Identically Distributed Research Articles

Out-of-Distribution Detection by Cross-Class Vicinity Distribution of In-Distribution Data.

Deep neural networks for image classification only learn to map in-distribution inputs to their corresponding ground-truth labels in training without differentiating out-of-distribution samples from in-distribution ones. This results from the assumption that all samples are independent and identically distributed (IID) without distributional distinction. Therefore, a pretrained network learned from in-distribution samples treats out-of-distribution samples as in-distribution and makes high-confidence predictions on them in the test phase. To address this issue, we draw out-of-distribution samples from the vicinity distribution of training in-distribution samples for learning to reject the prediction on out-of-distribution inputs. A cross-class vicinity distribution is introduced by assuming that an out-of-distribution sample generated by mixing multiple in-distribution samples does not share the same classes of its constituents. We, thus, improve the discriminability of a pretrained network by finetuning it with out-of-distribution samples drawn from the cross-class vicinity distribution, where each out-of-distribution input corresponds to a complementary label. Experiments on various in-/out-of-distribution datasets show that the proposed method significantly outperforms the existing methods in improving the capacity of discriminating between in- and out-of-distribution samples.

Minimizing distance between distribution functions: discrete counterparts to continuous random variables with applications in non-life insurance and stochastic reliability

In this work, we propose a novel family of procedures for deriving a discrete counterpart to a continuous probability distribution. They are based on a class of distances between cumulative distribution functions, including the Cramér, the Cramér-von Mises, and the Anderson-Darling distances as particular cases. The discrete counterpart is defined and derived as the random variable which minimizes its distance to the assigned continuous probability distribution among all the discrete random variables supported on the set of integers (or positive integers). Applications are provided with reference to the exponential and the normal distributions, among others; the discrete counterparts are derived, and their main properties are discussed, also in comparison with the one obtained through an existing discretization technique based on the preservation of the cumulative distribution function at integer values. Parameter estimation for these discrete analogs is discussed, along with an analysis of two real datasets, where they are compared in terms of goodness-of-fit with some popular discrete distributions. Furthermore, in order to highlight the effectiveness and the benefits derived from the proposed discretization procedures, we illustrate two practical applications in actuarial science and in reliability engineering. In the former case, the problem of determining the distribution of the total claims amount for a non-life insurance portfolio is considered, where the claim sizes can be modelled as iid random variables, and the number of claims is random as well. Actuaries use a recursive calculation method based on Panjer's formula, which requires an appropriate discretization of the individual claim distribution, and therefore the proposed procedures can be used. Since we consider two simple cases where the distribution of the total claims amount is analytically acquirable, the efficacy of the discretization procedures in the final approximation can be easily assessed and turns out to be satisfactory, especially when compared to the existing discretization. The latter case considers the determination of the reliability parameter for a complex stress-strength model. Here, the approximation by discretization is compared to Monte Carlo simulation and shown to be relevant: with a comparable if not smaller computational effort, discretization leads to similar results as simulation. Such discretizations can also naturally be applied to more complex problems such as scenario generation in stochastic programming. R code for this article is provided as supplementary material.

Domain Adaptation With Self-Supervised Learning and Feature Clustering for Intelligent Fault Diagnosis.

Domain adaptation indeed promotes the progress of intelligent fault diagnosis in industrial scenarios. The abundant labeled samples are not necessary. The identical distribution between the training and testing datasets is not any more the prerequisite for intelligent fault diagnosis working. However, two issues arise subsequently: Feature learning in domain adaptation framework tends to be biased to the source domain, and unreliable pseudolabeling seriously impacts on the conditional domain adaptation. In this article, a new domain adaptation approach with self-supervised learning and feature clustering (DASSL-FC) is proposed, trying to alleviate the issues by unbiased feature learning and pseudolabels updating strategy. Taking different transformation methods as pretext, the transformed data and its pretext train a neural network in an SSL way. As to pseudolabeling, clusters are taken as the auxiliary information to correct the network predicted labels in terms of the "strong cluster" rule. Then, the updated pseudolabels and their confidence are enforced to further estimate the conditional distribution discrepancy and its confidence weight. To verify the effectiveness of the proposed method, the experiments are implemented on intraplatform and interplatforms for simulating the practical scenarios.

