RESUMO
OBJECT: Transformer-based neural networks have been applied to the electroencephalography (EEG) decoding for motor imagery (MI). However, most networks focus on applying the self-attention mechanism to extract global temporal information, while the cross-frequency coupling features between different frequencies have been neglected. Additionally, effectively integrating different neural networks poses challenges for the advanced design of decoding algorithms. METHODS: This study proposes a novel end-to-end Multi-Scale Vision Transformer Neural Network (MSVTNet) for MI-EEG classification. MSVTNet first extracts local spatio-temporal features at different filtered scales through convolutional neural networks (CNNs). Then, these features are concatenated along the feature dimension to form local multi-scale spatio-temporal feature tokens. Finally, Transformers are utilized to capture cross-scale interaction information and global temporal correlations, providing more distinguishable feature embeddings for classification. Moreover, auxiliary branch loss is leveraged for intermediate supervision to ensure the effective integration of CNNs and Transformers. RESULTS: The performance of MSVTNet was assessed through subject-dependent (session-dependent and session-independent) and subject-independent experiments on three MI datasets, i.e., the BCI competition IV 2a, 2b and OpenBMI datasets. The experimental results demonstrate that MSVTNet achieves state-of-the-art performance in all analyses. CONCLUSION: MSVTNet shows superiority and robustness in enhancing MI decoding performance. The source code for MSVTNet is available at https://github.com/SheepTAO/MSVTNet.
RESUMO
Granular-ball support vector machine (GBSVM) is a significant attempt to construct a classifier using the coarse-to-fine granularity of a granular ball as input, rather than a single data point. It is the first classifier whose input contains no points. However, the existing model has some errors, and its dual model has not been derived. As a result, the current algorithm cannot be implemented or applied. To address these problems, we fix the errors of the original model of the existing GBSVM and derive its dual model. Furthermore, a particle swarm optimization (PSO) algorithm is designed to solve the dual problem. The sequential minimal optimization (SMO) algorithm is also carefully designed to solve the dual problem. The latter is faster and more stable. The experimental results on the UCI benchmark datasets demonstrate that GBSVM is more robust and efficient. All codes have been released in the open source library available at: http://www.cquptshuyinxia.com/GBSVM.html or https://github.com/syxiaa/GBSVM.
RESUMO
In the real world, data distributions often exhibit multiple granularities. However, the majority of existing neighbor-based machine-learning methods rely on manually setting a single-granularity for neighbor relationships. These methods typically handle each data point using a single-granularity approach, which severely affects their accuracy and efficiency. This paper adopts a dual-pronged approach: it constructs a multi-granularity representation of the data using the granular-ball computing model, thereby boosting the algorithm's time efficiency. It leverages the multi-granularity representation of the data to create tailored, multi-granularity neighborhood relationships for different task scenarios, resulting in improved algorithmic accuracy. The experimental results convincingly demonstrate that the proposed multi-granularity neighbor relationship effectively enhances KNN classification and clustering methods. The source code has been publicly released and is now accessible on GitHub at https://github.com/xjnine/MGNR.
RESUMO
Due to the great successes of Graph Neural Networks (GNN) in numerous fields, growing research interests have been devoted to applying GNN to molecular learning tasks. The molecule structure can be naturally represented as graphs where atoms and bonds refer to nodes and edges respectively. However, the atoms are not haphazardly stacked together but combined into various spatial geometries. Meanwhile, since chemical reactions mainly occur in substructures such as functional groups, the substructure plays a decisive role in the molecule's properties. Therefore, directly applying GNN to molecular representation learning could ignore the molecular spatial structure and the substructure properties which in turn degrades the performance of downstream tasks. In this paper, we propose Knowledge-Driven Self-Supervised Model for Molecular Representation Learning (KSMRL) to address above problems. The KSMRL consists of two major pathways: (1) the Spatial Information (SI) based pathway which preserves the spatial information of molecular structure, (2) the Subgraph Constraint (SC) based pathway which retains the properties of substructures into the molecular representation. In this manner, both the atomic level and substructure level information can be included in modeling. According to the experimental results on multiple datasets, the proposed KSMRL can generate discriminative molecular representations. In molecular generation tasks, KSMRL combined with Autoregressive Flow (AF) models or Discrete Flow (DF) models outperforms the state-of-the-art baselines over all datasets. In addition, we demonstrate the effectiveness of KSMRL with property optimization experiments. To indicate the ability of predicting specified potential Drug-Target Interactions (DTIs), a case study for discriminating the interactions between molecule generated by KSMRL and targets is also given.
