Search

The HADAR experiment, which will be constructed in Tibet, China, combines the wide-angle advantages of traditional EAS array detectors with the high sensitivity advantages of focused Cherenkov detectors. Its physics objective is to observe transient sources such as gamma-ray bursts and counterparts of gravitational waves. The aim of this study is to utilize the latest AI technology to enhance the sensitivity of the HADAR experiment. We have built training datasets and models with distinctive creativity by incorporating relevant physical theories for various applications. They are able to determine the kind, energy, and direction of incident particles after careful design. We have obtained a background identification accuracy of 98.6%, a relative energy reconstruction error of 10.0%, and an angular resolution of 0.22-degrees in a test dataset at 10 TeV. These findings demonstrate the enormous potential for enhancing the precision and dependability of detector data analysis in astrophysical research. Thanks to deep learning techniques, the HADAR experiment's observational sensitivity to the Crab Nebula has surpassed that of MAGIC and H.E.S.S. at energies below 0.5 TeV and remains competitive with conventional narrow-field Cherenkov telescopes at higher energies. Additionally, our experiment offers a fresh approach to dealing with strongly connected scattered data.

Topological materials hosting metallic edges characterized by integer quantized conductivity in an insulating bulk have revolutionized our understanding of transport in matter. The topological protection of these edge states is based on symmetries and dimensionality. However, only integer-dimensional models have been classified, and the interplay of topology and fractals, which may have a non-integer dimension, remained largely unexplored. Quantum fractals have recently been engineered in metamaterials, but up to present no topological states were unveiled in fractals realized in real materials. Here, we show theoretically and experimentally that topological edge and corner modes arise in fractals formed upon depositing thin layers of bismuth on an indium antimonide substrate. Scanning tunneling microscopy reveals the appearance of (nearly) zero-energy modes at the corners of Sierpi\'nski triangles, as well as the formation of outer and inner edge modes at higher energies. Unexpectedly, a robust and sharp depleted mode appears at the outer and inner edges of the samples at negative bias voltages. The experimental findings are corroborated by theoretical calculations in the framework of a continuum muffin-tin and a lattice tight-binding model. The stability of the topological features to the introduction of a Rashba spin-orbit coupling and disorder is discussed. This work opens the perspective to novel electronics in real materials at non-integer dimensions with robust and protected topological states., Comment: Main manuscript 14 pages, supplementary material 34 pages

Large language models (LLMs) are highly adept at question answering and reasoning tasks, but when reasoning in a situational context, human expectations vary depending on the relevant cultural common ground. As languages are associated with diverse cultures, LLMs should also be culturally-diverse reasoners. In this paper, we study the ability of a wide range of state-of-the-art multilingual LLMs (mLLMs) to reason with proverbs and sayings in a conversational context. Our experiments reveal that: (1) mLLMs "know" limited proverbs and memorizing proverbs does not mean understanding them within a conversational context; (2) mLLMs struggle to reason with figurative proverbs and sayings, and when asked to select the wrong answer (instead of asking it to select the correct answer); and (3) there is a "culture gap" in mLLMs when reasoning about proverbs and sayings translated from other languages. We construct and release our evaluation dataset MAPS (MulticultrAl Proverbs and Sayings) for proverb understanding with conversational context for six different languages., Comment: NAACL

Dataset distillation methods have demonstrated remarkable performance for neural networks trained with very limited training data. However, a significant challenge arises in the form of architecture overfitting: the distilled training data synthesized by a specific network architecture (i.e., training network) generates poor performance when trained by other network architectures (i.e., test networks). This paper addresses this issue and proposes a series of approaches in both architecture designs and training schemes which can be adopted together to boost the generalization performance across different network architectures on the distilled training data. We conduct extensive experiments to demonstrate the effectiveness and generality of our methods. Particularly, across various scenarios involving different sizes of distilled data, our approaches achieve comparable or superior performance to existing methods when training on the distilled data using networks with larger capacities.