Brave the Wind and the Waves: Discovering Robust and Generalizable Graph Lottery Tickets.

The training and inference of Graph Neural Networks (GNNs) are costly when scaling up to large-scale graphs. Graph Lottery Ticket (GLT) has presented the first attempt to accelerate GNN inference on large-scale graphs by jointly pruning the graph structure and the model weights. Though promising, GLT encounters robustness and generalization issues when deployed in real-world scenarios, which are also long-standing and critical problems in deep learning ideology. In real-world scenarios, the distribution of unseen test data is typically diverse. We attribute the failures on out-of-distribution (OOD) data to the incapability of discerning causal patterns, which remain stable amidst distribution shifts. In traditional spase graph learning, the model performance deteriorates dramatically as the graph/network sparsity exceeds a certain high level. Worse still, the pruned GNNs are hard to generalize to unseen graph data due to limited training set at hand. To tackle these issues, we propose the Resilient Graph Lottery Ticket (RGLT) to find more robust and generalizable GLT in GNNs. Concretely, we reactivate a fraction of weights/edges by instantaneous gradient information at each pruning point. After sufficient pruning, we conduct environmental interventions to extrapolate potential test distribution. Finally, we perform last several rounds of model averages to further improve generalization. We provide multiple examples and theoretical analyses that underpin the universality and reliability of our proposal. Further, RGLT has been experimentally verified across various independent identically distributed (IID) and out-of-distribution (OOD) graph benchmarks.

Rotation in stellar evolution: Probing the influence on population synthesis in high-redshift galaxies

Stellar population synthesis (SPS) is essential for understanding galaxy formation and evolution. However, the recent discovery of rotation-driven phenomena in star clusters warrants a review of uncertainties in SPS models caused by overlooked factors, including stellar rotation. In this study, we investigated the impact of rotation on SPS specifically using the PARSEC V2.0 rotation model and its implications for high-redshift galaxies with the JWST. Rotation enhances the ultraviolet (UV) flux for up to ∼400 Myr after the starburst, with the slope of UV increasing as the population becomes faster rotating and metal-poorer. Using the Prospector tool, we constructed simulated galaxies and deduce their properties associated with dust and star formation. Our results suggest that rapid rotation models result in a gradual UV slope up to 0.1 dex higher and an approximately 50% increase in dust attenuation for identical wide-band spectral energy distributions. Furthermore, we investigated biases if the stellar population was characterized by rapid rotation and demonstrate that accurate estimation can be achieved for rotation rates up to ωi = 0.6. Accounting for the bias in the case of rapid rotation aligns specific star formation rates more closely with predictions from theoretical models. Notably, this also implies a slightly higher level of dust attenuation than previously anticipated, while still allowing for a “dust-free” interpretation of the galaxy. The impact of rapid rotation SPS models on the rest-UV luminosity function is found to be minimal. Overall, our findings have potentially important implications for comprehending dust attenuation and mass assembly history in the high-redshift Universe.

Reweighted Extreme Learning Machine-Based Clutter Suppression and Range Compensation Algorithm for Non-Side-Looking Airborne Radar

