapley-Whittle indices, which extend Whittle indices from RMABs to RMAB-Gs. We prove approximation bounds but also point out how these indices could fail when reward functions are highly non-linear. To overcome this, we propose two sets of adaptive policies: the first computes indices iteratively, and the second combines indices with Monte-Carlo Tree Search (MCTS). Empirically, we demonstrate that our proposed policies outperform baselines and index-based policies with synthetic data and real-world data from food rescue., Comment: 27 pages

Diffusion models show a remarkable ability in generating images that closely mirror the training distribution. However, these models are prone to training data memorization, leading to significant privacy, ethical, and legal concerns, particularly in sensitive fields such as medical imaging. We hypothesize that memorization is driven by the overparameterization of deep models, suggesting that regularizing model capacity during fine-tuning could be an effective mitigation strategy. Parameter-efficient fine-tuning (PEFT) methods offer a promising approach to capacity control by selectively updating specific parameters. However, finding the optimal subset of learnable parameters that balances generation quality and memorization remains elusive. To address this challenge, we propose a bi-level optimization framework that guides automated parameter selection by utilizing memorization and generation quality metrics as rewards. Our framework successfully identifies the optimal parameter set to be updated to satisfy the generation-memorization tradeoff. We perform our experiments for the specific task of medical image generation and outperform existing state-of-the-art training-time mitigation strategies by fine-tuning as few as 0.019% of model parameters. Furthermore, we show that the strategies learned through our framework are transferable across different datasets and domains. Our proposed framework is scalable to large datasets and agnostic to the choice of reward functions. Finally, we show that our framework can be combined with existing approaches for further memorization mitigation.

Concept-based explainability methods provide insight into deep learning systems by constructing explanations using human-understandable concepts. While the literature on human reasoning demonstrates that we exploit relationships between concepts when solving tasks, it is unclear whether concept-based methods incorporate the rich structure of inter-concept relationships. We analyse the concept representations learnt by concept-based models to understand whether these models correctly capture inter-concept relationships. First, we empirically demonstrate that state-of-the-art concept-based models produce representations that lack stability and robustness, and such methods fail to capture inter-concept relationships. Then, we develop a novel algorithm which leverages inter-concept relationships to improve concept intervention accuracy, demonstrating how correctly capturing inter-concept relationships can improve downstream tasks., Comment: Accepted at ICML 2024

The introduction of geometry has proven instrumental in the efforts towards more realistic models for real-world networks. In Geometric Inhomogeneous Random Graphs (GIRGs), Euclidean Geometry induces clustering of the vertices, which is widely observed in networks in the wild. Euclidean Geometry in multiple dimensions however restricts proximity of vertices to those cases where vertices are close in each coordinate. We introduce a large class of GIRG extensions, called BDF-GIRGs, which capture arbitrary hierarchies of the coordinates within the distance function of the vertex feature space. These distance functions have the potential to allow more realistic modeling of the complex formation of social ties in real-world networks, where similarities between people lead to connections. Here, similarity with respect to certain features, such as familial kinship or a shared workplace, suffices for the formation of ties. It is known that - while many key properties of GIRGs, such as log-log average distance and sparsity, are independent of the distance function - the Euclidean metric induces small separators, i.e. sublinear cuts of the unique giant component in GIRGs, whereas no such sublinear separators exist under the component-wise minimum distance. Building on work of Lengler and Todorovi\'{c}, we give a complete classification for the existence of small separators in BDF-GIRGs. We further show that BDF-GIRGs all fulfill a stochastic triangle inequality and thus also exhibit clustering.

We study online classification under smoothed adversaries. In this setting, at each time point, the adversary draws an example from a distribution that has a bounded density with respect to a fixed base measure, which is known apriori to the learner. For binary classification and scalar-valued regression, previous works \citep{haghtalab2020smoothed, block2022smoothed} have shown that smoothed online learning is as easy as learning in the iid batch setting under PAC model. However, we show that smoothed online classification can be harder than the iid batch classification when the label space is unbounded. In particular, we construct a hypothesis class that is learnable in the iid batch setting under the PAC model but is not learnable under the smoothed online model. Finally, we identify a condition that ensures that the PAC learnability of a hypothesis class is sufficient for its smoothed online learnability., Comment: 18 pages

We study online classification when the learner has access to predictions about future examples. We design an online learner whose expected regret is never worse than the worst-case regret, gracefully improves with the quality of the predictions, and can be significantly better than the worst-case regret when the predictions of future examples are accurate. As a corollary, we show that if the learner is always guaranteed to observe data where future examples are easily predictable, then online learning can be as easy as transductive online learning. Our results complement recent work in online algorithms with predictions and smoothed online classification, which go beyond a worse-case analysis by using machine-learned predictions and distributional assumptions respectively., Comment: 24 pages