Magnetic insulators such as yttrium iron garnets (YIGs) are of paramount importance for spin-wave or magnonic devices as their ultralow damping enables ultralow power dissipation that is free of Joule heating, exotic magnon quantum state, and coherent coupling to other wave excitations. Magnetic insulator heterostructures bestow superior structural and magnetic properties and house immense design space thanks to the strong and engineerable exchange interaction between individual layers. To fully unleash their potential, realizing low damping and strong exchange coupling simultaneously is critical, which often requires high quality interface. Here, we show that such a demand is realized in an all-insulator thulium iron garnet (TmIG)/YIG bilayer system. The ultralow dissipation rates in both YIG and TmIG, along with their significant spin-spin interaction at the interface, enable strong and coherent magnon-magnon coupling with a benchmarking cooperativity value larger than the conventional ferromagnetic metal-based heterostructures. The coupling strength can be tuned by varying the magnetic insulator layer thickness and magnon modes, which is consistent with analytical calculations and micromagnetic simulations. Our results demonstrate TmIG/YIG as a novel platform for investigating hybrid magnonic phenomena and open opportunities in magnon devices comprising all-insulator heterostructures., Comment: 45 pages, 18 figures, and 2 tables

For years, researchers investigated the applications of deep learning in forecasting financial time series. However, they continued to rely on the conventional econometric approach for model training that optimizes the deep learning models on individual assets. This study highlights the importance of global training, where the deep learning model is optimized across a wide spectrum of stocks. Focusing on stock volatility forecasting as an exemplar, we show that global training is not only beneficial but also necessary for deep learning-based financial time series forecasting. We further demonstrate that, given a sufficient amount of training data, a globally trained deep learning model is capable of delivering accurate zero-shot forecasts for any stocks., Comment: 25 pages, 5 figures

Modern deep learning heavily relies on large labeled datasets, which often comse with high costs in terms of both manual labeling and computational resources. To mitigate these challenges, researchers have explored the use of informative subset selection techniques, including coreset selection and active learning. Specifically, coreset selection involves sampling data with both input ($\bx$) and output ($\by$), active learning focuses solely on the input data ($\bx$). In this study, we present a theoretically optimal solution for addressing both coreset selection and active learning within the context of linear softmax regression. Our proposed method, COPS (unCertainty based OPtimal Sub-sampling), is designed to minimize the expected loss of a model trained on subsampled data. Unlike existing approaches that rely on explicit calculations of the inverse covariance matrix, which are not easily applicable to deep learning scenarios, COPS leverages the model's logits to estimate the sampling ratio. This sampling ratio is closely associated with model uncertainty and can be effectively applied to deep learning tasks. Furthermore, we address the challenge of model sensitivity to misspecification by incorporating a down-weighting approach for low-density samples, drawing inspiration from previous works. To assess the effectiveness of our proposed method, we conducted extensive empirical experiments using deep neural networks on benchmark datasets. The results consistently showcase the superior performance of COPS compared to baseline methods, reaffirming its efficacy.

Learning from bounding-boxes annotations has shown great potential in weakly-supervised 3D point cloud instance segmentation. However, we observed that existing methods would suffer severe performance degradation with perturbed bounding box annotations. To tackle this issue, we propose a complementary image prompt-induced weakly-supervised point cloud instance segmentation (CIP-WPIS) method. CIP-WPIS leverages pretrained knowledge embedded in the 2D foundation model SAM and 3D geometric prior to achieve accurate point-wise instance labels from the bounding box annotations. Specifically, CP-WPIS first selects image views in which 3D candidate points of an instance are fully visible. Then, we generate complementary background and foreground prompts from projections to obtain SAM 2D instance mask predictions. According to these, we assign the confidence values to points indicating the likelihood of points belonging to the instance. Furthermore, we utilize 3D geometric homogeneity provided by superpoints to decide the final instance label assignments. In this fashion, we achieve high-quality 3D point-wise instance labels. Extensive experiments on both Scannet-v2 and S3DIS benchmarks demonstrate that our method is robust against noisy 3D bounding-box annotations and achieves state-of-the-art performance.