Non-side-looking airborne radar provides important applications on account of its all-round multi-angle airspace coverage. However, it suffers clutter range dependence that makes the samples fail to satisfy the condition of being independent and identically distributed (IID), and it severely degrades traditional approaches to clutter suppression and target detection. In this paper, a novel reweighted extreme learning machine (ELM)-based clutter suppression and range compensation algorithm is proposed for non-side-looking airborne radar. The proposed method involves first designing the pre-processing stage, the special reweighted complex-valued activation function containing an unknown range compensation matrix, and two new objective outputs for constructing an initial reweighted ELM-based network with its training. Then, two other objective outputs, a new loss function, and a reverse feedback framework driven by the specifically designed objectives are proposed for the unknown range compensation matrix. Finally, aiming to estimate and reconstruct the unknown compensation matrix, special processes of the complex-valued structures and the theoretical derivations are designed and analyzed in detail. Consequently, with the updated and compensated samples, further processing including space–time adaptive processing (STAP) can be performed for clutter suppression and target detection. Compared with the classic relevant methods, the proposed algorithm achieves significantly superior performance with reasonable computation time. It provides an obviously higher detection probability and better improvement factor (IF). The simulation results verify that the proposed algorithm is effective and has many advantages.

Conditional Strong Law of Large Numbers under G-Expectations

In this paper, we investigate two types of the conditional strong law of large numbers with a new notion of conditionally independent random variables under G-expectation which are related to the symmetry G-function. Our limit theorem demonstrates that the cluster points of empirical averages fall within the bounds of the lower and upper conditional expectations with lower probability one. Moreover, for conditionally independent random variables with identical conditional distributions, we show the existence of two cluster points of empirical averages that correspond to the essential minimum and essential maximum expectations, respectively, with G-capacity one.

HSG-MGAF Net: Heterogeneous subgraph-guided multiscale graph attention fusion network for interpretable prediction of whole-slide image

Background and objectivePathological whole slide image (WSI) prediction and region of interest (ROI) localization are important issues in computer-aided diagnosis and postoperative analysis in clinical applications. Existing computer-aided methods for predicting WSI are mainly based on multiple instance learning (MIL) and its variants. However, most of the methods are based on instance independence and identical distribution assumption and performed at a single scale, which not fully exploit the hierarchical multiscale heterogeneous information contained in WSI. MethodsHeterogeneous Subgraph-Guided Multiscale Graph Attention Fusion Network (HSG-MGAF Net) is proposed to build the topology of critical image patches at two scales for adaptive WSI prediction and lesion localization. The HSG-MGAF Net simulates the hierarchical heterogeneous information of WSI through graph and hypergraph at two scales, respectively. This framework not only fully exploits the low-order and potential high-order correlations of image patches at each scale, but also leverages the heterogeneous information of the two scales for adaptive WSI prediction. ResultsWe validate the superiority of the proposed method on the CAMELYON16 and the TCGA- NSCLC, and the results show that HSG-MGAF Net outperforms the state-of-the-art method on both datasets. The average ACC, AUC and F1 score of HSG-MGAF Net can reach 92.7 %/0.951/0.892 and 92.2 %/0.957/0.919, respectively. The obtained heatmaps can also localize the positive regions more accurately, which have great consistency with the pixel-level labels. ConclusionsThe results demonstrate that HSG-MGAF Net outperforms existing weakly supervised learning methods by introducing critical heterogeneous information between the two scales. This approach paves the way for further research on light weighted heterogeneous graph-based WSI prediction and ROI localization.

Clarification of the identity and geographic distribution of Phyllanthus claussenii and typification of two names in Phyllanthaceae (Phyllanthoideae)

Phyllanthus claussenii is one of the most frequently mentioned species of Phyllanthus sect. Phyllanthus subsect. Clausseniani in various studies. However, it is often mistaken for P. subemarginatus in most collections. In the present work, we present a clarification of the identity and geographic distribution of P. claussenii and the typification of Phyllanthus acuminatus and Savia sessiliflora. The proposal of lectotypes for P. acuminatus and S. sessiliflora will enable the accurate identification of the species and avoid potential ambiguity. To achieve this, protologues and type specimens, including digital images, were analyzed, along with specimens from collections of national and international herbaria.