RESUMO
Electroencephalographic (EEG) source imaging (ESI) is a powerful method for studying brain functions and surgical resection of epileptic foci. However, accurately estimating the location and extent of brain sources remains challenging due to noise and background interference in EEG signals. To reconstruct extended brain sources, we propose a new ESI method called Variation Sparse Source Imaging based on Generalized Gaussian Distribution (VSSI-GGD). VSSI-GGD uses the generalized Gaussian prior as a sparse constraint on the spatial variation domain and embeds it into the Bayesian framework for source estimation. Using a variational technique, we approximate the intractable true posterior with a Gaussian density. Through convex analysis, the Bayesian inference problem is transformed entirely into a series of regularized L2p -norm ( ) optimization problems, which are efficiently solved with the ADMM algorithm. Imaging results of numerical simulations and human experimental dataset analysis reveal the superior performance of VSSI-GGD, which provides higher spatial resolution with clear boundaries compared to benchmark algorithms. VSSI-GGD can potentially serve as an effective and robust spatiotemporal EEG source imaging method. The source code of VSSI-GGD is available at https://github.com/Mashirops/VSSI-GGD.git.
Assuntos
Encéfalo , Eletroencefalografia , Humanos , Teorema de Bayes , Distribuição Normal , Eletroencefalografia/métodos , Encéfalo/diagnóstico por imagem , Mapeamento Encefálico/métodos , Algoritmos , Magnetoencefalografia/métodosRESUMO
Deep learning has excelled in single-image super-resolution (SISR) applications, yet the lack of interpretability in most deep learning-based SR networks hinders their applicability, especially in fields like medical imaging that require transparent computation. To address these problems, we present an interpretable frequency division SR network that operates in the image frequency domain. It comprises a frequency division module and a step-wise reconstruction method, which divides the image into different frequencies and performs reconstruction accordingly. We develop a frequency division loss function to ensure that each reconstruction module (ReM) operates solely at one image frequency. These methods establish an interpretable framework for SR networks, visualizing the image reconstruction process and reducing the black box nature of SR networks. Additionally, we revisited the subpixel layer upsampling process by deriving its inverse process and designing a displacement generation module. This interpretable upsampling process incorporates subpixel information and is similar to pre-upsampling frameworks. Furthermore, we develop a new ReM based on interpretable Hessian attention to enhance network performance. Extensive experiments demonstrate that our network, without the frequency division loss, outperforms state-of-the-art methods qualitatively and quantitatively. The inclusion of the frequency division loss enhances the network's interpretability and robustness, and only slightly decreases the PSNR and SSIM metrics by an average of 0.48 dB and 0.0049, respectively.
RESUMO
Deep generative models have advantages in modeling complex time series and are widely used in anomaly detection. Nevertheless, the existing deep generative approaches mainly concentrate on the investigation of models' reconstruction capability rather than customizing a model suitable for anomaly detection. Meanwhile, VAE-based models suffer from posterior collapse, which can lead to a series of undesirable consequences, such as high false positive rate etc. Based on these considerations, in this paper, we propose a novel self-adversarial variational auto-encoder combined with contrast learning, short for ACVAE, to address these challenges. ACVAE consist of three parts ãT,E,Gã, wherein the transformation network T is employed to generate abnormal latent representations similar to those normal latent representations encoded by the encoder E, and the decoder G is used to distinguish the two representations. In the framework of this model, the normal reconstructions are considered as positive samples and abnormal reconstructions as negative samples, and the contrast learning is executed on the part E to measure the similarities between inputs and positive samples, dissimilarities between inputs and negative samples. Thus, an improved objective function is proposed by integrating two novel regularizers, one refers to adversarial mechanism and the other involves contrast learning, in which the encoder E and decoder G hold the capability to distinguish, and decoder G is constrained to mitigate the posterior collapse. We perform several experiments on five datasets, whose results show ACVAE outperforms state-of-the-art methods.