Convolutional neural networks excel in histopathological image classification, yet their pixel-level focus hampers explainability. Conversely, emerging graph convolutional networks spotlight cell-level features and medical implications. However, limited by their shallowness and suboptimal use of high-dimensional pixel data, GCNs underperform in multi-class histopathological image classification. To make full use of pixel-level and cell-level features dynamically, we propose an asymmetric co-training framework combining a deep graph convolutional network and a convolutional neural network for multi-class histopathological image classification. To improve the explainability of the entire framework by embedding morphological and topological distribution of cells, we build a 14-layer deep graph convolutional network to handle cell graph data. For the further utilization and dynamic interactions between pixel-level and cell-level information, we also design a co-training strategy to integrate the two asymmetric branches. Notably, we collect a private clinically acquired dataset termed LUAD7C, including seven subtypes of lung adenocarcinoma, which is rare and more challenging. We evaluated our approach on the private LUAD7C and public colorectal cancer datasets, showcasing its superior performance, explainability, and generalizability in multi-class histopathological image classification.

Given an audio-visual pair, audio-visual segmentation (AVS) aims to locate sounding sources by predicting pixel-wise maps. Previous methods assume that each sound component in an audio signal always has a visual counterpart in the image. However, this assumption overlooks that off-screen sounds and background noise often contaminate the audio recordings in real-world scenarios. They impose significant challenges on building a consistent semantic mapping between audio and visual signals for AVS models and thus impede precise sound localization. In this work, we propose a two-stage bootstrapping audio-visual segmentation framework by incorporating multi-modal foundation knowledge. In a nutshell, our BAVS is designed to eliminate the interference of background noise or off-screen sounds in segmentation by establishing the audio-visual correspondences in an explicit manner. In the first stage, we employ a segmentation model to localize potential sounding objects from visual data without being affected by contaminated audio signals. Meanwhile, we also utilize a foundation audio classification model to discern audio semantics. Considering the audio tags provided by the audio foundation model are noisy, associating object masks with audio tags is not trivial. Thus, in the second stage, we develop an audio-visual semantic integration strategy (AVIS) to localize the authentic-sounding objects. Here, we construct an audio-visual tree based on the hierarchical correspondence between sounds and object categories. We then examine the label concurrency between the localized objects and classified audio tags by tracing the audio-visual tree. With AVIS, we can effectively segment real-sounding objects. Extensive experiments demonstrate the superiority of our method on AVS datasets, particularly in scenarios involving background noise. Our project website is https://yenanliu.github.io/AVSS.github.io/.

Ferrimagnets have received renewed attention as a promising platform for spintronic applications. Of particular interest is the Mn4N from the ${\epsilon}$-phase of the manganese nitride as an emergent rare-earth-free spintronic material due to its perpendicular magnetic anisotropy, small saturation magnetization, high thermal stability, and large domain wall velocity. We have achieved high-quality (001)-ordered Mn$_{4}$N thin film by sputtering Mn onto ${\eta}$-phase Mn$_{3}$N$_{2}$ seed layers on Si substrates. As the deposited Mn thickness varies, nitrogen ion migration across the Mn$_{3}$N$_{2}$/Mn layers leads to a continuous evolution of the layers to Mn$_{3}$N$_{2}$/Mn$_{2}$N/Mn$_{4}$N, Mn$_{2}$N/Mn$_{4}$N, and eventually Mn$_{4}$N alone. The ferrimagnetic Mn$_{4}$N indeed exhibits perpendicular magnetic anisotropy, and forms via a nucleation-and-growth mechanism. The nitrogen ion migration is also manifested in a significant exchange bias, up to 0.3 T at 5 K, due to the interactions between ferrimagnetic Mn$_{4}$N and antiferromagnetic Mn$_{3}$N$_{2}$ and Mn$_{2}$N. These results demonstrate a promising all-nitride magneto-ionic platform with remarkable tunability for device applications., Comment: 21 pages, 5 figures, 7 pages of supplementary material with 5 figures