Exploring biometric domain adaptation in human action recognition models for unconstrained environments

AbstractIn conventional machine learning (ML), a fundamental assumption is that the training and test sets share identical feature distributions, a reasonable premise drawn from the same dataset. However, real-world scenarios often defy this assumption, as data may originate from diverse sources, causing disparities between training and test data distributions. This leads to a domain shift, where variations emerge between the source and target domains. This study delves into human action recognition (HAR) models within an unconstrained, real-world setting, scrutinizing the impact of input data variations related to contextual information and video encoding. The objective is to highlight the intricacies of model performance and interpretability in this context. Additionally, the study explores the domain adaptability of HAR models, specifically focusing on their potential for re-identifying individuals within uncontrolled environments. The experiments involve seven pre-trained backbone models and introduce a novel analytical approach by linking domain-related (HAR) and domain-unrelated (re-identification (re-ID)) tasks. Two key analyses addressing contextual information and encoding strategies reveal that maintaining the same encoding approach during training results in high task correlation while incorporating richer contextual information enhances performance. A notable outcome of this study is the comprehensive evaluation of a novel transformer-based architecture driven by a HAR backbone, which achieves a robust re-ID performance superior to state-of-the-art (SOTA). However, it faces challenges when other encoding schemes are applied, highlighting the role of the HAR classifier in performance variations.

Addressing Heterogeneity in Federated Learning with Client Selection via Submodular Optimization

Federated learning (FL) has been proposed as a privacy-preserving distributed learning paradigm, which differs from traditional distributed learning in two main aspects: the systems heterogeneity, meaning that clients participating in training have significant differences in systems performance including CPU frequency, dataset size, and transmission power, and the statistical heterogeneity, indicating that the data distribution among clients exhibits Non-Independent Identical Distribution. Therefore, the random selection of clients will significantly reduce the training efficiency of FL. In this article, we propose a client selection mechanism considering both systems and statistical heterogeneity, which aims to improve the time-to-accuracy performance by trading off the impact of systems performance differences and data distribution differences among the clients on training efficiency. First, client selection is formulated as a combinatorial optimization problem that jointly optimizes systems and statistical performance. Then, we generalize it to a submodular maximization problem with knapsack constraint, and propose the Iterative Greedy with Partial Enumeration (IGPE) algorithm to greedily select the suitable clients. Then, the approximation ratio of IGPE is analyzed theoretically. Extensive experiments verify that the time-to-accuracy performance of the IGPE algorithm outperforms other compared algorithms in a variety of heterogeneous environments.

Refining intra-patch connectivity measures in landscape fragmentation and connectivity indices

ContextMeasuring intra-patch connectivity, i.e. the connectivity within a habitat patch, is important to evaluate landscape fragmentation and connectivity. However, intra-patch connectivity is mainly measured with patch size, which can conceal diverse intra-patch connectivity patterns for similar patch size distributions.ObjectivesWe suggest a method to refine the intra-patch connectivity component of fragmentation and connectivity indices. This method allows for distinguishing different intra-patch connectivity patterns for similar patch size distributions.MethodsWe used normalized patch complexity indices to weight patch size in common fragmentation and connectivity indices. Patch complexity indices included two existing geometrical indices (SHAPE and FRAC), and a new index derived from spatial network analysis, the mean detour index (MDI). We analyzed the behaviours of adjusted fragmentation and connectivity indices theoretically and empirically on both artificial and real landscapes.ResultsWhile maintaining the mathematical properties of fragmentation and connectivity indices, our method could distinguish landscapes with identical patch size distributions but different spatial configurations. The mean detour index had a different response than geometrical indices. This result indicates that, at the patch level, topological complexity can exhibit different patterns from geometrical complexity.ConclusionsMeasuring intra-patch connectivity with patch size in fragmentation and connectivity indices cannot distinguish landscapes having similar patch sizes distribution but different spatial configurations. This paper introduces a method to distinguish such patterns relying on geometrical and topological indices and shows to which extent it can impact conservation planning.

