Noninterference offers a rigorous end-to-end guarantee for secure propagation of information. However, real-world systems almost always involve security requirements that change during program execution, making noninterference inapplicable. Prior wor
ks alleviate the limitation to some extent, but even for a veteran in information flow security, understanding the subtleties in the syntax and semantics of each policy is challenging, largely due to very different policy specification languages, and more fundamentally, semantic requirements of each policy. We take a top-down approach and present a novel information flow policy, called Dynamic Release, which allows information flow restrictions to downgrade and upgrade in arbitrary ways. Dynamic Release is formalized on a novel framework that, for the first time, allows us to compare and contrast various dynamic policies in the literature. We show that Dynamic Release generalizes declassification, erasure, delegation and revocation. Moreover, it is the only dynamic policy that is both applicable and correct on a benchmark of tests with dynamic policy.
Domain boundaries in ferroelectric materials exhibit rich and diverse physical properties distinct from their parent materials and have been proposed for novel applications in nanoelectronics and quantum information technology. Due to their complexit
y and diversity, the internal atomic and electronic structure of domain boundaries that governs the electronic properties as well as the kinetics of domain switching remains far from being elucidated. By using scanning tunneling microscopy and spectroscopy (STM/S) combined with density functional theory (DFT) calculations, we directly visualize the atomic structure of domain boundaries in two-dimensional (2D) ferroelectric beta In2Se3 down to the monolayer limit and reveal a double-barrier energy potential of the 60{deg} tail to tail domain boundaries for the first time. We further controllably manipulate the domain boundaries with atomic precision by STM and show that the movements of domain boundaries can be driven by the electric field from an STM tip and proceed by the collective shifting of atoms at the domain boundaries. The results will deepen our understanding of domain boundaries in 2D ferroelectric materials and stimulate innovative applications of these materials.
Quantifying the heterogeneity is an important issue in meta-analysis, and among the existing measures, the $I^2$ statistic is the most commonly used measure in the literature. In this paper, we show that the $I^2$ statistic was, in fact, defined as p
roblematic or even completely wrong from the very beginning. To confirm this statement, we first present a motivating example to show that the $I^2$ statistic is heavily dependent on the study sample sizes, and consequently it may yield contradictory results for the amount of heterogeneity. Moreover, by drawing a connection between ANOVA and meta-analysis, the $I^2$ statistic is shown to have, mistakenly, applied the sampling errors of the estimators rather than the variances of the study populations. Inspired by this, we introduce an Intrinsic measure for Quantifying the heterogeneity in meta-analysis, and meanwhile study its statistical properties to clarify why it is superior to the existing measures. We further propose an optimal estimator, referred to as the IQ statistic, for the new measure of heterogeneity that can be readily applied in meta-analysis. Simulations and real data analysis demonstrate that the IQ statistic provides a nearly unbiased estimate of the true heterogeneity and it is also independent of the study sample sizes.
Modern neural networks can assign high confidence to inputs drawn from outside the training distribution, posing threats to models in real-world deployments. While much research attention has been placed on designing new out-of-distribution (OOD) det
ection methods, the precise definition of OOD is often left in vagueness and falls short of the desired notion of OOD in reality. In this paper, we present a new formalization and model the data shifts by taking into account both the invariant and environmental (spurious) features. Under such formalization, we systematically investigate how spurious correlation in the training set impacts OOD detection. Our results suggest that the detection performance is severely worsened when the correlation between spurious features and labels is increased in the training set. We further show insights on detection methods that are more effective in reducing the impact of spurious correlation and provide theoretical analysis on why reliance on environmental features leads to high OOD detection error. Our work aims to facilitate a better understanding of OOD samples and their formalization, as well as the exploration of methods that enhance OOD detection.
The dynamics of spherical laser-induced cavitation bubbles in water is investigated by plasma photography, time-resolved shadowgraphs, and single-shot probe beam scattering enabling to portray the transition from initial nonlinear to late linear osci
llations. The frequency of late oscillations yields the bubbles gas content. Simulations with the Gilmore model using plasma size as input and oscillation times as fit parameter provide insights into experimentally not accessible bubble parameters and shock wave emission. The model is extended by a term covering the initial shock-driven acceleration of the bubble wall, an automated method determining shock front position and pressure decay, and an energy balance for the partitioning of absorbed laser energy into vaporization, bubble and shock wave energy, and dissipation through viscosity and condensation. These tools are used for analysing a scattering signal covering 102 oscillation cycles. The bubble was produced by a plasma with 1550 K average temperature and had 36 $mu$m maximum radius. Predicted bubble wall velocities during expansion agree well with experimental data. Upon first collapse, most energy was stored in the compressed liquid around the bubble and then radiated away acoustically. The collapsed bubble contained more vapour than gas, and its pressure was 13.5 GPa. The pressure of the rebound shock wave initially decayed $propto r^{-1.8}$, and energy dissipation at the shock front heated liquid near the bubble wall above the superheat limit. The shock-induced temperature rise reduces damping during late bubble oscillations. Bubble dynamics changes significantly for small bubbles with less than 10 $mu$m radius.