Data privacy has become a major concern in healthcare due to the increasing digitization of medical records and data-driven medical research. Protecting sensitive patient information from breaches and unauthorized access is critical, as such incidents can have severe legal and ethical complications. Federated Learning (FL) addresses this concern by enabling multiple healthcare institutions to collaboratively learn from decentralized data without sharing it. FL's scope in healthcare covers areas such as disease prediction, treatment customization, and clinical trial research. However, implementing FL poses challenges, including model convergence in non-IID (independent and identically distributed) data environments, communication overhead, and managing multi-institutional collaborations. A systematic review of FL in healthcare is necessary to evaluate how effectively FL can provide privacy while maintaining the integrity and usability of medical data analysis. In this study, we analyze existing literature on FL applications in healthcare. We explore the current state of model security practices, identify prevalent challenges, and discuss practical applications and their implications. Additionally, the review highlights promising future research directions to refine FL implementations, enhance data security protocols, and expand FL's use to broader healthcare applications, which will benefit future researchers and practitioners.

During the first half of the fourth observing run (O4a) of the International Gravitational Wave Network (IGWN), the Zwicky Transient Facility (ZTF) conducted a systematic search for kilonova (KN) counterparts to binary neutron star (BNS) and neutron star-black hole (NSBH) merger candidates. Here, we present a comprehensive study of the five high-significance (FAR < 1 per year) BNS and NSBH candidates in O4a. Our follow-up campaigns relied on both target-of-opportunity observations (ToO) and re-weighting of the nominal survey schedule to maximize coverage. We describe the toolkit we have been developing, Fritz, an instance of SkyPortal, instrumental in coordinating and managing our telescope scheduling, candidate vetting, and follow-up observations through a user-friendly interface. ZTF covered a total of 2841 deg$^2$ within the skymaps of the high-significance GW events, reaching a median depth of g~20.2 mag. We circulated 15 candidates, but found no viable KN counterpart to any of the GW events. Based on the ZTF non-detections of the high-significance events in O4a, we used a Bayesian approach, nimbus, to quantify the posterior probability of KN model parameters that are consistent with our non-detections. Our analysis favors KNe with initial absolute magnitude fainter than -16 mag. The joint posterior probability of a GW170817-like KN associated with all our O4a follow-ups was 64%. Additionally, we use a survey simulation software, simsurvey, to determine that our combined filtered efficiency to detect a GW170817-like KN is 36%, when considering the 5 confirmed astrophysical events in O3 (1 BNS and 4 NSBH), along with our O4a follow-ups. Following Kasliwal et al. (2020), we derived joint constraints on the underlying KN luminosity function based on our O3 and O4a follow-ups, determining that no more than 76% of KNe fading at 1 mag/day can peak at a magnitude brighter than -17.5 mag., Comment: submitted

We define the notion of a marked moduli space as the parameter space of a physical theory together with all of its observables. In geometric examples, this coincides with the mathematical notion of Teichm\"uller space. We propose two new Swampland principles about the geometry of marked moduli spaces: We conjecture that a marked moduli space is always contractible, and moreover, that there is a unique geodesic connecting any pair of points in it with respect to its physical metric. We provide strong evidence for these conjectures for theories with 8 or more supercharges., Comment: 39 pages, 3 figures

We present a comparison review that evaluates popular techniques for garment draping for 3D fashion design, virtual try-ons, and animations. A comparative study is performed between various methods for garment draping of clothing over the human body. These include numerous models, such as physics and machine learning based techniques, collision handling, and more. Performance evaluations and trade-offs are discussed to ensure informed decision-making when choosing the most appropriate approach. These methods aim to accurately represent deformations and fine wrinkles of digital garments, considering the factors of data requirements, and efficiency, to produce realistic results. The research can be insightful to researchers, designers, and developers in visualizing dynamic multi-layered 3D clothing.

Multiple imputation (MI) models can be improved by including auxiliary covariates (AC), but their performance in high-dimensional data is not well understood. We aimed to develop and compare high-dimensional MI (HDMI) approaches using structured and natural language processing (NLP)-derived AC in studies with partially observed confounders. We conducted a plasmode simulation study using data from opioid vs. non-steroidal anti-inflammatory drug (NSAID) initiators (X) with observed serum creatinine labs (Z2) and time-to-acute kidney injury as outcome. We simulated 100 cohorts with a null treatment effect, including X, Z2, atrial fibrillation (U), and 13 other investigator-derived confounders (Z1) in the outcome generation. We then imposed missingness (MZ2) on 50% of Z2 measurements as a function of Z2 and U and created different HDMI candidate AC using structured and NLP-derived features. We mimicked scenarios where U was unobserved by omitting it from all AC candidate sets. Using LASSO, we data-adaptively selected HDMI covariates associated with Z2 and MZ2 for MI, and with U to include in propensity score models. The treatment effect was estimated following propensity score matching in MI datasets and we benchmarked HDMI approaches against a baseline imputation and complete case analysis with Z1 only. HDMI using claims data showed the lowest bias (0.072). Combining claims and sentence embeddings led to an improvement in the efficiency displaying the lowest root-mean-squared-error (0.173) and coverage (94%). NLP-derived AC alone did not perform better than baseline MI. HDMI approaches may decrease bias in studies with partially observed confounders where missingness depends on unobserved factors.