The audio-visual segmentation (AVS) task aims to segment sounding objects from a given video. Existing works mainly focus on fusing audio and visual features of a given video to achieve sounding object masks. However, we observed that prior arts are prone to segment a certain salient object in a video regardless of the audio information. This is because sounding objects are often the most salient ones in the AVS dataset. Thus, current AVS methods might fail to localize genuine sounding objects due to the dataset bias. In this work, we present an audio-visual instance-aware segmentation approach to overcome the dataset bias. In a nutshell, our method first localizes potential sounding objects in a video by an object segmentation network, and then associates the sounding object candidates with the given audio. We notice that an object could be a sounding object in one video but a silent one in another video. This would bring ambiguity in training our object segmentation network as only sounding objects have corresponding segmentation masks. We thus propose a silent object-aware segmentation objective to alleviate the ambiguity. Moreover, since the category information of audio is unknown, especially for multiple sounding sources, we propose to explore the audio-visual semantic correlation and then associate audio with potential objects. Specifically, we attend predicted audio category scores to potential instance masks and these scores will highlight corresponding sounding instances while suppressing inaudible ones. When we enforce the attended instance masks to resemble the ground-truth mask, we are able to establish audio-visual semantics correlation. Experimental results on the AVS benchmarks demonstrate that our method can effectively segment sounding objects without being biased to salient objects., Comment: This paper has been received by ACM MM 23

For time-dependent PDEs, the numerical schemes can be rendered bound-preserving without losing conservation and accuracy, by a post processing procedure of solving a constrained minimization in each time step. Such a constrained optimization can be formulated as a nonsmooth convex minimization, which can be efficiently solved by first order optimization methods, if using the optimal algorithm parameters. By analyzing the asymptotic linear convergence rate of the generalized Douglas-Rachford splitting method, optimal algorithm parameters can be approximately expressed as a simple function of the number of out-of-bounds cells. We demonstrate the efficiency of this simple choice of algorithm parameters by applying such a limiter to cell averages of a discontinuous Galerkin scheme solving phase field equations for 3D demanding problems. Numerical tests on a sophisticated 3D Cahn-Hilliard-Navier-Stokes system indicate that the limiter is high order accurate, very efficient, and well-suited for large-scale simulations. For each time step, it takes at most $20$ iterations for the Douglas-Rachford splitting to enforce bounds and conservation up to the round-off error, for which the computational cost is at most $80N$ with $N$ being the total number of cells.

Power analysis is a class of side-channel attacks, where power consumption data is used to infer sensitive information and extract secrets from a system. Traditionally, such attacks required physical access to the target, as well as specialized devices to measure the power consumption with enough precision. The PLATYPUS attack has shown that on-chip power meter capabilities exposed to a software interface might form a new class of power side-channel attacks. This paper presents a software-based power side-channel attack on Apple Silicon M1/M2 platforms, exploiting the System Management Controller (SMC) and its power-related keys, which provides access to the on-chip power meters through a software interface to user space software. We observed data-dependent power consumption reporting from such keys and analyzed the correlations between the power consumption and the processed data. Our work also demonstrated how an unprivileged user mode application successfully recovers bytes from an AES encryption key from a cryptographic service supported by a kernel mode driver in macOS. Furthermore, we discuss the impact of software-based power side-channels in the industry, possible countermeasures, and the overall implications of software interfaces for modern on-chip power management systems., Comment: 6 pages, 4 figures, 5 tables

Generating vivid and diverse 3D co-speech gestures is crucial for various applications in animating virtual avatars. While most existing methods can generate gestures from audio directly, they usually overlook that emotion is one of the key factors of authentic co-speech gesture generation. In this work, we propose EmotionGesture, a novel framework for synthesizing vivid and diverse emotional co-speech 3D gestures from audio. Considering emotion is often entangled with the rhythmic beat in speech audio, we first develop an Emotion-Beat Mining module (EBM) to extract the emotion and audio beat features as well as model their correlation via a transcript-based visual-rhythm alignment. Then, we propose an initial pose based Spatial-Temporal Prompter (STP) to generate future gestures from the given initial poses. STP effectively models the spatial-temporal correlations between the initial poses and the future gestures, thus producing the spatial-temporal coherent pose prompt. Once we obtain pose prompts, emotion, and audio beat features, we will generate 3D co-speech gestures through a transformer architecture. However, considering the poses of existing datasets often contain jittering effects, this would lead to generating unstable gestures. To address this issue, we propose an effective objective function, dubbed Motion-Smooth Loss. Specifically, we model motion offset to compensate for jittering ground-truth by forcing gestures to be smooth. Last, we present an emotion-conditioned VAE to sample emotion features, enabling us to generate diverse emotional results. Extensive experiments demonstrate that our framework outperforms the state-of-the-art, achieving vivid and diverse emotional co-speech 3D gestures. Our code and dataset will be released at the project page: https://xingqunqi-lab.github.io/Emotion-Gesture-Web/, Comment: Under review

