arXiv subjects

Rui Ye

Publications and source records attributed to Rui Ye.

63 records · Page 4Linked to original sources

arXiv2023

FedDisco: Federated Learning with Discrepancy-Aware Collaboration

This work considers the category distribution heterogeneity in federated learning. This issue is due to biased labeling preferences at multiple clients and is a typical setting of data heterogeneity. To alleviate this issue, most previous works consider either regularizing local models or fine-tuning the global model, while they ignore the adjustment of aggregation weights and simply assign weights based on the dataset size. However, based on our empirical observations and theoretical analysis, we find that the dataset size is not optimal and the discrepancy between local and global category distributions could be a beneficial and complementary indicator for determining aggregation weights. We thus propose a novel aggregation method, Federated Learning with Discrepancy-aware Collaboration (FedDisco), whose aggregation weights not only involve both the dataset size and the discrepancy value, but also contribute to a tighter theoretical upper bound of the optimization error. FedDisco also promotes privacy-preservation, communication and computation efficiency, as well as modularity. Extensive experiments show that our FedDisco outperforms several state-of-the-art methods and can be easily incorporated with many existing methods to further enhance the performance. Our code will be available at https://github.com/MediaBrain-SJTU/FedDisco.

cs.LG↗

arXiv2023

Observation of non-Hermitian antichiral edge currents

Non-Hermitian topological photonics is of great interest in bridging topological matter with gain/dissipation engineering in optics. A key problem in this direction is the interplay between the effective gauge potential and the non-Hermiticity. Here we tackle this problem in a synthetic non-Hermitian Hall ladder and experimentally observe antichiral edge currents (ACECs) of photons, by tuning the locally uniform effective magnetic flux and the on-site gain/loss. Such ACECs provide a topological method to probe the signatures of the non-Hermitian skin effect (NHSE) from steady-state bulk dynamics. The universality of this method is verified by its generalization to three dimensions. This study paves a way to investigate exotic non-Hermitian topological physics and has potential applications in topological photonics engineering.

physics.optics↗

arXiv2022

FedFM: Anchor-based Feature Matching for Data Heterogeneity in Federated Learning

One of the key challenges in federated learning (FL) is local data distribution heterogeneity across clients, which may cause inconsistent feature spaces across clients. To address this issue, we propose a novel method FedFM, which guides each client's features to match shared category-wise anchors (landmarks in feature space). This method attempts to mitigate the negative effects of data heterogeneity in FL by aligning each client's feature space. Besides, we tackle the challenge of varying objective function and provide convergence guarantee for FedFM. In FedFM, to mitigate the phenomenon of overlapping feature spaces across categories and enhance the effectiveness of feature matching, we further propose a more precise and effective feature matching loss called contrastive-guiding (CG), which guides each local feature to match with the corresponding anchor while keeping away from non-corresponding anchors. Additionally, to achieve higher efficiency and flexibility, we propose a FedFM variant, called FedFM-Lite, where clients communicate with server with fewer synchronization times and communication bandwidth costs. Through extensive experiments, we demonstrate that FedFM with CG outperforms several works by quantitative and qualitative comparisons. FedFM-Lite can achieve better performance than state-of-the-art methods with five to ten times less communication costs.

cs.LG↗

arXiv2021

MetaNODE: Prototype Optimization as a Neural ODE for Few-Shot Learning

Few-Shot Learning (FSL) is a challenging task, \emph{i.e.}, how to recognize novel classes with few examples? Pre-training based methods effectively tackle the problem by pre-training a feature extractor and then predicting novel classes via a cosine nearest neighbor classifier with mean-based prototypes. Nevertheless, due to the data scarcity, the mean-based prototypes are usually biased. In this paper, we attempt to diminish the prototype bias by regarding it as a prototype optimization problem. To this end, we propose a novel meta-learning based prototype optimization framework to rectify prototypes, \emph{i.e.}, introducing a meta-optimizer to optimize prototypes. Although the existing meta-optimizers can also be adapted to our framework, they all overlook a crucial gradient bias issue, \emph{i.e.}, the mean-based gradient estimation is also biased on sparse data. To address the issue, we regard the gradient and its flow as meta-knowledge and then propose a novel Neural Ordinary Differential Equation (ODE)-based meta-optimizer to polish prototypes, called MetaNODE. In this meta-optimizer, we first view the mean-based prototypes as initial prototypes, and then model the process of prototype optimization as continuous-time dynamics specified by a Neural ODE. A gradient flow inference network is carefully designed to learn to estimate the continuous gradient flow for prototype dynamics. Finally, the optimal prototypes can be obtained by solving the Neural ODE. Extensive experiments on miniImagenet, tieredImagenet, and CUB-200-2011 show the effectiveness of our method.

cs.CV↗

arXiv2021

SGMNet: Scene Graph Matching Network for Few-Shot Remote Sensing Scene Classification