GW230529 is the first compact binary coalescence detected by the LIGO-Virgo-KAGRA collaboration with at least one component mass confidently in the lower mass-gap, corresponding to the range 3-5$M_{\odot}$. If interpreted as a neutron star-black hole merger, this event has the most symmetric mass ratio detected so far and therefore has a relatively high probability of producing electromagnetic (EM) emission. However, no EM counterpart has been reported. At the merger time $t_0$, Swift-BAT and Fermi-GBM together covered 100$\%$ of the sky. Performing a targeted search in a time window $[t_0-20 \text{s},t_0+20 \text{s}]$, we report no detection by the Swift-BAT and the Fermi-GBM instruments. Combining the position-dependent $\gamma-$ray flux upper limits and the gravitational-wave posterior distribution of luminosity distance, sky localization and inclination angle of the binary, we derive constraints on the characteristic luminosity and structure of the jet possibly launched during the merger. Assuming a top-hat jet structure, we exclude at 90$\%$ credibility the presence of a jet which has at the same time an on-axis isotropic luminosity $\gtrsim 10^{48}$ erg s$^{-1}$, in the bolometric band 1 keV-10 MeV, and a jet opening angle $\gtrsim 15$ deg. Similar constraints are derived testing other assumptions about the jet structure profile. Excluding GRB 170817A, the luminosity upper limits derived here are below the luminosity of any GRB observed so far., Comment: 18 pages, 1 table, 11 figures

We consider structural parameterizations of the fundamental Dominating Set problem and its variants in the parameter ecology program. We give improved FPT algorithms and lower bounds under well-known conjectures for dominating set in graphs that are k vertices away from a cluster graph or a split graph. These are graphs in which there is a set of k vertices (called the modulator) whose deletion results in a cluster graph or a split graph. We also call k as the deletion distance (to the appropriate class of graphs). When parameterized by the deletion distance k to cluster graphs - we can find a minimum dominating set (DS) in 3^k n^{O(1)}-time. Within the same time, we can also find a minimum independent dominating set (IDS) or a minimum dominating clique (DC) or a minimum efficient dominating set (EDS) or a minimum total dominating set (TDS). We also show that most of these variants of dominating set do not have polynomial sized kernel. Additionally, we show that when parameterized by the deletion distance k to split graphs - IDS can be solved in 2^k n^{O(1)}-time and EDS can be solved in 3^{k/2}n^{O(1)}., Comment: Some of the results appeared in proceedings of CSR 2018

Deep learning models, particularly Convolutional Neural Networks (CNNs), have demonstrated exceptional performance in diagnosing skin diseases, often outperforming dermatologists. However, they have also unveiled biases linked to specific demographic traits, notably concerning diverse skin tones or gender, prompting concerns regarding fairness and limiting their widespread deployment. Researchers are actively working to ensure fairness in AI-based solutions, but existing methods incur an accuracy loss when striving for fairness. To solve this issue, we propose a `two-biased teachers' (i.e., biased on different sensitive attributes) based approach to transfer fair knowledge into the student network. Our approach mitigates biases present in the student network without harming its predictive accuracy. In fact, in most cases, our approach improves the accuracy of the baseline model. To achieve this goal, we developed a weighted loss function comprising biasing and debiasing loss terms. We surpassed available state-of-the-art approaches to attain fairness and also improved the accuracy at the same time. The proposed approach has been evaluated and validated on two dermatology datasets using standard accuracy and fairness evaluation measures. We will make source code publicly available to foster reproducibility and future research.

Although simplices are trivial from a linear optimization standpoint, the simplex algorithm can exhibit quite complex behavior. In this paper we study the behavior of max-slope pivot rules on (products of) simplices and describe the associated pivot rule polytopes. For simplices, the pivot rule polytopes are combinatorially isomorphic to associahedra. To prove this correspondence, we interpret max-slope pivot rules in terms of the combinatorics of colliding particles on a line. For prisms over simplices, we recover Stasheff's multiplihedra. For products of two simplices we get new realizations of constrainahedra, that capture the combinatorics of certain particle systems in the plane., Comment: 20 pages, 13 figures

