Search for: All records

Uncertainty Quantification (UQ) is vital for decision makers as it offers insights into the potential reliability of data and model, enabling more informed and risk-aware decision-making. Graphical models, capable of representing data with complex dependencies, are widely used across domains. Existing sampling-based UQ methods are unbiased but cannot guarantee convergence and are time-consuming on large-scale graphs. There are fast UQ methods for graphical models with closed-form solutions and convergence guarantee but with uncertainty underestimation. We propose LinUProp, a UQ method that utilizes a novel linear propagation of uncertainty to model uncertainty among related nodes additively instead of multiplicatively, to offer linear scalability, guaranteed convergence, and closed-form solutions without underestimating uncertainty. Theoretically, we decompose the expected prediction error of the graphical model and prove that the uncertainty computed by LinUProp is the generalized variance component of the decomposition. Experimentally, we demonstrate that LinUProp is consistent with the sampling-based method but with linear scalability and fast convergence. Moreover, LinUProp outperforms competitors in uncertainty-based active learning on four real-world graph datasets, achieving higher accuracy with a lower labeling budget. 

An environmentally responsible synthesis of Na_2/3[Ni_1/3Mn_2/3]O₂(NNMO) enabled by direct conversion of metallic Ni powder, releasing only O₂and H₂O as byproducts. The resulting NNMO exhibits excellent cycling performance as a Na-ion battery cathode. 

User-generated product reviews are essential for online platforms like Amazon and Yelp. However, the presence of fake reviews misleads customers. GNN is the state-of-the-art method that detects suspicious reviewers by exploiting the topologies of the graph connecting reviewers, reviews, and products. Nevertheless, the discrepancy in the detection accuracy over different groups of reviewers degrades reviewer engagement and customer trust in the review websites. Unlike the previous belief that the difference between the groups causes unfairness, we study the subgroup structures within the groups that can also cause discrepancies in treating different groups. This paper addresses the challenges of defining, approximating, and utilizing a new subgroup structure for fair spam detection. We first identify subgroup structures in the review graph that lead to discrepant accuracy in the groups. The complex dependencies over the review graph create difficulties in teasing out subgroups hidden within larger groups. We design a model that can be trained to jointly infer the hidden subgroup memberships and exploits the membership for calibrating the detection accuracy across groups. Comprehensive comparisons against baselines on three large Yelp review datasets demonstrate that the subgroup membership can be identified and exploited for group fairness. 

Accelerator-based x-ray free-electron lasers (XFELs) are the latest addition to the revolutionary tools of discovery for the 21st century. The two major components of an XFEL are an accelerator-produced electron beam and a magnetic undulator, which tend to be kilometer-scale long and expensive. A proof-of-principle demonstration of free-electron lasing at 27 nm using beams from compact laser wakefield accelerators was shown recently by using a magnetic undulator. However, scaling these concepts to x-ray wavelengths is far from straightforward as the requirements on the beam quality and jitters become much more stringent. Here, we present an ultracompact scheme to produce tens of attosecond x-ray pulses with several GW peak power utilizing a novel aspect of the FEL instability using a highly chirped, prebunched, and ultrabright tens of MeVelectron beam from a plasma-based accelerator interacting with an optical undulator. The FEL resonant relation between the prebunched period and the energy selects resonant electrons automatically from the highly chirped beam which leads to a stable generation of attosecond x-ray pulses. Furthermore, two-color attosecond pulses with subfemtosecond separation can be produced by adjusting the energy distribution of the electron beam so that multiple FEL resonances occur at different locations within the beam. Such a tunable coherent attosecond x-ray sources may open up a new area of attosecond science enabled by x-ray attosecond pump/probe techniques 

The quality of electron beams produced from plasma-based accelerators, i.e., normalized brightness and energy spread, has made transformative progress in the past several decades in both simulation and experiment. Recently, full-scale particle-in-cell (PIC) simulations have shown that electron beams with unprecedented brightness (1020–1021 A=m2=rad2) and 0.1–1 MeVenergy spread can be produced through controlled injection in a slowly expanding bubble that arises when a particle beam or laser pulse propagates in density gradient, or when a particle beam self-focuses in uniform plasma or has a superluminal flying focus. However, in previous simulations of work on self-injection triggered by an evolving laser driver in a uniform plasma, the resulting beams did not exhibit comparable brightnesses and energy spreads. Here, we demonstrate through the use of large-scale high-fidelity PIC simulations that a slowly expanding bubble driven by a laser pulse in a uniform plasma can indeed produce self-injected electron beams with similar brightness and energy spreads as for an evolving bubble driven by an electron beam driver. We consider laser spot sizes roughly equal to the matched spot sizes in a uniform plasma and find that the evolution of the bubble occurs naturally through the evolution of the laser. The effects of the electron beam quality on the choice of physical as well as numerical parameters, e.g., grid sizes and field solvers used in the PIC simulations are presented. It is found that this original and simplest injection scheme can produce electron beams with beam quality exceeding that of the more recent concepts. 

Social networks are frequently polluted by rumors, which can be detected by advanced models such as graph neural networks. However, the models are vulnerable to attacks, and discovering and understanding the vulnerabilities is critical to robust rumor detection. To discover subtle vulnerabilities, we design a attacking algorithm based on reinforcement learning to camouflage rumors against black-box detectors. We address exponentially large state spaces, high-order graph dependencies, and ranking dependencies, which are unique to the problem setting but fundamentally challenging for the state-of-the-art end-to-end approaches. We design domain-specific features that have causal effect on the reward, so that even a linear policy can arrive at powerful attacks with additional interpretability. To speed up policy optimization, we devise: (i) a credit assignment method that proportionally decomposes delayed and aggregated rewards to atomic attacking actions for enhance feature-reward associations; (ii) a time-dependent control variate to reduce prediction variance due to large state-action spaces and long attack horizon, based on reward variance analysis and a Bayesian analysis of the prediction distribution. On two real world datasets of rumor detection tasks, we demonstrate: (i) the effectiveness of the learned attacking policy on a wide spectrum of target models compared to both rule-based and end-to-end attacking approaches; (ii) the usefulness of the proposed credit assignment strategy and variance reduction components; (iii) the interpretability of the attacking policy.