Search

Although achieving promising performance, recent analyses show that current generative large language models (LLMs) may still capture dataset biases and utilize them for generation, leading to poor generalizability and harmfulness of LLMs. However, due to the diversity of dataset biases and the over-optimization problem, previous prior-knowledge-based debiasing methods and fine-tuning-based debiasing methods may not be suitable for current LLMs. To address this issue, we explore combining active learning with the causal mechanisms and propose a casual-guided active learning (CAL) framework, which utilizes LLMs itself to automatically and autonomously identify informative biased samples and induce the bias patterns. Then a cost-effective and efficient in-context learning based method is employed to prevent LLMs from utilizing dataset biases during generation. Experimental results show that CAL can effectively recognize typical biased instances and induce various bias patterns for debiasing LLMs., Comment: Accepted as ACL 2024 main conference & Rewared as Outstanding Paper

The Rezzolla-Zhidenko (RZ) and Konoplya-Rezzolla-Zhidenko (KRZ) frameworks provide an efficient approach to characterize agnostically spherically symmetric or stationary black-hole spacetimes in arbitrary metric theories. In their original construction, these metrics were defined only in the spacetime region outside of the event horizon, where they can reproduce any black-hole metric with percent precision and a few parameters only. At the same time, numerical simulations of accreting black holes often require metric functions that are regular across the horizon, so that the inner boundary of the computational domain can be placed in a region that is causally disconnected from the exterior. We present a novel formulation of the RZ/KRZ parametrized metrics in coordinate systems that are regular at the horizon and defined everywhere in the interior. We compare the horizon-penetrating form of the KRZ and RZ metrics with the corresponding forms of the Kerr metric in Kerr-Schild coordinates and of the Schwarzschild metric in Eddington-Finkelstein coordinates, remarking the similarities and differences. We expect the horizon-penetrating formulations of the RZ/KRZ metrics to represent new tools to study via simulations the physical processes that occur near the horizon of an arbitrary black hole., Comment: 15 pages, submitted to PRD, matching the accepted version

The evolution of excitons from 2D to 3D is of great importance in photo-physics, yet the layer-dependent exciton polarizability has not been investigated in 2D semiconductors. Here, we determine the exciton polarizabilities for 3- to 11-layer black phosphorus-a direct bandgap semiconductor regardless of the thickness-through frequency-resolved photocurrent measurements on dual-gate devices and unveil the carrier screening effect in relatively thicker samples. By taking advantage of the broadband photocurrent spectra, we are also able to reveal the exciton response for higher-index subbands under the gate electrical field. Surprisingly, dark excitons are brightened with intensity even stronger than the allowed transitions above certain electrical field. Our study not only sheds light on the exciton evolution with sample thickness, but also paves a way for optoelectronic applications of few-layer BP in modulators, tunable photodetectors, emitters and lasers., Comment: 30 pages, 4 figures

Contrastive learning on graphs aims at extracting distinguishable high-level representations of nodes. In this paper, we theoretically illustrate that the entropy of a dataset can be approximated by maximizing the lower bound of the mutual information across different views of a graph, \ie, entropy is estimated by a neural network. Based on this finding, we propose a simple yet effective subset sampling strategy to contrast pairwise representations between views of a dataset. In particular, we randomly sample nodes and edges from a given graph to build the input subset for a view. Two views are fed into a parameter-shared Siamese network to extract the high-dimensional embeddings and estimate the information entropy of the entire graph. For the learning process, we propose to optimize the network using two objectives, simultaneously. Concretely, the input of the contrastive loss function consists of positive and negative pairs. Our selection strategy of pairs is different from previous works and we present a novel strategy to enhance the representation ability of the graph encoder by selecting nodes based on cross-view similarities. We enrich the diversity of the positive and negative pairs by selecting highly similar samples and totally different data with the guidance of cross-view similarity scores, respectively. We also introduce a cross-view consistency constraint on the representations generated from the different views. This objective guarantees the learned representations are consistent across views from the perspective of the entire graph. We conduct extensive experiments on seven graph benchmarks, and the proposed approach achieves competitive performance compared to the current state-of-the-art methods. The source code will be publicly released once this paper is accepted., Comment: ACM MM 2023 accepted