Redacted emails satisfy most privacy requirements but they make it more difficult to detect anomalous emails that may be indicative of data exfiltration. In this paper we develop an enhanced method of Active Learning using an information gain maximizing heuristic, and we evaluate its effectiveness in a real world setting where only redacted versions of email could be labeled by human analysts due to privacy concerns. In the first case study we examined how Active Learning should be carried out. We found that model performance was best when a single highly skilled (in terms of the labelling task) analyst provided the labels. In the second case study we used confidence ratings to estimate the labeling uncertainty of analysts and then prioritized instances for labeling based on the expected information gain (the difference between model uncertainty and analyst uncertainty) that would be provided by labelling each instance. We found that the information maximization gain heuristic improved model performance over existing sampling methods for Active Learning. Based on the results obtained, we recommend that analysts should be screened, and possibly trained, prior to implementation of Active Learning in cybersecurity applications. We also recommend that the information gain maximizing sample method (based on expert confidence) should be used in early stages of Active Learning, providing that well-calibrated confidence can be obtained. We also note that the expertise of analysts should be assessed prior to Active Learning, as we found that analysts with lower labelling skill had poorly calibrated (over-) confidence in their labels., Comment: arXiv admin note: substantial text overlap with arXiv:2303.00870

Long gamma-ray bursts (GRBs) are believed to originate from core collapse of massive stars. High-redshift GRBs can probe the star formation and reionization history of the early universe, but their detection remains rare. Here we report the detection of a GRB triggered in the 0.5--4 keV band by the Wide-field X-ray Telescope (WXT) on board the Einstein Probe (EP) mission, designated as EP240315a, whose bright peak was also detected by the Swift Burst Alert Telescope and Konus-Wind through off-line analyses. At a redshift of $z=4.859$, EP240315a showed a much longer and more complicated light curve in the soft X-ray band than in gamma-rays. Benefiting from a large field-of-view ($\sim$3600 deg$^2$) and a high sensitivity, EP-WXT captured the earlier engine activation and extended late engine activity through a continuous detection. With a peak X-ray flux at the faint end of previously known high-$z$ GRBs, the detection of EP240315a demonstrates the great potential for EP to study the early universe via GRBs., Comment: 41 pages, 8 figures, 7 tables

Additive manufacturing (AM), particularly 3D printing, has revolutionized the production of complex structures across various industries. However, ensuring quality and detecting defects in 3D-printed objects remain significant challenges. This study focuses on improving defect detection in 3D-printed cylinders by integrating novel pre-processing techniques such as Region of Interest (ROI) selection, Histogram Equalization (HE), and Details Enhancer (DE) with Convolutional Neural Networks (CNNs), specifically the modified VGG16 model. The approaches, ROIN, ROIHEN, and ROIHEDEN, demonstrated promising results, with the best model achieving an accuracy of 1.00 and an F1-score of 1.00 on the test set. The study also explored the models' interpretability through Local Interpretable Model-Agnostic Explanations and Gradient-weighted Class Activation Mapping, enhancing the understanding of the decision-making process. Furthermore, the modified VGG16 model showed superior computational efficiency with 30713M FLOPs and 15M parameters, the lowest among the compared models. These findings underscore the significance of tailored pre-processing and CNNs in enhancing defect detection in AM, offering a pathway to improve manufacturing precision and efficiency. This research not only contributes to the advancement of 3D printing technology but also highlights the potential of integrating machine learning with AM for superior quality control.

The intricate interplay between spin and lattice degrees of freedom in two-dimensional magnetic materials plays a pivotal role in modifying their magnetic characteristics, engendering hybrid quasiparticles, and implementing functional devices. Herein, we present our comprehensive and in-depth investigations on magnetic and lattice excitations of MnPSe3-xSx (x = 0, 0.5, and 1.5) alloys, utilizing temperature- and polarization-dependent Raman scattering. Our experimental results reveal the occurrence of multiple phase transitions, evidenced by notable changes in phonon self-energy and the appearance or splitting of phonon modes. These emergent phases are tied to the development of long and short-range spin-spin correlations, as well as to spin reorientations or magnetic instabilities. Our analysis of two-magnon excitations as a function of temperature and composition showcases their hybridization with phonons whose degree weakens with increasing x. Moreover, the suppression of spin-dependent phonon intensity in chemically most-disordered MnPSe3-xSx (x = 1.5) suggests that chalcogen substitution offers a control knob of tuning spin and phonon dynamics by modulating concurrently superexchange pathways and a degree of trigonal distortions.

In the contemporary landscape of technological advancements, the automation of manual processes is crucial, compelling the demand for huge datasets to effectively train and test machines. This research paper is dedicated to the exploration and implementation of an automated approach to generate test cases specifically using Large Language Models. The methodology integrates the use of Open AI to enhance the efficiency and effectiveness of test case generation for training and evaluating Large Language Models. This formalized approach with LLMs simplifies the testing process, making it more efficient and comprehensive. Leveraging natural language understanding, LLMs can intelligently formulate test cases that cover a broad range of REST API properties, ensuring comprehensive testing. The model that is developed during the research is trained using manually collected postman test cases or instances for various Rest APIs. LLMs enhance the creation of Postman test cases by automating the generation of varied and intricate test scenarios. Postman test cases offer streamlined automation, collaboration, and dynamic data handling, providing a user-friendly and efficient approach to API testing compared to traditional test cases. Thus, the model developed not only conforms to current technological standards but also holds the promise of evolving into an idea of substantial importance in future technological advancements.