Flexible load at the demand-side has been regarded as an effective measure to cope with volatile distributed renewable generations. To unlock the demand-side flexibility, this paper proposes a peer-to-peer energy sharing mechanism that facilitates en
ergy exchange among users while preserving privacy. We prove the existence and partial uniqueness of the energy sharing market equilibrium and provide a centralized optimization to obtain the equilibrium. The centralized optimization is further linearized by a convex combination approach, turning into a multi-parametric linear program (MP-LP) with renewable output deviations being the parameters. The flexibility requirement of individual users is calculated based on this MP-LP. To be specific, an adaptive vertex generation algorithm is established to construct a piecewise linear estimator of the optimal total cost subject to a given error tolerance. Critical regions and optimal strategies are retrieved from the obtained approximate cost function to evaluate the flexibility requirement. The proposed algorithm does not rely on the exact characterization of optimal basis invariant sets and thus is not influenced by model degeneracy, a common difficulty faced by existing approaches. Case studies validate the theoretical results and show that the proposed method is scalable.
Federated learning is widely used to learn intelligent models from decentralized data. In federated learning, clients need to communicate their local model updates in each iteration of model learning. However, model updates are large in size if the m
odel contains numerous parameters, and there usually needs many rounds of communication until model converges. Thus, the communication cost in federated learning can be quite heavy. In this paper, we propose a communication efficient federated learning method based on knowledge distillation. Instead of directly communicating the large models between clients and server, we propose an adaptive mutual distillation framework to reciprocally learn a student and a teacher model on each client, where only the student model is shared by different clients and updated collaboratively to reduce the communication cost. Both the teacher and student on each client are learned on its local data and the knowledge distilled from each other, where their distillation intensities are controlled by their prediction quality. To further reduce the communication cost, we propose a dynamic gradient approximation method based on singular value decomposition to approximate the exchanged gradients with dynamic precision. Extensive experiments on benchmark datasets in different tasks show that our approach can effectively reduce the communication cost and achieve competitive results.
Deep learning models are modern tools for spatio-temporal graph (STG) forecasting. Despite their effectiveness, they require large-scale datasets to achieve better performance and are vulnerable to noise perturbation. To alleviate these limitations,
an intuitive idea is to use the popular data augmentation and contrastive learning techniques. However, existing graph contrastive learning methods cannot be directly applied to STG forecasting due to three reasons. First, we empirically discover that the forecasting task is unable to benefit from the pretrained representations derived from contrastive learning. Second, data augmentations that are used for defeating noise are less explored for STG data. Third, the semantic similarity of samples has been overlooked. In this paper, we propose a Spatio-Temporal Graph Contrastive Learning framework (STGCL) to tackle these issues. Specifically, we improve the performance by integrating the forecasting loss with an auxiliary contrastive loss rather than using a pretrained paradigm. We elaborate on four types of data augmentations, which disturb data in terms of graph structure, time domain, and frequency domain. We also extend the classic contrastive loss through a rule-based strategy that filters out the most semantically similar negatives. Our framework is evaluated across three real-world datasets and four state-of-the-art models. The consistent improvements demonstrate that STGCL can be used as an off-the-shelf plug-in for existing deep models.
Attributed to the ever-increasing large image datasets, Convolutional Neural Networks (CNNs) have become popular for vision-based tasks. It is generally admirable to have larger-sized datasets for higher network training accuracies. However, the impa
ct of dataset quality has not to be involved. It is reasonable to assume the near-duplicate images exist in the datasets. For instance, the Street View House Numbers (SVHN) dataset having cropped house plate digits from 0 to 9 are likely to have repetitive digits from the same/similar house plates. Redundant images may take up a certain portion of the dataset without consciousness. While contributing little to no accuracy improvement for the CNNs training, these duplicated images unnecessarily pose extra resource and computation consumption. To this end, this paper proposes a framework to assess the impact of the near-duplicate images on CNN training performance, called CE-Dedup. Specifically, CE-Dedup associates a hashing-based image deduplication approach with downstream CNNs-based image classification tasks. CE-Dedup balances the tradeoff between a large deduplication ratio and a stable accuracy by adjusting the deduplication threshold. The effectiveness of CE-Dedup is validated through extensive experiments on well-known CNN benchmarks. On one hand, while maintaining the same validation accuracy, CE-Dedup can reduce the dataset size by 23%. On the other hand, when allowing a small validation accuracy drop (by 5%), CE-Dedup can trim the dataset size by 75%.
Neural networks (NNs) are widely used for classification tasks for their remarkable performance. However, the robustness and accuracy of NNs heavily depend on the training data. In many applications, massive training data is usually not available. To
address the challenge, this paper proposes an iterative adversarial data augmentation (IADA) framework to learn neural network models from an insufficient amount of training data. The method uses formal verification to identify the most confusing input samples, and leverages human guidance to safely and iteratively augment the training data with these samples. The proposed framework is applied to an artificial 2D dataset, the MNIST dataset, and a human motion dataset. By applying IADA to fully-connected NN classifiers, we show that our training method can improve the robustness and accuracy of the learned model. By comparing to regular supervised training, on the MNIST dataset, the average perturbation bound improved 107.4%. The classification accuracy improved 1.77%, 3.76%, 10.85% on the 2D dataset, the MNIST dataset, and the human motion dataset respectively.