Photoionization of the rubidium (Rb) atoms cooled in a magneto-optical trap, characterized by the coexistence of the ground 5$S_{1/2}$ and the excited 5$P_{3/2}$ states, is investigated experimentally and theoretically with the 400 nm femtosecond laser pulses at intensities of $I=3\times10^9$ W/cm$^2$ - $4.5\times10^{12}$ W/cm$^2$. Recoil-ion momentum distribution (RIMD) of Rb$^+$ exhibits rich ring-like structures and their energies correspond to one-photon ionization of the 5$P_{3/2}$ state, two-photon and three-photon ionizations of the 5$S_{1/2}$ state, respectively. With the increasing of $I$, we find that experimental signals near zero-momentum (NZM) in RIMDs resulted from the 5$P_{3/2}$ state enhance dramatically and its peaked Rb$^+$ momenta dwindle obviously while that from the 5$S_{1/2}$ state is maintained. Meanwhile, the ion-yield ratio of the 5$S_{1/2}$ over the 5$P_{3/2}$ states varies from $I$ to $I^{1.5}$ as $I$ increases. These features indicate a transition from perturbative ionization to strong-perturbative ionization for the 5$P_{3/2}$ state. Numerical simulations by solving the time-dependent Schr\"odinger equation (TDSE) can qualitatively explain the measurements of RIMD, photoion angular distributions, as well as ion-yield ratio. However, some discrepancies still exist, especially for the NZM dip, which could stem from the electron-electron correlation that is neglected in the present TDSE simulations since we have adopted the single-active-electron approximation.

Although many graph-based clustering methods attempt to model the stationary diffusion state in their objectives, their performance limits to using a predefined graph. We argue that the estimation of the stationary diffusion state can be achieved by gradient descent over neural networks. We specifically design the Stationary Diffusion State Neural Estimation (SDSNE) to exploit multiview structural graph information for co-supervised learning. We explore how to design a graph neural network specially for unsupervised multiview learning and integrate multiple graphs into a unified consensus graph by a shared self-attentional module. The view-shared self-attentional module utilizes the graph structure to learn a view-consistent global graph. Meanwhile, instead of using auto-encoder in most unsupervised learning graph neural networks, SDSNE uses a co-supervised strategy with structure information to supervise the model learning. The co-supervised strategy as the loss function guides SDSNE in achieving the stationary state. With the help of the loss and the self-attentional module, we learn to obtain a graph in which nodes in each connected component fully connect by the same weight. Experiments on several multiview datasets demonstrate effectiveness of SDSNE in terms of six clustering evaluation metrics., Comment: AAAI 2022

Driven by the high profit, Portable Executable (PE) malware has been consistently evolving in terms of both volume and sophistication. PE malware family classification has gained great attention and a large number of approaches have been proposed. With the rapid development of machine learning techniques and the exciting results they achieved on various tasks, machine learning algorithms have also gained popularity in the PE malware family classification task. Three mainstream approaches that use learning based algorithms, as categorized by the input format the methods take, are image-based, binary-based and disassembly-based approaches. Although a large number of approaches are published, there is no consistent comparisons on those approaches, especially from the practical industry adoption perspective. Moreover, there is no comparison in the scenario of concept drift, which is a fact for the malware classification task due to the fast evolving nature of malware. In this work, we conduct a thorough empirical study on learning-based PE malware classification approaches on 4 different datasets and consistent experiment settings. Based on the experiment results and an interview with our industry partners, we find that (1) there is no individual class of methods that significantly outperforms the others; (2) All classes of methods show performance degradation on concept drift (by an average F1-score of 32.23%); and (3) the prediction time and high memory consumption hinder existing approaches from being adopted for industry usage., Comment: 12 pages, 3 figures, conference

Dwarf spheroidal galaxies are dark matter dominated systems, and as such, ideal for indirect dark matter searches. If dark matter decays into high-energy photons in the dwarf galaxies, they will be a good target for current and future generations of X-ray and gamma-ray telescopes. By adopting the latest estimates of density profiles of dwarf galaxies in the Milky Way, we revise the estimates dark matter decay rates in dwarf galaxies; our results are more robust, but weaker than previous estimates. Applying these results, we study the detectability of dark matter decays with X-ray and very-high-energy gamma-ray telescopes, such as eROSITA, XRISM, Athena, HAWC, and CTA. Our projection shows that all of these X-ray telescopes will be able to critically assess the claim of the 7 keV sterile neutrino decays from stacked galaxy clusters and nearby galaxies. For TeV decaying dark matter, we can constrain its lifetime to be longer than $\sim$10$^{27}$-10$^{28}$ s. We also make projections for future dwarf galaxies that would be newly discovered with the Vera Rubin Observatory Legacy Survey of Space and Time, which will further improve the expected sensitivity to dark matter decays both in the keV and PeV mass ranges., Comment: 18 pages, 15 figures; comments welcome