Getting large language models (LLMs) to perform well on the downstream tasks requires pre-training over trillions of tokens. This typically demands a large number of powerful computational devices in addition to a stable distributed training framework to accelerate the training. The growing number of applications leveraging AI/ML had led to a scarcity of the expensive conventional accelerators (such as GPUs), which begs the need for the alternative specialized-accelerators that are scalable and cost-efficient. AWS Trainium is the second-generation machine learning accelerator that has been purposely built for training large deep learning models. Its corresponding instance, Amazon EC2 trn1, is an alternative to GPU instances for LLM training. However, training LLMs with billions of parameters on trn1 is challenging due to its relatively nascent software ecosystem. In this paper, we showcase HLAT: a 7 billion parameter decoder-only LLM pre-trained using trn1 instances over 1.8 trillion tokens. The performance of HLAT is benchmarked against popular open source baseline models including LLaMA and OpenLLaMA, which have been trained on NVIDIA GPUs and Google TPUs, respectively. On various evaluation tasks, we show that HLAT achieves model quality on par with the baselines. We also share the best practice of using the Neuron Distributed Training Library (NDTL), a customized distributed training library for AWS Trainium to achieve efficient training. Our work demonstrates that AWS Trainium powered by the NDTL is able to successfully pre-train state-of-the-art LLM models with high performance and cost-effectiveness.

A key challenge in contrastive learning is to generate negative samples from a large sample set to contrast with positive samples, for learning better encoding of the data. These negative samples often follow a softmax distribution which are dynamically updated during the training process. However, sampling from this distribution is non-trivial due to the high computational costs in computing the partition function. In this paper, we propose an Efficient Markov Chain Monte Carlo negative sampling method for Contrastive learning (EMC$^2$). We follow the global contrastive learning loss as introduced in SogCLR, and propose EMC$^2$ which utilizes an adaptive Metropolis-Hastings subroutine to generate hardness-aware negative samples in an online fashion during the optimization. We prove that EMC$^2$ finds an $\mathcal{O}(1/\sqrt{T})$-stationary point of the global contrastive loss in $T$ iterations. Compared to prior works, EMC$^2$ is the first algorithm that exhibits global convergence (to stationarity) regardless of the choice of batch size while exhibiting low computation and memory cost. Numerical experiments validate that EMC$^2$ is effective with small batch training and achieves comparable or better performance than baseline algorithms. We report the results for pre-training image encoders on STL-10 and Imagenet-100., Comment: 20 pages

Fine-tuning language models (LMs) has demonstrated success in a wide array of downstream tasks. However, as LMs are scaled up, the memory requirements for backpropagation become prohibitively high. Zeroth-order (ZO) optimization methods can leverage memory-efficient forward passes to estimate gradients. More recently, MeZO, an adaptation of ZO-SGD, has been shown to consistently outperform zero-shot and in-context learning when combined with suitable task prompts. In this work, we couple ZO methods with variance reduction techniques to enhance stability and convergence for inference-based LM fine-tuning. We introduce Memory-Efficient Zeroth-Order Stochastic Variance-Reduced Gradient (MeZO-SVRG) and demonstrate its efficacy across multiple LM fine-tuning tasks, eliminating the reliance on task-specific prompts. Evaluated across a range of both masked and autoregressive LMs on benchmark GLUE tasks, MeZO-SVRG outperforms MeZO with up to 20% increase in test accuracies in both full- and partial-parameter fine-tuning settings. MeZO-SVRG benefits from reduced computation time as it often surpasses MeZO's peak test accuracy with a $2\times$ reduction in GPU-hours. MeZO-SVRG significantly reduces the required memory footprint compared to first-order SGD, i.e. by $2\times$ for autoregressive models. Our experiments highlight that MeZO-SVRG's memory savings progressively improve compared to SGD with larger batch sizes., Comment: 29 pages, 25 tables, 9 figures

Trained models are often composed with post-hoc transforms such as temperature scaling (TS), ensembling and stochastic weight averaging (SWA) to improve performance, robustness, uncertainty estimation, etc. However, such transforms are typically applied only after the base models have already been finalized by standard means. In this paper, we challenge this practice with an extensive empirical study. In particular, we demonstrate a phenomenon that we call post-hoc reversal, where performance trends are reversed after applying these post-hoc transforms. This phenomenon is especially prominent in high-noise settings. For example, while base models overfit badly early in training, both conventional ensembling and SWA favor base models trained for more epochs. Post-hoc reversal can also suppress the appearance of double descent and mitigate mismatches between test loss and test error seen in base models. Based on our findings, we propose post-hoc selection, a simple technique whereby post-hoc metrics inform model development decisions such as early stopping, checkpointing, and broader hyperparameter choices. Our experimental analyses span real-world vision, language, tabular and graph datasets from domains like satellite imaging, language modeling, census prediction and social network analysis. On an LLM instruction tuning dataset, post-hoc selection results in > 1.5x MMLU improvement compared to naive selection. Code is available at https://github.com/rishabh-ranjan/post-hoc-reversal., Comment: 9 pages + references + appendix, 7 figures