Few-Shot Remote Sensing Scene Classification (FSRSSC) is an important task, which aims to recognize novel scene classes with few examples. Recently, several studies attempt to address the FSRSSC problem by following few-shot natural image classification methods. These existing methods have made promising progress and achieved superior performance. However, they all overlook two unique characteristics of remote sensing images: (i) object co-occurrence that multiple objects tend to appear together in a scene image and (ii) object spatial correlation that these co-occurrence objects are distributed in the scene image following some spatial structure patterns. Such unique characteristics are very beneficial for FSRSSC, which can effectively alleviate the scarcity issue of labeled remote sensing images since they can provide more refined descriptions for each scene class. To fully exploit these characteristics, we propose a novel scene graph matching-based meta-learning framework for FSRSSC, called SGMNet. In this framework, a scene graph construction module is carefully designed to represent each test remote sensing image or each scene class as a scene graph, where the nodes reflect these co-occurrence objects meanwhile the edges capture the spatial correlations between these co-occurrence objects. Then, a scene graph matching module is further developed to evaluate the similarity score between each test remote sensing image and each scene class. Finally, based on the similarity scores, we perform the scene class prediction via a nearest neighbor classifier. We conduct extensive experiments on UCMerced LandUse, WHU19, AID, and NWPU-RESISC45 datasets. The experimental results show that our method obtains superior performance over the previous state-of-the-art methods.

cs.CV↗

arXiv2021

RAP-Net: Region Attention Predictive Network for Precipitation Nowcasting

Natural disasters caused by heavy rainfall often cost huge loss of life and property. To avoid it, the task of precipitation nowcasting is imminent. To solve the problem, increasingly deep learning methods are proposed to forecast future radar echo images and then the predicted maps have converted the distribution of rainfall. The prevailing spatiotemporal sequence prediction methods apply ConvRNN structure which combines the Convolution and Recurrent neural network. Although improvements based on ConvRNN achieve remarkable success, these methods ignore capturing both local and global spatial features simultaneously, which degrades the nowcasting in the region of heavy rainfall. To address this issue, we proposed the Region Attention Block (RAB) and embed it into ConvRNN to enhance the forecast in the area with strong rainfall. Besides, the ConvRNN models are hard to memory longer history representations with limited parameters. Considering it, we propose Recall Attention Mechanism (RAM) to improve the prediction. By preserving longer temporal information, RAM contributes to the forecasting, especially in the middle rainfall intensity. The experiments show that the proposed model Region Attention Predictive Network (RAP-Net) has outperformed the state-of-art method.

cs.LG↗

arXiv2021

Observation of flat-band and band transition in the synthetic space

Constructions of synthetic lattices in photonics attract growingly attentions for exploring interesting physics beyond the geometric dimensionality, among which modulated ring resonator system has been proved as a powerful platform to create different kinds of connectivities between resonant modes along the synthetic frequency dimension with many theoretical proposals. Various experimental realizations are investigated in a single ring resonator, while building beyond simple synthetic lattices in multiple rings with different types remains lacking, which desires to be accomplished as an important step further. Here, we implement the experimental demonstration of generating the one-dimensional Lieb lattice along the frequency axis of light, realized in two coupled ring resonators while the larger ring undergoing dynamic modulation. Such synthetic photonic structure naturally exhibits the physics of flat band. We show that the time-resolved band structure read out from the drop-port output of the excited ring is the intensity projection of the band structure onto specific resonant mode in the synthetic momentum space, where gapless flat band, mode localization effect, and flat to non-flat band transition are observed in experiments and verified by simulations. Our work gives a direct evidence for the constructing synthetic Lieb lattice with two rings, which hence makes a solid step towards experimentally constructing more complicated lattices in multiple rings associated with synthetic frequency dimension.

physics.optics↗

arXiv2021

Investigating the effect of expected travel distance on individual descent speed in the stairwell with super long distance

Currently, there is an increasing number of super high-rise buildings in urban cities, the issue of evacuation in emergencies from such buildings comes to the fore. An evacuation experiment was carried out by our group in Shanghai Tower, it was found that the evacuation speed of pedestrians evacuated from the 126th floor was always slower than that of those from the 117th floor. Therefore, we propose a hypothesis that the expected evacuation distance will affect pedestrians' movement speed. In order to verify our conjecture, we conduct an experiment in a 12-story office building, that is, to study whether there would be an influence and what kind of influence would be caused on speed by setting the evacuation distance for participants in advance. According to the results, we find that with the increase of expected evacuation distance, the movement speed of pedestrians will decrease, which confirms our hypothesis. At the same time, we give the relation between the increase rate of evacuation distance and the decrease rate of speed. It also can be found that with the increase of expected evacuation distance, the speed decrease rate of the male is greater than that for female. In addition, we study the effects of actual evacuation distance, gender, BMI on evacuation speed. Finally, we obtain the correlation between heart rate and speed during evacuation. The results in this paper are beneficial to the study of pedestrian evacuation in super high-rise buildings.

physics.soc-ph↗

arXiv2017

Commissioning of te China-ADS injector-I testing facility

The 10 MeV accelerator-driven subcritical system (ADS) Injector-I test stand at Institute of High Energy Physics (IHEP) is a testing facility dedicated to demonstrate one of the two injector design schemes [Injector Scheme-I, which works at 325 MHz], for the ADS project in China. The Injector adopted a four vane copper structure RFQ with output energy of 3.2 MeV and a superconducting (SC) section accommodating fourteen \b{eta}g=0.12 single spoke cavities, fourteen SC solenoids and fourteen cold BPMs. The ion source was installed since April of 2014, periods of commissioning are regularly scheduled between installation phases of the rest of the injector. Continuous wave (CW) beam was shooting through the injector and 10 MeV CW proton beam with average beam current around 2 mA was obtained recently. This contribution describe the results achieved so far and the difficulties encountered in CW commissioning.

physics.acc-ph↗