Super-heavy ion beams generated by a multi-PW femtosecond laser

The numerical investigations into the acceleration of superheavy ions driven by a multi-PW, 30 fs laser pulse with a peak intensity from 5 × 1022 to 2 × 1023 W/cm2 were carried out using an advanced 2D3V particle-in-cell code. The properties of laser-accelerated Au, Pb, Bi, and U ion beams, such as ionization and ion energy spectra, ion beam energies, angular distributions of the beam fluence, the ion pulse shapes, and peak intensities, were examined and compared. It was found that for a laser intensity of 1023 W/cm2, a common feature of the Au, Pb, Bi, and U ion beams was the dominance in the beam of Ne-like ions that carry the vast majority (≥90%) of the energy of all accelerated ions and have by far the highest mean and maximum ion energy. The Ne-like ion beams for Au, Pb, Bi, and U have almost identical angular fluence distributions and ion pulse shapes, as well as peak fluencies and intensities. However, the dependence of the parameters of the Ne-like ion beam on the laser intensity is different for ions with different masses. In the considered laser intensity range, the heaviest ions (U ions) ensured the achievement of the highest beam parameters, such as the mean and maximum ion energy, the ion beam energy, or the laser-to-ions energy conversion efficiency. The mono-charge superheavy ion beams demonstrated in this paper open the prospect for novel applications of heavy ions in high energy-density physics, nuclear physics, and possibly in other fields.

Heterogeneous Domain Adaptation Framework for Logging Lithofacies Identification

Summary Reservoir lithofacies type is an important indicator of reservoir quality and oiliness, and understanding lithofacies type can help geologists and engineers make informed decisions about exploration and development activities. The use of well-log data to establish machine learning models for lithofacies identification has gained popularity; however, the assumption that data are independent identical distribution followed by these models is often unrealistic. Additionally, there is a possible incompatibility between the training and test data in terms of feature space dimensions. We propose the heterogeneous domain adaptation framework for logging lithofacies identification (HDAFLI) to address these problems. The framework comprises three main contributions: (i) The denoising autoencoder feature mapping (DAFM) module is adopted to resolve the incompatibility issue in feature space between training and test data. The connection between training and test data can be effectively established to improve the performance and generalization ability. (ii) The transferability and discriminative joint probability distribution adaptive (TDJPDA) module addresses the issue of data distribution differences. It improves the transferability of training and test data by minimizing the maximum mean difference (MMD) of the joint probabilities of the source and target domains and enhances their discriminative ability by maximizing the joint probability MMD of different lithofacies categories. (iii) Bayesian optimization is used to optimize hyperparameters in the light gradient boosting machine (LightGBM) model for high computational efficiency in determining the best accuracy. We selected well-logging data from eight wells in the Pearl River Mouth Basin of the South China Sea to design four tasks and compared HDAFLI with various baseline machine learning algorithms and baseline domain adaptive algorithms. The results show that HDAFLI has the highest average accuracy among the four tasks. It is 19.76% and 8.94% higher than the best-performing baseline machine learning algorithm and baseline domain adaptive method among the comparison algorithms, respectively. For HDAFLI, we also conducted ablation experiments, time cost and convergence performance analysis, parameter sensitivity experiments, and feature visualization experiments. The results of ablation experiments show that the three modules of HDAFLI all play an active role, working together to achieve the best results. In addition, HDAFLI has a reasonable time cost, can become stable after several iterations, and has good convergence performance. The results of parameter sensitivity experiments confirm that the accuracy of HDAFLI does not change significantly with changes in hyperparameters, which is robust. The results of feature visualization experiments show that the data of the training set and the test set are concentrated together to a certain extent, which indicates that HDAFLI has completed the task of data distribution alignment very well. The findings of this study can help for a better understanding of how to address the challenge of reservoir lithofacies identification through a heterogeneous domain adaptation framework. By solving the problem of feature space incompatibility and data distribution difference between training data and test data, the application of HDAFLI provides geologists and engineers with more accurate lithofacies classification tools. This study has practical application value for reservoir quality assessment, oiliness prediction, and exploration and development decision-making.