Sentiment analysis, a vital component in natural language processing, plays a crucial role in understanding the underlying emotions and opinions expressed in textual data. In this paper, we propose an innovative ensemble approach for sentiment analysis for finding fake reviews that amalgamate the predictive capabilities of Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Decision Tree classifiers. Our ensemble architecture strategically combines these diverse models to capitalize on their strengths while mitigating inherent weaknesses, thereby achieving superior accuracy and robustness in fake review prediction. By combining all the models of our classifiers, the predictive performance is boosted and it also fosters adaptability to varied linguistic patterns and nuances present in real-world datasets. The metrics accounted for on fake reviews demonstrate the efficacy and competitiveness of the proposed ensemble method against traditional single-model approaches. Our findings underscore the potential of ensemble techniques in advancing the state-of-the-art in finding fake reviews using hybrid algorithms, with implications for various applications in different social media and e-platforms to find the best reviews and neglect the fake ones, eliminating puffery and bluffs.

As large language models (LLMs) expand the power of natural language processing to handle long inputs, rigorous and systematic analyses are necessary to understand their abilities and behavior. A salient application is summarization, due to its ubiquity and controversy (e.g., researchers have declared the death of summarization). In this paper, we use financial report summarization as a case study because financial reports not only are long but also use numbers and tables extensively. We propose a computational framework for characterizing multimodal long-form summarization and investigate the behavior of Claude 2.0/2.1, GPT-4/3.5, and Command. We find that GPT-3.5 and Command fail to perform this summarization task meaningfully. For Claude 2 and GPT-4, we analyze the extractiveness of the summary and identify a position bias in LLMs. This position bias disappears after shuffling the input for Claude, which suggests that Claude has the ability to recognize important information. We also conduct a comprehensive investigation on the use of numeric data in LLM-generated summaries and offer a taxonomy of numeric hallucination. We employ prompt engineering to improve GPT-4's use of numbers with limited success. Overall, our analyses highlight the strong capability of Claude 2 in handling long multimodal inputs compared to GPT-4.

The recent development of decentralised and interoperable social networks (such as the "fediverse") creates new challenges for content moderators. This is because millions of posts generated on one server can easily "spread" to another, even if the recipient server has very different moderation policies. An obvious solution would be to leverage moderation tools to automatically tag (and filter) posts that contravene moderation policies, e.g. related to toxic speech. Recent work has exploited the conversational context of a post to improve this automatic tagging, e.g. using the replies to a post to help classify if it contains toxic speech. This has shown particular potential in environments with large training sets that contain complete conversations. This, however, creates challenges in a decentralised context, as a single conversation may be fragmented across multiple servers. Thus, each server only has a partial view of an entire conversation because conversations are often federated across servers in a non-synchronized fashion. To address this, we propose a decentralised conversation-aware content moderation approach suitable for the fediverse. Our approach employs a graph deep learning model (GraphNLI) trained locally on each server. The model exploits local data to train a model that combines post and conversational information captured through random walks to detect toxicity. We evaluate our approach with data from Pleroma, a major decentralised and interoperable micro-blogging network containing 2 million conversations. Our model effectively detects toxicity on larger instances, exclusively trained using their local post information (0.8837 macro-F1). Our approach has considerable scope to improve moderation in decentralised and interoperable social networks such as Pleroma or Mastodon., Comment: Accepted at International AAAI Conference on Web and Social Media (ICWSM) 2024. Please cite accordingly!

A state-feedback watermarking signal design for the detection of replay attacks in linear systems is proposed. The control input is augmented with a random time-delayed term of the system state estimate, in order to secure the system against attacks of replay type. We outline the basic analysis of the closed-loop response of the state-feedback watermarking in a LQG controlled system. Our theoretical results are applied on a temperature process control example. While the proposed secure control scheme requires very involved analysis, it, nevertheless, holds promise of being superior to conventional, feed-forward, watermarking schemes, in both its ability to detect attacks as well as the secured system performance., Comment: 6 pages, 3 figures, manuscript to appear in the 2024 American Control Conference

The electrical double layer (EDL) at aqueous solution-metal oxide interfaces critically affects many fundamental processes in electrochemistry, geology and biology, yet understanding its microscopic structure is challenging for both theory and experiments. Here we employ ab initio-based machine learning potentials including long-range electrostatics in large-scale atomistic simulations of the EDL at the TiO2-electrolyte interface. Our simulations provide a molecular-scale picture of the EDL that demonstrates the limitations of standard mean-field models. We further develop a method to accurately calculate the electrostatic potential drop at the interface. The computed capacitance originating from the adsorbed charges and the potential drop agrees with experiments, supporting the reliability of our description of the EDL. The larger interfacial capacitance of basic relative to acidic solutions originates from the higher affinity of the cations for the oxide surface and gives rise to distinct charging mechanisms on negative and positive surfaces.