Assuntos
Aprendizagem , Fatores de TempoRESUMO
Pawlak rough set (PRS) and neighborhood rough set (NRS) are the two most common rough set theoretical models. Although the PRS can use equivalence classes to represent knowledge, it is unable to process continuous data. On the other hand, NRSs, which can process continuous data, rather lose the ability of using equivalence classes to represent knowledge. To remedy this deficit, this article presents a granular-ball rough set (GBRS) based on the granular-ball computing combining the robustness and the adaptability of the granular-ball computing. The GBRS can simultaneously represent both the PRS and the NRS, enabling it not only to be able to deal with continuous data and to use equivalence classes for knowledge representation as well. In addition, we propose an implementation algorithm of the GBRS by introducing the positive region of GBRS into the PRS framework. The experimental results on benchmark datasets demonstrate that the learning accuracy of the GBRS has been significantly improved compared with the PRS and the traditional NRS. The GBRS also outperforms nine popular or the state-of-the-art feature selection methods. We have open-sourced all the source codes of this article at http://www.cquptshuyinxia.com/GBRS.html, https://github.com/syxiaa/GBRS.
RESUMO
Density peaks clustering algorithm (DP) has difficulty in clustering large-scale data, because it requires the distance matrix to compute the density and δ -distance for each object, which has O(n2) time complexity. Granular ball (GB) is a coarse-grained representation of data. It is based on the fact that an object and its local neighbors have similar distribution and they have high possibility of belonging to the same class. It has been introduced into supervised learning by Xia et al. to improve the efficiency of supervised learning, such as support vector machine, k -nearest neighbor classification, rough set, etc. Inspired by the idea of GB, we introduce it into unsupervised learning for the first time and propose a GB-based DP algorithm, called GB-DP. First, it generates GBs from the original data with an unsupervised partitioning method. Then, it defines the density of GBs, instead of the density of objects, according to the centers, radius, and distances between its members and centers, without setting any parameters. After that, it computes the distance between the centers of GBs as the distance between GBs and defines the δ -distance of GBs. Finally, it uses GBs' density and δ -distance to plot the decision graph, employs DP algorithm to cluster them, and expands the clustering result to the original data. Since there is no need to calculate the distance between any two objects and the number of GBs is far less than the scale of a data, it greatly reduces the running time of DP algorithm. By comparing with k -means, ball k -means, DP, DPC-KNN-PCA, FastDPeak, and DLORE-DP, GB-DP can get similar or even better clustering results in much less running time without setting any parameters. The source code is available at https://github.com/DongdongCheng/GB-DP.
RESUMO
Due to simplicity, K-means has become a widely used clustering method. However, its clustering result is seriously affected by the initial centers and the allocation strategy makes it hard to identify manifold clusters. Many improved K-means are proposed to accelerate it and improve the quality of initialize cluster centers, but few researchers pay attention to the shortcoming of K-means in discovering arbitrary-shaped clusters. Using graph distance (GD) to measure the dissimilarity between objects is a good way to solve this problem, but computing the GD is time-consuming. Inspired by the idea that granular ball uses a ball to represent the local data, we select representatives from a local neighborhood, called natural density peaks (NDPs). On the basis of NDPs, we propose a novel K-means algorithm for identifying arbitrary-shaped clusters, called NDP-Kmeans. It defines neighbor-based distance between NDPs and takes advantage of the neighbor-based distance to compute the GD between NDPs. Afterward, an improved K-means with high-quality initial centers and GD is used to cluster NDPs. Finally, each remaining object is assigned according to its representative. The experimental results show that our algorithms can not only recognize spherical clusters but also manifold clusters. Therefore, NDP-Kmeans has more advantages in detecting arbitrary-shaped clusters than other excellent algorithms.
