Search

Federated learning (FL) has emerged as a new paradigm for privacy-preserving collaborative training. Under domain skew, the current FL approaches are biased and face two fairness problems. 1) Parameter Update Conflict: data disparity among clients leads to varying parameter importance and inconsistent update directions. These two disparities cause important parameters to potentially be overwhelmed by unimportant ones of dominant updates. It consequently results in significant performance decreases for lower-performing clients. 2) Model Aggregation Bias: existing FL approaches introduce unfair weight allocation and neglect domain diversity. It leads to biased model convergence objective and distinct performance among domains. We discover a pronounced directional update consistency in Federated Learning and propose a novel framework to tackle above issues. First, leveraging the discovered characteristic, we selectively discard unimportant parameter updates to prevent updates from clients with lower performance overwhelmed by unimportant parameters, resulting in fairer generalization performance. Second, we propose a fair aggregation objective to prevent global model bias towards some domains, ensuring that the global model continuously aligns with an unbiased model. The proposed method is generic and can be combined with other existing FL methods to enhance fairness. Comprehensive experiments on Digits and Office-Caltech demonstrate the high fairness and performance of our method., Comment: Accepted by CVPR2024

Ultra-massive multiple-input multiple-output (UM-MIMO) is the enabler of Terahertz (THz) communications in next-generation wireless networks. In THz UM-MIMO systems, a new paradigm of cross-field communications spanning from near-field to far-field is emerging, since the near-field range expands with higher frequencies and larger array apertures. Precise beam alignment in cross-field is critical but challenging. Specifically, unlike far-field beams that rely only on the angle domain, the incorporation of dual-domain (angle and distance) training significantly increases overhead. A natural question arises of whether far-field beam training can be deployed for cross-field beam alignment. In this paper, this question is answered, by demonstrating that the far-field training enables sufficient signal-to-noise ratio (SNR) in both far- and near-field scenarios, while exciting all channel dimensions. Based on that, we propose a subarray-coordinated hierarchical (SCH) training with greatly reduced overhead. To further obtain high-precision beam designs, we propose a two-phase angle and distance beam estimator (TPBE). Extensive simulations demonstrate the effectiveness of the proposed methods. Compared to near-field exhaustive search, the SCH possesses 0.2\% training overhead. The TPBE achieves 0.01~degrees and 0.02~m estimation root-mean-squared errors for angle and distance. Furthermore, with the estimated beam directions, a near-optimal SNR with 0.11~dB deviation is attained after beam alignment.

We study the orbifold Hirzebruch-Riemann-Roch (HRR) theorem for quotient Deligne-Mumford stacks, explore its relation with the representation theory of finite groups, and derive a new orbifold HRR formula via an orbifold Mukai pairing., Comment: This paper is part of the author's PhD thesis. arXiv admin note: substantial text overlap with arXiv:2204.09824

Automated log analysis is crucial in modern software-intensive systems for facilitating program comprehension throughout software maintenance and engineering life cycles. Existing methods perform tasks such as log parsing and log anomaly detection by providing a single prediction value without interpretation. However, given the increasing volume of system events, the limited interpretability of analysis results hinders analysts' comprehension of program status and their ability to take appropriate actions. Moreover, these methods require substantial in-domain training data, and their performance declines sharply (by up to 62.5%) in online scenarios involving unseen logs from new domains, a common occurrence due to rapid software updates. In this paper, we propose LogPrompt, a novel interpretable log analysis approach for online scenarios. LogPrompt employs large language models (LLMs) to perform online log analysis tasks via a suite of advanced prompt strategies tailored for log tasks, which enhances LLMs' performance by up to 380.7% compared with simple prompts. Experiments on nine publicly available evaluation datasets across two tasks demonstrate that LogPrompt, despite requiring no in-domain training, outperforms existing approaches trained on thousands of logs by up to 55.9%. We also conduct a human evaluation of LogPrompt's interpretability, with six practitioners possessing over 10 years of experience, who highly rated the generated content in terms of usefulness and readability (averagely 4.42/5). LogPrompt also exhibits remarkable compatibility with open-source and smaller-scale LLMs, making it flexible for practical deployment. Code of LogPrompt is available at https://github.com/lunyiliu/LogPrompt., Comment: Accepted by ICPC 2024