Graphene aerogel (GA) is a promising material for thermal management applications across many fields due to its lightweight and thermally insulative properties. However, standard values for important thermal properties, such as thermal conductivity, remain elusive due to the lack of reliable characterization techniques for highly porous materials. Comparative infrared thermal microscopy (CITM) is an attractive technique to obtain thermal conductance values of porous materials like GA, due to its non-invasive character, which requires no probing of, or contact with, the often-delicate structures and frameworks. In this study, we improve upon CITM by utilizing a higher resolution imaging setup and reducing the need for pore-filling coating of the sample (previously used to adjust for emissivity). This upgraded setup, verified by characterizing porous silica aerogel, allows for a more accurate confirmation of the fundamental thermal conductivity value of GA while still accounting for the thermal resistance at material boundaries. Using this improved method, we measure a thermal conductivity below 0.036 W/m$\cdot$K for commercial GA using multiple reference materials. These measurements demonstrate the impact of higher resolution thermal imaging to improve accuracy in low density, highly porous materials characterization. This study also reports thermal conductivity for much lower density (less than 15 mg/cm$^3$) GA than previously published studies while maintaining the robustness of the CITM technique., Comment: v3: Submitted to Journal of Porous Materials v2: Resubmitted to correct MathJax typos in abstract

In this paper, we are interested in constructing a scheme solving compressible Navier--Stokes equations, with desired properties including high order spatial accuracy, conservation, and positivity-preserving of density and internal energy under a standard hyperbolic type CFL constraint on the time step size, e.g., $\Delta t=\mathcal O(\Delta x)$. Strang splitting is used to approximate convection and diffusion operators separately. For the convection part, i.e., the compressible Euler equation, the high order accurate postivity-preserving Runge--Kutta discontinuous Galerkin method can be used. For the diffusion part, the equation of internal energy instead of the total energy is considered, and a first order semi-implicit time discretization is used for the ease of achieving positivity. A suitable interior penalty discontinuous Galerkin method for the stress tensor can ensure the conservation of momentum and total energy for any high order polynomial basis. In particular, positivity can be proven with $\Delta t=\mathcal{O}(\Delta x)$ if the Laplacian operator of internal energy is approximated by the $\mathbb{Q}^k$ spectral element method with $k=1,2,3$. So the full scheme with $\mathbb{Q}^k$ ($k=1,2,3$) basis is conservative and positivity-preserving with $\Delta t=\mathcal{O}(\Delta x)$, which is robust for demanding problems such as solutions with low density and low pressure induced by high-speed shock diffraction. Even though the full scheme is only first order accurate in time, numerical tests indicate that higher order polynomial basis produces much better numerical solutions, e.g., better resolution for capturing the roll-ups during shock reflection.

In this paper, we propose an efficient self-supervised arbitrary-scale super-resolution (SR) framework to reconstruct isotropic magnetic resonance (MR) images from anisotropic MRI inputs without involving external training data. The proposed framework builds a training dataset using in-the-wild anisotropic MR volumes with arbitrary image resolution. We then formulate the 3D volume SR task as a SR problem for 2D image slices. The anisotropic volume's high-resolution (HR) plane is used to build the HR-LR image pairs for model training. We further adapt the implicit neural representation (INR) network to implement the 2D arbitrary-scale image SR model. Finally, we leverage the well-trained proposed model to up-sample the 2D LR plane extracted from the anisotropic MR volumes to their HR views. The isotropic MR volumes thus can be reconstructed by stacking and averaging the generated HR slices. Our proposed framework has two major advantages: (1) It only involves the arbitrary-resolution anisotropic MR volumes, which greatly improves the model practicality in real MR imaging scenarios (e.g., clinical brain image acquisition); (2) The INR-based SR model enables arbitrary-scale image SR from the arbitrary-resolution input image, which significantly improves model training efficiency. We perform experiments on a simulated public adult brain dataset and a real collected 7T brain dataset. The results indicate that our current framework greatly outperforms two well-known self-supervised models for anisotropic MR image SR tasks., Comment: 10 pages, 5 figures