RESUMO
Hierarchical quotient space structure (HQSS), as a typical description of granular computing (GrC), focuses on hierarchically granulating fuzzy data and mining hidden knowledge. The key step of constructing HQSS is to transform the fuzzy similarity relation into fuzzy equivalence relation. However, on one hand, the transformation process has high time complexity. On the other hand, it is difficult to mine knowledge directly from fuzzy similarity relation due to its information redundancy, i.e., sparsity of effective information. Therefore, this article mainly focuses on proposing an efficient granulation approach for constructing HQSS by quickly extracting the effective value of fuzzy similarity relation. First, the effective value and effective position of fuzzy similarity relation are defined according to whether they could be retained in fuzzy equivalence relation. Second, the number and composition of effective values are presented to confirm that which elements are effective values. Based on these above theories, redundant information and sparse effective information in fuzzy similarity relation could be completely distinguished. Next, both isomorphism and similarity between two fuzzy similarity relations are researched based on the effective value. The isomorphism between two fuzzy equivalence relations is discussed based on the effective value. Then, the algorithm with low time complexity for extracting effective values of fuzzy similarity relation is introduced. On the basis, the algorithm for constructing HQSS is presented to realize efficient granulation of fuzzy data. The proposed algorithms could accurately extract effective information from the fuzzy similarity relation and construct the same HQSS with the fuzzy equivalence relation while greatly reducing the time complexity. Finally, relevant experiments on 15 UCI datasets, 3 UKB datasets, and 5 image datasets are shown and analyzed to verify the effectiveness and efficiency of the proposed algorithm.
RESUMO
Hashing methods have sparked a great revolution in cross-modal retrieval due to the low cost of storage and computation. Benefiting from the sufficient semantic information of labeled data, supervised hashing methods have shown better performance compared with unsupervised ones. Nevertheless, it is expensive and labor intensive to annotate the training samples, which restricts the feasibility of supervised methods in real applications. To deal with this limitation, a novel semisupervised hashing method, i.e., three-stage semisupervised hashing (TS3H) is proposed in this article, where both labeled and unlabeled data are seamlessly handled. Different from other semisupervised approaches that learn the pseudolabels, hash codes, and hash functions simultaneously, the new approach is decomposed into three stages as the name implies, in which all of the stages are conducted individually to make the optimization cost-effective and precise. Specifically, the classifiers of different modalities are learned via the provided supervised information to predict the labels of unlabeled data at first. Then, hash code learning is achieved with a simple but efficient scheme by unifying the provided and the newly predicted labels. To capture the discriminative information and preserve the semantic similarities, we leverage pairwise relations to supervise both classifier learning and hash code learning. Finally, the modality-specific hash functions are obtained by transforming the training samples to the generated hash codes. The new approach is compared with the state-of-the-art shallow and deep cross-modal hashing (DCMH) methods on several widely used benchmark databases, and the experiment results verify its efficiency and superiority.
RESUMO
Understanding the genetic basis of the node of the first fruiting branch (NFFB) improves early-maturity cotton breeding. Here we report QTL mapping on 200 F2 plants and derivative F2:3 and F2:4 populations by genotyping by sequencing (GBS). BC1F2 population was constructed by backcrossing one F2:4 line with the maternal parent JF914 and used for BSA-seq for further QTL mapping. A total of 1,305,642 SNPs were developed between the parents by GBS, and 2,907,790 SNPs were detected by BSA-seq. A high-density genetic map was constructed containing 11,488 SNPs and spanning 4,202.12 cM in length. A total of 13 QTL were mapped in the 3 tested populations. JF914 conferred favorable alleles for 11 QTL, and JF173 conferred favorable alleles for the other 2 QTL. Two stable QTL were repeatedly mapped in F2:3 and F2:4, including qNFFB-D3-1 and qNFFB-D6-1. Only qNFFB-D3-1 contributed more than 10% of the phenotypic variation. This QTL covered about 24.7 Mb (17,130,008-41,839,226 bp) on chromosome D3. Two regions on D3 (41,779,195-41,836,120 bp, 41,836,768-41,872,287 bp) were found by BSA-seq and covered about 92.4 Kb. This 92.4 Kb region overlapped with the stable QTL qNFFB-D3-1 and contained 8 annotated genes. By qRT-PCR, Ghir_D03G012430 showed a lower expression level from the 1- to 2-leaf stage and a higher expression level from the 3- to 6-leaf stage in the buds of JF173 than that of JF914. Ghir_D03G012390 reached the highest level at the 3- and 5-leaf stages in the buds of JF173 and JF914, respectively. As JF173 has lower NFFB and more early maturity than JF914, these two genes might be important in cell division and differentiation during NFFB formation in the seedling stage. The results of this study will facilitate a better understanding of the genetic basis of NFFB and benefit cotton molecular breeding for improving earliness traits.