In this paper we study equivariant moduli spaces of sheaves on a $ K3 $ surface $ X $ under a symplectic action of a finite group. We prove that under some mild conditions, equivariant moduli spaces of sheaves on $ X $ are irreducible symplectic manifolds deformation equivalent to Hilbert schemes of points on $ X $ via a connection between Gieseker and Bridgeland moduli spaces, as well as the derived McKay correspondence., Comment: This paper is part of the author's PhD thesis [arXiv:2204.09824]

Anomaly detection in multivariate time series data is of paramount importance for ensuring the efficient operation of large-scale systems across diverse domains. However, accurately detecting anomalies in such data poses significant challenges. Existing approaches, including forecasting and reconstruction-based methods, struggle to address these challenges effectively. To overcome these limitations, we propose a novel anomaly detection framework named ImDiffusion, which combines time series imputation and diffusion models to achieve accurate and robust anomaly detection. The imputation-based approach employed by ImDiffusion leverages the information from neighboring values in the time series, enabling precise modeling of temporal and inter-correlated dependencies, reducing uncertainty in the data, thereby enhancing the robustness of the anomaly detection process. ImDiffusion further leverages diffusion models as time series imputers to accurately capturing complex dependencies. We leverage the step-by-step denoised outputs generated during the inference process to serve as valuable signals for anomaly prediction, resulting in improved accuracy and robustness of the detection process. We evaluate the performance of ImDiffusion via extensive experiments on benchmark datasets. The results demonstrate that our proposed framework significantly outperforms state-of-the-art approaches in terms of detection accuracy and timeliness. ImDiffusion is further integrated into the real production system in Microsoft and observe a remarkable 11.4% increase in detection F1 score compared to the legacy approach. To the best of our knowledge, ImDiffusion represents a pioneering approach that combines imputation-based techniques with time series anomaly detection, while introducing the novel use of diffusion models to the field., Comment: To appear in VLDB 2024.Code: https://github.com/17000cyh/IMDiffusion.git

The loss function for bounding box regression (BBR) is essential to object detection. Its good definition will bring significant performance improvement to the model. Most existing works assume that the examples in the training data are high-quality and focus on strengthening the fitting ability of BBR loss. If we blindly strengthen BBR on low-quality examples, it will jeopardize localization performance. Focal-EIoU v1 was proposed to solve this problem, but due to its static focusing mechanism (FM), the potential of non-monotonic FM was not fully exploited. Based on this idea, we propose an IoU-based loss with a dynamic non-monotonic FM named Wise-IoU (WIoU). The dynamic non-monotonic FM uses the outlier degree instead of IoU to evaluate the quality of anchor boxes and provides a wise gradient gain allocation strategy. This strategy reduces the competitiveness of high-quality anchor boxes while also reducing the harmful gradient generated by low-quality examples. This allows WIoU to focus on ordinary-quality anchor boxes and improve the detector's overall performance. When WIoU is applied to the state-of-the-art real-time detector YOLOv7, the AP-75 on the MS-COCO dataset is improved from 53.03% to 54.50%. Code is available at https://github.com/Instinct323/wiou.

Terahertz (THz) band owning the abundant multi-ten-GHz bandwidth is capable to support Terabit-per-second wireless communications, which is a pillar technology for 6G and beyond systems. With sub-millimeter-long antennas, ultra-massive (UM) MIMO and intelligent surface (IS) systems with thousands of array elements are exploited to effectively combat the distance limitation and blockage problems, which compose a promising THz ultra-large antenna array (ULAA) system. As a combined effect of wavelength and array aperture, the resulting coverage of THz systems ranges from near-field to far-field, leading to a new paradigm of cross-field communications. Although channel models, communications theories, and networking strategies have been studied for far-field and near-field separately, the unified design of cross-field communications that achieve high spectral efficiency and low complexity is still missing. In this article, the challenges and features of THz ULAA cross-field communications are investigated. Furthermore, cross-field solutions in three perspectives are presented, including a hybrid spherical- and planar-wave channel model, cross-field channel estimation, and widely-spaced multi-subarray hybrid beamforming, where a subarray as a basic unit in THz ULAA systems is exploited. The approximation error of channel modeling accuracy, spectral efficiency, and estimation error of these designs are numerically evaluated. Finally, as a roadmap of THz ULAA cross-field communications, multiple open problems and potential research directions are elaborated.