This paper considers the infinite horizon optimal control problem for nonlinear systems. Under the condition of nonlinear controllability of the system to any terminal set containing the origin and forward invariance of the terminal set, we establish a regularized solution approach consisting of a ``finite free final time" optimal transfer problem to the terminal set which renders the set globally asymptotically stable. Further, we show that the approximations converge to the optimal infinite horizon cost as the size of the terminal set decreases to zero. We also perform the analysis for the discounted problem and show that the terminal set is asymptotically stable only for a subset of the state space and not globally. The theory is empirically evaluated on various nonholonomic robotic systems to show that the cost of our approximate problem converges and the transfer time into the terminal set is dependent on the initial state of the system, necessitating the free final time formulation.

Let $\Gamma$ be a finite set of Jordan curves in the plane. For any curve $\gamma \in \Gamma$, we denote the bounded region enclosed by $\gamma$ as $\tilde{\gamma}$. We say that $\Gamma$ is a non-piercing family if for any two curves $\alpha , \beta \in \Gamma$, $\tilde{\alpha} \setminus \tilde{\beta}$ is a connected region. A non-piercing family of curves generalizes a family of $2$-intersecting curves in which each pair of curves intersect in at most two points. Snoeyink and Hershberger (``Sweeping Arrangements of Curves'', SoCG '89) proved that if we are given a family $\mathcal{C}$ of $2$-intersecting curves and a fixed curve $C\in\mathcal{C}$, then the arrangement can be \emph{swept} by $C$, i.e., $C$ can be continuously shrunk to any point $p \in \tilde{C}$ in such a way that the we have a family of $2$-intersecting curves throughout the process. In this paper, we generalize the result of Snoeyink and Hershberger to the setting of non-piercing curves. We show that given an arrangement of non-piercing curves $\Gamma$, and a fixed curve $\gamma\in \Gamma$, the arrangement can be swept by $\gamma$ so that the arrangement remains non-piercing throughout the process. We also give a shorter and simpler proof of the result of Snoeyink and Hershberger and cite applications of their result, where our result leads to a generalization.

We study offline multitask representation learning in reinforcement learning (RL), where a learner is provided with an offline dataset from different tasks that share a common representation and is asked to learn the shared representation. We theoretically investigate offline multitask low-rank RL, and propose a new algorithm called MORL for offline multitask representation learning. Furthermore, we examine downstream RL in reward-free, offline and online scenarios, where a new task is introduced to the agent that shares the same representation as the upstream offline tasks. Our theoretical results demonstrate the benefits of using the learned representation from the upstream offline task instead of directly learning the representation of the low-rank model.

We introduce a novel approach to make the tracking error of a class of nonlinear systems differentially private in addition to guaranteeing the tracking error performance. We use funnel control to make the tracking error evolve within a performance funnel that is pre-specified by the user. We make the performance funnel differentially private by adding a bounded continuous noise generated from an Ornstein-Uhlenbeck-type process. Since the funnel controller is a function of the performance funnel, the noise adds randomized perturbation to the control input. We show that, as a consequence of the differential privacy of the performance funnel, the tracking error is also differentially private. As a result, the tracking error is bounded by the noisy funnel boundary while maintaining privacy. We show a simulation result to demonstrate the framework.

This paper introduces a novel approach to concurrently design dynamic controllers and correlated differential privacy noise in dynamic control systems. An increase in privacy noise increases the system's privacy but adversely affects the system's performance. Our approach optimizes the noise distribution while shaping closed-loop system dynamics such that the privacy noise has the least impact on system performance and the most effect on system privacy. We further add privacy noise to both control input and system output to privatize the system's state for an adversary with access to both communication channels and direct output measurements. The study also suggests tailored privacy bounds for different states, providing a comprehensive framework for jointly optimizing system performance and privacy in the context of differential privacy.