The YOLOv3 target detection algorithm is widely used in industry due to its high speed and high accuracy, but it has some limitations, such as the accuracy degradation of unbalanced datasets. The YOLOv3 target detection algorithm is based on a Gaussian fuzzy data augmentation approach to pre-process the data set and improve the YOLOv3 target detection algorithm. Through the efficient pre-processing, the confidence level of YOLOv3 is generally improved by 0.01-0.02 without changing the recognition speed of YOLOv3, and the processed images also perform better in image localization due to effective feature fusion, which is more in line with the requirement of recognition speed and accuracy in production.

We perform high-resolution measurements of momentum distribution on Rb$^{n+}$ recoil ions up to charge state $n=4$, where laser-cooled rubidium atoms are ionized by femtosecond elliptically polarized lasers with the pulse duration of 35 fs and the intensity of 3.3$\times$10$^{15}$ W/cm$^2$ in the over-barrier ionization (OBI) regime. The momentum distributions of the recoil ions are found to exhibit multi-band structures as the ellipticity varies from the linear to circular polarizations. The origin of these band structures can be explained quantitatively by the classical OBI model and dedicated classical trajectory Monte Carlo simulations with Heisenberg potential. Specifically, with back analysis of the classical trajectories, we reveal the ionization time and the OBI geometry of the sequentially released electrons, disentangling the mechanisms behind the tilted angle of the band structures. These results indicate that the classical treatment can describe the strong-field multiple ionization processes of alkali atoms.

Employing recent developed magneto-optical trap recoil ion momentum spectroscopy (MOTRIMS) combining cold atom, strong laser pulse, and ultrafast technologies, we study momentum distributions of the multiply ionized cold rubidium (Rb) induced by the elliptically polarized laser pulses (35 fs, $1.3 \times 10^{15}$ W/cm$^2$). The complete vector momenta of Rbn+ ions up to charge state n = 4 are recorded with extremely high resolution (0.12 a.u. for Rb$^+$). Variations of characteristic multi-bands displayed in momentum distributions, as the ellipticity varies from the linear to circular polarization, are interpreted qualitatively with the classical over-barrier ionization model. Present momentum spectroscopy of cold heavy alkali atoms presents novel strong-field phenomena beyond the noble gases.

A key concept underlying the specific functionalities of metasurfaces, i.e. arrays of subwavelength nanoparticles, is the use of constituent components to shape the wavefront of the light, on-demand. Metasurfaces are versatile and novel platforms to manipulate the scattering, colour, phase or the intensity of the light. Currently, one of the typical approaches for designing a metasurface is to optimize one or two variables, among a vast number of fixed parameters, such as various materials' properties and coupling effects, as well as the geometrical parameters. Ideally, it would require a multi-dimensional space optimization through direct numerical simulations. Recently, an alternative approach became quite popular allowing to reduce the computational cost significantly based on a deep-learning-assisted method. In this paper, we utilize a deep-learning approach for obtaining high-quality factor (high-Q) resonances with desired characteristics, such as linewidth, amplitude and spectral position. We exploit such high-Q resonances for the enhanced light-matter interaction in nonlinear optical metasurfaces and optomechanical vibrations, simultaneously. We demonstrate that optimized metasurfaces lead up to 400+ folds enhancement of the third harmonic generation (THG); at the same time, they also contribute to 100+ folds enhancement in optomechanical vibrations. This approach can be further used to realize structures with unconventional scattering responses., Comment: 29 pages, 7 figures

We study photoionization of cold rubidium atoms in a strong infrared laser field using a magneto-optical trap (MOT) recoil ion momentum spectrometer. Three types of cold rubidium target are provided, operating in two-dimension (2D) MOT, 2D molasses, and 3D MOT with densities in the orders of $10^7$ atoms/cm$^3$, $10^8$ atoms/cm$^3$, and $10^9$ atoms/cm$^3$, respectively. The density profile and the temperature of 3D MOT are characterized using the absorption imaging and photoionization. The momentum distributions of Rb$^+$ created by absorption of two- or three-photon illuminate a dipole-like double-peak structure, in good agreement with the results in the strong field approximation. The yielding momentum resolution of $0.12 \pm 0.03$ a.u. is achieved in comparison with theoretical calculations, exhibiting the great prospects for the study of electron correlations in alkali metal atoms through interaction with strong laser pulses.