Photons with spin angular momentum possess intrinsic chirality which underpins many phenomena including nonlinear optics1, quantum optics2, topological photonics3 and chiroptics4. Intrinsic chirality is weak in natural materials, and recent theoretical proposals5-7 aimed to enlarge circular dichroism by resonant metasurfaces supporting bound states in the continuum that enhance substantially chiral light-matter interaction. Those insightful works resort to three-dimensional sophisticated geometries, which are too challenging to be realized for optical frequencies8. Therefore, most of the experimental attempts9-11 showing strong circular dichroism rely on false/extrinsic chirality by employing either oblique incidence9, 10 or structural anisotropy11. Here, we report on the experimental realization of true/intrinsic chiral response with resonant metasurfaces where the engineered slant geometry breaks both in-plane and out-of-plane symmetries. Our result marks the first observation of intrinsic chiral bound states in the continuum with near-unity chiral dichroism of 0.93 and record-high quality factor exceeding 2663 for visible frequencies. Our chiral metasurfaces promise a plethora of applications in chiral light sources and detectors, chiral sensing, valleytronics and asymmetric photocatalysis.

Due to the complicated background and noise of infrared images, infrared small target detection is one of the most difficult problems in the field of computer vision. In most existing studies, semantic segmentation methods are typically used to achieve better results. The centroid of each target is calculated from the segmentation map as the detection result. In contrast, we propose a novel end-to-end framework for infrared small target detection and segmentation in this paper. First, with the use of UNet as the backbone to maintain resolution and semantic information, our model can achieve a higher detection accuracy than other state-of-the-art methods by attaching a simple anchor-free head. Then, a pyramid pool module is used to further extract features and improve the precision of target segmentation. Next, we use semantic segmentation tasks that pay more attention to pixel-level features to assist in the training process of object detection, which increases the average precision and allows the model to detect some targets that were previously not detectable. Furthermore, we develop a multi-task framework for infrared small target detection and segmentation. Our multi-task learning model reduces complexity by nearly half and speeds up inference by nearly twice compared to the composite single-task model, while maintaining accuracy. The code and models are publicly available at https://github.com/Chenastron/MTUNet.

Integrated ultra-massive multiple-input multiple-output (UM-MIMO) and intelligent reflecting surface (IRS) systems are promising for 6G and beyond Terahertz (0.1-10 THz) communications, to effectively bypass the barriers of limited coverage and line-of-sight blockage. However, excessive dimensions of UM-MIMO and IRS enlarge the near-field region, while strong THz channel sparsity in far-field is detrimental to spatial multiplexing. Moreover, channel estimation (CE) requires recovering the large-scale channel from severely compressed observations due to limited RF-chains. To tackle these challenges, a hybrid spherical- and planar-wave channel model (HSPM) is developed for the cascaded channel of the integrated system. The spatial multiplexing gains under near-field and far-field regions are analyzed, which are found to be limited by the segmented channel with a lower rank. Furthermore, a compressive sensing-based CE framework is developed, including a sparse channel representation method, a separate-side estimation (SSE) and a dictionary-shrinkage estimation (DSE) algorithms. Numerical results verify the effectiveness of the HSPM, the capacity of which is only $5\times10^{-4}$ bits/s/Hz deviated from that obtained by the ground-truth spherical-wave-model, with 256 elements. While the SSE achieves improved accuracy for CE than benchmark algorithms, the DSE is more attractive in noisy environments, with 0.8 dB lower normalized-mean-square-error than SSE.

We study the orbifold Hirzebruch-Riemann-Roch (HRR) theorem for quotient Deligne-Mumford stacks, explore its relation with the representation theory of finite groups, and derive a new orbifold HRR formula via an orbifold Mukai pairing. As a first application, we use this formula to compute the dimensions of $ G $-equivariant moduli spaces of stable sheaves on a $ K3 $ surface $ X $ under the action of a finite subgroup $ G $ of its symplectic automorphism group. We then apply the orbifold HRR formula to reproduce the number of fixed points on $ X $ when $ G $ is cyclic without using the Lefschetz fixed point formula. We prove that under some mild conditions, equivariant moduli spaces of stable sheaves on $ X $ are irreducible symplectic manifolds deformation equivalent to Hilbert schemes of points on $ X $ via a connection between Gieseker and Bridgeland moduli spaces, as well as the derived McKay correspondence., Comment: Changed the title, corrected the definition of a Bridgeland stability condition, and made some minor changes