Towards Robust Learning with Noisy and Pseudo Labels for Text Classification

Unlike Positive Training (PT), Negative Training (NT) is an indirect learning technique that trains the model on a combination of clean and noisy data using complementary labels, which are randomly generated from the label space except for the actual label. Although clean samples have identical distributions to the test samples, they are treated with the same level of uncertainty as noisy samples because of the complementary labeling of NT. Consequently, their contribution to the overall performance is relatively lower. We propose a Learning with Noisy and Pseudo Label (LNPL) framework, which jointly trains the model using PT and NT on clean and noisy data, respectively. We aim to enable direct learning on clean samples while leveraging the robustness of NT against noise in a unified framework. To mitigate the abundance of noisy instances, we leverage a gradient reversal layer at the top of LNPL as a regularization term to mislead the recognition of the source of the instance (e.g., clean or noisy). Moreover, we introduce a self-training LNPL that performs a semi-supervised text classification task as a learning with noisy pseudo-label problem. Extensive experiments on various textual benchmark datasets demonstrate that LNPL is robust and consistently outperforms the alternatives. The code is available on GitHub.1

Comparative genomic analyses of Cutibacterium granulosum provide insights into genomic diversity.

Cutibacterium granulosum, a commensal bacterium found on human skin, formerly known as Propionibacterium granulosum, rarely causes infections and is generally considered non-pathogenic. Recent research has revealed the transferability of the multidrug-resistant plasmid pTZC1 between C. granulosum and Cutibacterium acnes, the latter being an opportunistic pathogen in surgical site infections. However, there is a noticeable lack of research on the genome of C. granulosum, and the genetic landscape of this species remains largely uncharted. We investigated the genomic features and evolutionary structure of C. granulosum by analyzing a total of 30 Metagenome-Assembled Genomes (MAGs) and isolate genomes retrieved from public databases, as well as those generated in this study. A pan-genome of 6,077 genes was identified for C. granulosum. Remarkably, the 'cloud genes' constituted 62.38% of the pan-genome. Genes associated with mobilome: prophages, transposons [X], defense mechanisms [V] and replication, recombination and repair [L] were enriched in the cloud genome. Phylogenomic analysis revealed two distinct mono-clades, highlighting the genomic diversity of C. granulosum. The genomic diversity was further confirmed by the distribution of Average Nucleotide Identity (ANI) values. The functional profiles analysis of C. granulosum unveiled a wide range of potential Antibiotic Resistance Genes (ARGs) and virulence factors, suggesting its potential tolerance to various environmental challenges. Subtype I-E of the CRISPR-Cas system was the most abundant in these genomes, a feature also detected in C. acnes genomes. Given the widespread distribution of C. granulosum strains within skin microbiome, our findings make a substantial contribution to our broader understanding of the genetic diversity, which may open new avenues for investigating the mechanisms and treatment of conditions such as acne vulgaris.

Towards estimating the number of strains that make up a natural bacterial population

What a strain is and how many strains make up a natural bacterial population remain elusive concepts despite their apparent importance for assessing the role of intra-population diversity in disease emergence or response to environmental perturbations. To advance these concepts, we sequenced 138 randomly selected Salinibacter ruber isolates from two solar salterns and assessed these genomes against companion short-read metagenomes from the same samples. The distribution of genome-aggregate average nucleotide identity (ANI) values among these isolates revealed a bimodal distribution, with four-fold lower occurrence of values between 99.2% and 99.8% relative to ANI >99.8% or <99.2%, revealing a natural “gap” in the sequence space within species. Accordingly, we used this ANI gap to define genomovars and a higher ANI value of >99.99% and shared gene-content >99.0% to define strains. Using these thresholds and extrapolating from how many metagenomic reads each genomovar uniquely recruited, we estimated that –although our 138 isolates represented about 80% of the Sal. ruber population– the total population in one saltern pond is composed of 5,500 to 11,000 genomovars, the great majority of which appear to be rare in-situ. These data also revealed that the most frequently recovered isolate in lab media was often not the most abundant genomovar in-situ, suggesting that cultivation biases are significant, even in cases that cultivation procedures are thought to be robust. The methodology and ANI thresholds outlined here should represent a useful guide for future microdiversity surveys of additional microbial species.