Quantum Local Search (QLS) is a promising approach that employs small-scale quantum computers to tackle large combinatorial optimization problems through local search on quantum hardware, starting from an initial point. However, the random selection of the sub-problem to solve in QLS may not be efficient. In this study, we propose a reinforcement learning (RL) based approach to train an agent for improved subproblem selection in QLS, beyond random selection. Our results demonstrate that the RL agent effectively enhances the average approximation ratio of QLS on fully-connected random Ising problems, indicating the potential of combining RL techniques with Noisy Intermediate-scale Quantum (NISQ) algorithms. This research opens a promising direction for integrating RL into quantum computing to enhance the performance of optimization tasks., Comment: 5 pages, 2 figures

In the past few years, more and more AI applications have been applied to edge devices. However, models trained by data scientists with machine learning frameworks, such as PyTorch or TensorFlow, can not be seamlessly executed on edge. In this paper, we develop an end-to-end code generator parsing a pre-trained model to C source libraries for the backend using MicroTVM, a machine learning compiler framework extension addressing inference on bare metal devices. An analysis shows that specific compute-intensive operators can be easily offloaded to the dedicated accelerator with a Universal Modular Accelerator (UMA) interface, while others are processed in the CPU cores. By using the automatically generated ahead-of-time C runtime, we conduct a hand gesture recognition experiment on an ARM Cortex M4F core., Comment: CODAI 2022 Workshop - Embedded System Week (ESWeek)

Grover search is a renowned quantum search algorithm that leverages quantum superposition to find a marked item with quadratic speedup. However, when implemented on Noisy Intermediate-scale Quantum (NISQ) hardware, the required repeated iterations of the oracle and diffusion operators increase exponentially with the number of qubits, resulting in significant noise accumulation. To address this, we propose a hybrid quantum-classical architecture that replaces quantum iterations with updates from a classical optimizer. This optimizer minimizes the expectation value of an oracle Hamiltonian with respect to a parameterized quantum state representing the target bit string. Our parameterized quantum circuit is much shallower than Grover search circuit, and we found that it outperforms Grover search on noisy simulators and NISQ hardware. When the number of qubits is greater than 5, our approach still maintains usable success probability, while the success probability of Grover search is at the same level as random guessing., Comment: 7 pages, 2 figures

Spatial attention has been widely used to improve the performance of convolutional neural networks. However, it has certain limitations. In this paper, we propose a new perspective on the effectiveness of spatial attention, which is that the spatial attention mechanism essentially solves the problem of convolutional kernel parameter sharing. However, the information contained in the attention map generated by spatial attention is not sufficient for large-size convolutional kernels. Therefore, we propose a novel attention mechanism called Receptive-Field Attention (RFA). Existing spatial attention, such as Convolutional Block Attention Module (CBAM) and Coordinated Attention (CA) focus only on spatial features, which does not fully address the problem of convolutional kernel parameter sharing. In contrast, RFA not only focuses on the receptive-field spatial feature but also provides effective attention weights for large-size convolutional kernels. The Receptive-Field Attention convolutional operation (RFAConv), developed by RFA, represents a new approach to replace the standard convolution operation. It offers nearly negligible increment of computational cost and parameters, while significantly improving network performance. We conducted a series of experiments on ImageNet-1k, COCO, and VOC datasets to demonstrate the superiority of our approach. Of particular importance, we believe that it is time to shift focus from spatial features to receptive-field spatial features for current spatial attention mechanisms. In this way, we can further improve network performance and achieve even better results. The code and pre-trained models for the relevant tasks can be found at https://github.com/Liuchen1997/RFAConv., Comment: 12 pages, 11figures