Millimeter-wave (mmWave) and Terahertz (THz)-band communications exploit the abundant bandwidth to fulfill the increasing data rate demands of 6G wireless communications. To compensate for the high propagation loss with reduced hardware costs, ultra-massive multiple-input multiple-output (UM-MIMO) with a hybrid beamforming structure is a promising technology in the mmWave and THz bands. However, channel estimation (CE) is challenging for hybrid UM-MIMO systems, which requires recovering the high-dimensional channels from severely few channel observations. In this paper, a Pruned Approximate Message Passing (AMP) Integrated Deep Convolutional-neural-network (DCNN) CE (PRINCE) method is firstly proposed, which enhances the estimation accuracy of the AMP method by appending a DCNN network. Moreover, by truncating the insignificant feature maps in the convolutional layers of the DCNN network, a pruning method including training with regularization, pruning and refining procedures is developed to reduce the network scale. Simulation results show that the PRINCE achieves a good trade-off between the CE accuracy and significantly low complexity, with normalized-mean-square-error (NMSE) of $-10$ dB at signal-to-noise-ratio (SNR) as $10$ dB after eliminating $80\%$ feature maps.

Reconfigurable intelligent surfaces (RISs) have emerged as a prospective technology for next-generation wireless networks due to their potential in coverage and capacity enhancement. Previous works on achievable rate analysis of RIS-assisted communication systems have mainly focused on the rich-scattering environment where Rayleigh and Rician channel models can be applied. This work studies the ergodic achievable rate of RIS-assisted multiple-input multiple-output communication systems in millimeter-wave band with limited scattering under the Saleh-Valenzuela channel model. Firstly, we derive an upper bound of the ergodic achievable rate by means of majorization theory and Jensen's inequality. The upper bound shows that the ergodic achievable rate increases logarithmically with the number of antennas at the base station (BS) and user, the number of the reflection units at the RIS, and the eigenvalues of the steering matrices associated with the BS, user and RIS. Then, we aim to maximize the ergodic achievable rate by jointly optimizing the transmit covariance matrix at the BS and the reflection coefficients at the RIS. Specifically, the transmit covariance matrix is optimized by the water-filling algorithm and the reflection coefficients are optimized using the Riemannian conjugate gradient algorithm. Simulation results validate the effectiveness of the proposed optimization algorithms., Comment: 30 pages, 11 figures

ComOpT is an open-source research tool for coverage-driven testing of autonomous driving systems, focusing on planning and control. Starting with (i) a meta-model characterizing discrete conditions to be considered and (ii) constraints specifying the impossibility of certain combinations, ComOpT first generates constraint-feasible abstract scenarios while maximally increasing the coverage of k-way combinatorial testing. Each abstract scenario can be viewed as a conceptual equivalence class, which is then instantiated into multiple concrete scenarios by (1) randomly picking one local map that fulfills the specified geographical condition, and (2) assigning all actors accordingly with parameters within the range. Finally, ComOpT evaluates each concrete scenario against a set of KPIs and performs local scenario variation via spawning a new agent that might lead to a collision at designated points. We use ComOpT to test the Apollo~6 autonomous driving software stack. ComOpT can generate highly diversified scenarios with limited test budgets while uncovering problematic situations such as inabilities to make simple right turns, uncomfortable accelerations, and dangerous driving patterns. ComOpT participated in the 2021 IEEE AI Autonomous Vehicle Testing Challenge and won first place among more than 110 contending teams.

As random walk is a powerful tool in many graph processing, mining and learning applications, this paper proposes an efficient in-memory random walk engine named ThunderRW. Compared with existing parallel systems on improving the performance of a single graph operation, ThunderRW supports massive parallel random walks. The core design of ThunderRW is motivated by our profiling results: common RW algorithms have as high as 73.1% CPU pipeline slots stalled due to irregular memory access, which suffers significantly more memory stalls than the conventional graph workloads such as BFS and SSSP. To improve the memory efficiency, we first design a generic step-centric programming model named Gather-Move-Update to abstract different RW algorithms. Based on the programming model, we develop the step interleaving technique to hide memory access latency by switching the executions of different random walk queries. In our experiments, we use four representative RW algorithms including PPR, DeepWalk, Node2Vec and MetaPath to demonstrate the efficiency and programming flexibility of ThunderRW. Experimental results show that ThunderRW outperforms state-of-the-art approaches by an order of magnitude, and the step interleaving technique significantly reduces the CPU pipeline stall from 73.1% to 15.0%.