RESUMO
This article presents a general sampling method, called granular-ball sampling (GBS), for classification problems by introducing the idea of granular computing. The GBS method uses some adaptively generated hyperballs to cover the data space, and the points on the hyperballs constitute the sampled data. GBS is the first sampling method that not only reduces the data size but also improves the data quality in noisy label classification. In addition, because the GBS method can be used to exactly describe the boundary, it can obtain almost the same classification accuracy as the results on the original datasets, and it can obtain an obviously higher classification accuracy than random sampling. Therefore, for the data reduction classification task, GBS is a general method that is not especially restricted by any specific classifier or dataset. Moreover, the GBS can be effectively used as an undersampling method for imbalanced classification. It has a time complexity that is close to O( N ), so it can accelerate most classifiers. These advantages make GBS powerful for improving the performance of classifiers. All codes have been released in the open source GBS library at http://www.cquptshuyinxia.com/GBS.html.
RESUMO
OBJECT: Motor imagery (MI) is a mental process widely utilized as the experimental paradigm for brain-computer interfaces (BCIs) across a broad range of basic science and clinical studies. However, decoding intentions from MI remains challenging due to the inherent complexity of brain patterns relative to the small sample size available for machine learning. APPROACH: This paper proposes an end-to-end Filter-Bank Multiscale Convolutional Neural Network (FBMSNet) for MI classification. A filter bank is first employed to derive a multiview spectral representation of the EEG data. Mixed depthwise convolution is then applied to extract temporal features at multiple scales, followed by spatial filtering to mitigate volume conduction. Finally, with the joint supervision of cross-entropy and center loss, FBMSNet obtains features that maximize interclass dispersion and intraclass compactness. MAIN RESULTS: We compare FBMSNet with several state-of-the-art EEG decoding methods on two MI datasets: the BCI Competition IV 2a dataset and the OpenBMI dataset. FBMSNet significantly outperforms the benchmark methods by achieving 79.17% and 70.05% for four-class and two-class hold-out classification accuracy, respectively. SIGNIFICANCE: These results demonstrate the efficacy of FBMSNet in improving EEG decoding performance toward more robust BCI applications. The FBMSNet source code is available at https://github.com/Want2Vanish/FBMSNet.
Assuntos
Interfaces Cérebro-Computador , Imaginação , Redes Neurais de Computação , Aprendizado de Máquina , Encéfalo , Eletroencefalografia/métodos , AlgoritmosRESUMO
Granular-ball computing (GBC) is an efficient, robust, and scalable learning method for granular computing. The granular ball (GB) generation method is based on GB computing. This article proposes a method for accelerating GB generation using division to replace k -means. It can significantly improve the efficiency of GB generation while ensuring an accuracy similar to that of the existing methods. In addition, a new adaptive method for GB generation is proposed by considering the elimination of the GB overlap and other factors. This makes the GB generation process parameter-free and completely adaptive in the true sense. In addition, this study first provides mathematical models for the GB covering. The experimental results on some real datasets demonstrate that the two proposed GB generation methods have accuracies similar to those of the existing method in most cases, while adaptiveness or acceleration is realized. All the codes were released in the open-source GBC library at http://www.cquptshuyinxia.com/GBC.html or https://github.com/syxiaa/gbc.
RESUMO
BACKGROUND: Coronavirus disease 2019 (COVID-19) has spurred a boom in uncovering repurposable existing drugs. Drug repurposing is a strategy for identifying new uses for approved or investigational drugs that are outside the scope of the original medical indication. MOTIVATION: Current works of drug repurposing for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) are mostly limited to only focusing on chemical medicines, analysis of single drug targeting single SARS-CoV-2 protein, one-size-fits-all strategy using the same treatment (same drug) for different infected stages of SARS-CoV-2. To dilute these issues, we initially set the research focusing on herbal medicines. We then proposed a heterogeneous graph embedding method to signaled candidate repurposing herbs for each SARS-CoV-2 protein, and employed the variational graph convolutional network approach to recommend the precision herb combinations as the potential candidate treatments against the specific infected stage. METHOD: We initially employed the virtual screening method to construct the 'Herb-Compound' and 'Compound-Protein' docking graph based on 480 herbal medicines, 12,735 associated chemical compounds and 24 SARS-CoV-2 proteins. Sequentially, the 'Herb-Compound-Protein' heterogeneous network was constructed by means of the metapath-based embedding approach. We then proposed the heterogeneous-information-network-based graph embedding method to generate the candidate ranking lists of herbs that target structural, nonstructural and accessory SARS-CoV-2 proteins, individually. To obtain precision synthetic effective treatments forvarious COVID-19 infected stages, we employed the variational graph convolutional network method to generate candidate herb combinations as the recommended therapeutic therapies. RESULTS: There were 24 ranking lists, each containing top-10 herbs, targeting 24 SARS-CoV-2 proteins correspondingly, and 20 herb combinations were generated as the candidate-specific treatment to target the four infected stages. The code and supplementary materials are freely available at https://github.com/fanyang-AI/TCM-COVID19.