Proximity coupling of ferromagnetic insulator EuS to the topological insulator Bi$_2$Se$_3$ has been proposed to break time-reversal symmetry near the surface of Bi$_2$Se$_3$, introducing an energy gap or a tilt in the surface Dirac cone. As an inverse proximity effect, strong spin-orbit coupling available in the topological surface states can enhance the Curie temperature of ferromagnetism in EuS largely beyond its bulk value, and also generate a magnetic anisotropy. This can result in a canting of the magnetic moment of Eu ions in a plane perpendicular to the interface. Here, we investigate theoretically electronic transport properties arising from the Bi$_2$Se$_3$/EuS interfaces in the planar Hall geometry. Our analysis, based on a realistic model Hamiltonian and a semi-classical formalism for the Boltzmann transport equation, reveals distinct intriguing features of anisotropic planar Hall conductivity, depending on different scenarios for the canting of the Eu moments: fixed Eu moment canting, and freely-orientable Eu moment in response to the external in-plane magnetic field. The anisotropy in the planar Hall conductivity arises from the asymmetric Berry curvature of the gapped topological surface states. We also explore topological Hall effect of the Dirac surface states, coupled to a skyrmion crystal which can emerge in the EuS due to the interplay of ferromagnetic Heisenberg exchange, interfacial Dzyaloshinskii-Moriya interaction, and perpendicular alignment of the Eu moment. Our study provides new impetus for probing complex interplay between magnetic exchange interactions and topological surface states via anisotropic planar Hall effects., Comment: 13 pages, 7 figures

We study the limits and capability of public-data assisted differentially private (PA-DP) algorithms. Specifically, we focus on the problem of stochastic convex optimization (SCO) with either labeled or unlabeled public data. For complete/labeled public data, we show that any $(\epsilon,\delta)$-PA-DP has excess risk $\tilde{\Omega}\big(\min\big\{\frac{1}{\sqrt{n_{\text{pub}}}},\frac{1}{\sqrt{n}}+\frac{\sqrt{d}}{n\epsilon} \big\} \big)$, where $d$ is the dimension, ${n_{\text{pub}}}$ is the number of public samples, ${n_{\text{priv}}}$ is the number of private samples, and $n={n_{\text{pub}}}+{n_{\text{priv}}}$. These lower bounds are established via our new lower bounds for PA-DP mean estimation, which are of a similar form. Up to constant factors, these lower bounds show that the simple strategy of either treating all data as private or discarding the private data, is optimal. We also study PA-DP supervised learning with \textit{unlabeled} public samples. In contrast to our previous result, we here show novel methods for leveraging public data in private supervised learning. For generalized linear models (GLM) with unlabeled public data, we show an efficient algorithm which, given $\tilde{O}({n_{\text{priv}}}\epsilon)$ unlabeled public samples, achieves the dimension independent rate $\tilde{O}\big(\frac{1}{\sqrt{{n_{\text{priv}}}}} + \frac{1}{\sqrt{{n_{\text{priv}}}\epsilon}}\big)$. We develop new lower bounds for this setting which shows that this rate cannot be improved with more public samples, and any fewer public samples leads to a worse rate. Finally, we provide extensions of this result to general hypothesis classes with finite fat-shattering dimension with applications to neural networks and non-Euclidean geometries.

This paper introduces a new stochastic optimization method based on the regularized Fisher information matrix (FIM), named SOFIM, which can efficiently utilize the FIM to approximate the Hessian matrix for finding Newton's gradient update in large-scale stochastic optimization of machine learning models. It can be viewed as a variant of natural gradient descent, where the challenge of storing and calculating the full FIM is addressed through making use of the regularized FIM and directly finding the gradient update direction via Sherman-Morrison matrix inversion. Additionally, like the popular Adam method, SOFIM uses the first moment of the gradient to address the issue of non-stationary objectives across mini-batches due to heterogeneous data. The utilization of the regularized FIM and Sherman-Morrison matrix inversion leads to the improved convergence rate with the same space and time complexities as stochastic gradient descent (SGD) with momentum. The extensive experiments on training deep learning models using several benchmark image classification datasets demonstrate that the proposed SOFIM outperforms SGD with momentum and several state-of-the-art Newton optimization methods in term of the convergence speed for achieving the pre-specified objectives of training and test losses as well as test accuracy.

Recognizing speaking in humans is a central task towards understanding social interactions. Ideally, speaking would be detected from individual voice recordings, as done previously for meeting scenarios. However, individual voice recordings are hard to obtain in the wild, especially in crowded mingling scenarios due to cost, logistics, and privacy concerns. As an alternative, machine learning models trained on video and wearable sensor data make it possible to recognize speech by detecting its related gestures in an unobtrusive, privacy-preserving way. These models themselves should ideally be trained using labels obtained from the speech signal. However, existing mingling datasets do not contain high quality audio recordings. Instead, speaking status annotations have often been inferred by human annotators from video, without validation of this approach against audio-based ground truth. In this paper we revisit no-audio speaking status estimation by presenting the first publicly available multimodal dataset with high-quality individual speech recordings of 33 subjects in a professional networking event. We present three baselines for no-audio speaking status segmentation: a) from video, b) from body acceleration (chest-worn accelerometer), c) from body pose tracks. In all cases we predict a 20Hz binary speaking status signal extracted from the audio, a time resolution not available in previous datasets. In addition to providing the signals and ground truth necessary to evaluate a wide range of speaking status detection methods, the availability of audio in REWIND makes it suitable for cross-modality studies not feasible with previous mingling datasets. Finally, our flexible data consent setup creates new challenges for multimodal systems under missing modalities.