Improved estimation of extreme floods with data pooling and mixed probability distribution

The accurate estimation of flood probability is crucial for designing water storage and flood retention structures. However, the assumption of identical distribution in flood samples is unrealistic, given the influence of various flood mechanisms. To address this challenge, we proposed a novel framework based on flood clustering and data pooling that encompasses the key steps such as 1) flood event separation based on a peak-detection flood separation algorithm, 2) grouping flood events using the k-prototypes algorithm, 3) application of the UNprecedented Simulated Extreme ENsemble (UNSEEN) approach to pool reforecast ensemble datasets, and 4) statistical mixing approach to derive common quantiles from all the flood groups. We applied the framework to the Dresden gauge in the Elbe River for a detailed case study. Various tests have been performed to assess the applicability of the UNSEEN approach and the reforecast dataset consistently shows the potential for data pooling. The proposed methodology outperformed the classical approach in terms of goodness-of-fit. The relative difference between the classical and the proposed approach ((classical-proposed)/proposed) for the 100-year return level is 0.16, with a reduction in root mean square error (RMSE) value from 163 to 98 m3/s. Further, replication of the approach to the gauges in North Germany exhibited a relative difference ranging from −0.3 to +0.15 and produced better estimates in terms of RMSE compared with the traditional model. In summary, the proposed framework offers a better estimation of flood probability by addressing the inherent sample inhomogeneity along with the inclusion of unprecedented flood samples.

A Gaussian Mixture CPHD Filter for Multi-Target Tracking in Target-Dependent False Alarms

The estimation of the target number and individual tracks are two major tasks in multi-target tracking. The main shortcoming of traditional tracking methods is the cumbersome data association between measurements and targets. The cardinalized probability hypothesis density filter (CPHD) proposed in recent years can achieve the requirement for multitarget tracking. This kind of filter jointly estimates the cardinality distribution and the posterior density, which can achieve a more stable estimate of the target number. However, targets with complex micro-Doppler signatures (drones, birds, etc.) may generate target-dependent false alarms, which is contrary to the traditional uniform distribution assumption. In this case, the estimates of traditional CPHD filter will suffer from the abnormal transfer of PHD mass, causing the degradation of filtering performance. This paper studies the individual tracking of group targets with an improved GM-CPHD filter. First, the target-dependent false alarms are modeled with a general independent and identically distributed (I.I.D.) cluster process. Second, the update equations of cardinality and PHD density in target-dependent false alarms are derived. Finally, a practical solution using the Gaussian mixture method is proposed. The effectiveness of the proposed filter is verified by the simulation and experimental results.

Federated learning on non-IID and globally long-tailed data via meta re-weighting networks

Under a federated learning environment, the training samples are generally collected and stored locally on each client’s device, which makes the machine learning procedure not meet the requirement of independent and identical distribution (IID). Existing federated learning methods to deal with non-IID data generally assume that the data is globally balanced. However, real-world multi-class data tend to exhibit long-tail distribution, where the majority of samples are in a few head classes and a large number of tail classes only have a small amount of data. This paper, therefore, focuses on addressing the problem of handling non-IID and globally long-tailed data in a federated learning scenario. Accordingly, we propose a new federated learning method called Federated meta re-weighting networks (FedReN), which assigns weights during the local training process from the class-level and instance-level perspectives, respectively. To deal with data non-IIDness and global long-tail, both of the two re-weighting functions are globally trained by the meta-learning approach to acquire the knowledge of global long-tail distribution. Experiments on several long-tailed image classification benchmarks show that FedReN outperforms the state-of-the-art federated learning methods. The code is available at https://github.com/pxqian/FedReN .