Assuntos
Tratamento Farmacológico da COVID-19 , Combinação de Medicamentos , Reposicionamento de Medicamentos/métodos , Drogas em Investigação , Humanos , SARS-CoV-2RESUMO
Mitigating label noise is a crucial problem in classification. Noise filtering is an effective method of dealing with label noise which does not need to estimate the noise rate or rely on any loss function. However, most filtering methods focus mainly on binary classification, leaving the more difficult counterpart problem of multiclass classification relatively unexplored. To remedy this deficit, we present a definition for label noise in a multiclass setting and propose a general framework for a novel label noise filtering learning method for multiclass classification. Two examples of noise filtering methods for multiclass classification, multiclass complete random forest (mCRF) and multiclass relative density, are derived from their binary counterparts using our proposed framework. In addition, to optimize the NI_threshold hyperparameter in mCRF, we propose two new optimization methods: a new voting cross-validation method and an adaptive method that employs a 2-means clustering algorithm. Furthermore, we incorporate SMOTE into our label noise filtering learning framework to handle the ubiquitous problem of imbalanced data in multiclass classification. We report experiments on both synthetic data sets and UCI benchmarks to demonstrate our proposed methods are highly robust to label noise in comparison with state-of-the-art baselines. All code and data results are available at https://github.com/syxiaa/Multiclass-Label-Noise-Filtering-Learning.
RESUMO
Multi-scale decision system (MDS) is an effective tool to describe hierarchical data in machine learning. Optimal scale combination (OSC) selection and attribute reduction are two key issues related to knowledge discovery in MDSs. However, searching for all OSCs may result in a combinatorial explosion, and the existing approaches typically incur excessive time consumption. In this study, searching for all OSCs is considered as an optimization problem with the scale space as the search space. Accordingly, a sequential three-way decision model of the scale space is established to reduce the search space by integrating three-way decision with the Hasse diagram. First, a novel scale combination is proposed to perform scale selection and attribute reduction simultaneously, and then an extended stepwise optimal scale selection (ESOSS) method is introduced to quickly search for a single local OSC on a subset of the scale space. Second, based on the obtained local OSCs, a sequential three-way decision model of the scale space is established to divide the search space into three pair-wise disjoint regions, namely the positive, negative, and boundary regions. The boundary region is regarded as a new search space, and it can be proved that a local OSC on the boundary region is also a global OSC. Therefore, all OSCs of a given MDS can be obtained by searching for the local OSCs on the boundary regions in a step-by-step manner. Finally, according to the properties of the Hasse diagram, a formula for calculating the maximal elements of a given boundary region is provided to alleviate space complexity. Accordingly, an efficient OSC selection algorithm is proposed to improve the efficiency of searching for all OSCs by reducing the search space. The experimental results demonstrate that the proposed method can significantly reduce computational time.
RESUMO
This paper presents a novel accelerated exact k-means called as "Ball k-means" by using the ball to describe each cluster, which focus on reducing the point-centroid distance computation. The "Ball k-means" can exactly find its neighbor clusters for each cluster, resulting distance computations only between a point and its neighbor clusters' centroids instead of all centroids. What's more, each cluster can be divided into "stable area" and "active area", and the latter one is further divided into some exact "annular area". The assignment of the points in the "stable area" is not changed while the points in each "annular area" will be adjusted within a few neighbor clusters. There are no upper or lower bounds in the whole process. Moreover, ball k-means uses ball clusters and neighbor searching along with multiple novel stratagems for reducing centroid distance computations. In comparison with the current state-of-the art accelerated exact bounded methods, the Yinyang algorithm and the Exponion algorithm, as well as other top-of-the-line tree-based and bounded methods, the ball k-means attains both higher performance and performs fewer distance calculations, especially for large-k problems. The faster speed, no extra parameters and simpler design of "Ball k-means" make it an all-around replacement of the naive k-means.