A single-cell atlas identifies pretreatment features of primary imatinib resistance in chronic myeloid leukemia.

Primary resistance to tyrosine kinase inhibitors (TKIs) is a significant barrier to optimal outcomes in chronic myeloid leukemia (CML), but factors contributing to response heterogeneity remain unclear. Using single-cell RNA (scRNA) sequencing, we identified 8 statistically significant features in pretreatment bone marrow, which correlated with either sensitivity (major molecular response or MMR) or extreme resistance to imatinib (eventual blast crisis [BC] transformation). Employing machine-learning, we identified leukemic stem cell (LSC) and natural killer (NK) cell gene expression profiles predicting imatinib response with >80% accuracy, including no false positives for predicting BC. A canonical erythroid-specifying (TAL1/KLF1/GATA1) regulon was a hallmark of LSCs from patients with MMR and was associated with erythroid progenitor [ERP] expansion in vivo (P < .05), and a 2- to 10-fold (6.3-fold in group A vs 1.09-fold in group C) erythroid over myeloid bias in vitro. Notably, ERPs demonstrated exquisite TKI sensitivity compared with myeloid progenitors (P < .001). These LSC features were lost with progressive resistance, and MYC- and IRF1-driven inflammatory regulons were evident in patients who progressed to transformation. Patients with MMR also exhibited a 56-fold expansion (P < .01) of a normally rare subset of hyperfunctional adaptive-like NK cells, which diminished with progressive resistance, whereas patients destined for BC accumulated inhibitory NKG2A+ NK cells favoring NK cell tolerance. Finally, we developed antibody panels to validate our scRNA-seq findings. These panels may be useful for prospective studies of primary resistance, and in assessing the contribution of predetermined vs acquired factors in TKI response heterogeneity.

[Chemical constituents from stems and leaves of Artocarpus tonkinensis and their inhibitory effects on proliferation of synovioblasts in vitro].

The chemical constituents from the stems and leaves of Artocarpus tonkinensis in Artocarpus of Moraceae were systematically studied by means of silica gel, octadecylsilyl(ODS), and Sephadex LH-20 gel column chromatographies, as well as preparative high-performance liquid chromatography(Pre-HPLC) and a variety of chromatographic separation techniques. The spectral data and physicochemical properties of the compounds were obtained from separation and compared with those of the compounds reported in the literature. As a result, 11 compounds isolated from the 90% ethanol extract of the stems and leaves of A. tonkinensis were identified as artocatonkine(1), 5,6,7,4'-tetramethoxyflavone(2), apigenin-4'-O-ß-D-glucoside(3), rayalinol(4), psorachalcone A(5), 4-ketopinoresinol(6), ficusesquilignan B(7), pinnatifidanin AI(8), pinnatifidanin A(9), O-methylmellein(10), and trans-4-hydroxymellein(11). Among these compounds, compound 1 was a new prenylated flavone, and compounds 2-11 were isolated from the plants belonging to the genus Artocarpus for the first time. Furthermore, all compounds 1-11 were evaluated for their anti-rheumatoid arthritis activities, and the MTS method was used to measure their inhibitory effects on the proliferation of synovioblasts in vitro. The results of activity evaluation showed that flavonoid compounds 1-3, 5, and lignan compounds 8 and 9 displayed significant anti-rheumatoid arthritis activities, showing the IC_(50) values in inhibiting the proliferation of synovioblasts MH7A from(6.38±0.06) µmol·L~(-1) to(168.58±0.28)µmol·L~(-1).

Tri-comb generation with a dual-ring structure.

By introducing a third measurement comb with different repetition frequencies (Δ f r e p ), the tri-comb spectroscopy technique overcomes the ambiguity problem of the original dual-comb spectroscopy technique and eliminates physical delay stages in multidimensional coherent spectroscopy. Nowadays, tri-comb generation based on three frequency-stabilized comb lasers is overly complicated and costly for many potential applications. Previous research on single-cavity dual-combs inspired research on single-cavity tri-combs. However, the currently reported tri-comb structures cannot achieve independently controllable pulses. This paper shows a dual-ring tri-comb seed-source structure using wavelength-based multiplexing in one of the rings. The wavelength and power of the output pulse are independently controlled by using the dual-ring structure. The Δ f r e p of wavelength multiplexing-based dual-comb output can be tuned by adjusting the intra-ring polarization controller (PC). In the case of single-wavelength mode-locking, the PC can be adjusted to achieve a wavelength tuning range of nearly 20 nm. The tri-comb source could offer an attractive alternative solution as a low-complexity light source for field-deployable multi-comb metrology applications.

Graph Clustering with High-Order Contrastive Learning.

Graph clustering is a fundamental and challenging task in unsupervised learning. It has achieved great progress due to contrastive learning. However, we find that there are two problems that need to be addressed: (1) The augmentations in most graph contrastive clustering methods are manual, which can result in semantic drift. (2) Contrastive learning is usually implemented on the feature level, ignoring the structure level, which can lead to sub-optimal performance. In this work, we propose a method termed Graph Clustering with High-Order Contrastive Learning (GCHCL) to solve these problems. First, we construct two views by Laplacian smoothing raw features with different normalizations and design a structure alignment loss to force these two views to be mapped into the same space. Second, we build a contrastive similarity matrix with two structure-based similarity matrices and force it to align with an identity matrix. In this way, our designed contrastive learning encompasses a larger neighborhood, enabling our model to learn clustering-friendly embeddings without the need for an extra clustering module. In addition, our model can be trained on a large dataset. Extensive experiments on five datasets validate the effectiveness of our model. For example, compared to the second-best baselines on four small and medium datasets, our model achieved an average improvement of 3% in accuracy. For the largest dataset, our model achieved an accuracy score of 81.92%, whereas the compared baselines encountered out-of-memory issues.

An integrative model of pathway convergence in genetically heterogeneous blast crisis chronic myeloid leukemia.

Targeted therapies against the BCR-ABL1 kinase have revolutionized treatment of chronic phase (CP) chronic myeloid leukemia (CML). In contrast, management of blast crisis (BC) CML remains challenging because BC cells acquire complex molecular alterations that confer stemness features to progenitor populations and resistance to BCR-ABL1 tyrosine kinase inhibitors. Comprehensive models of BC transformation have proved elusive because of the rarity and genetic heterogeneity of BC, but are important for developing biomarkers predicting BC progression and effective therapies. To better understand BC, we performed an integrated multiomics analysis of 74 CP and BC samples using whole-genome and exome sequencing, transcriptome and methylome profiling, and chromatin immunoprecipitation followed by high-throughput sequencing. Employing pathway-based analysis, we found the BC genome was significantly enriched for mutations affecting components of the polycomb repressive complex (PRC) pathway. While transcriptomically, BC progenitors were enriched and depleted for PRC1- and PRC2-related gene sets respectively. By integrating our data sets, we determined that BC progenitors undergo PRC-driven epigenetic reprogramming toward a convergent transcriptomic state. Specifically, PRC2 directs BC DNA hypermethylation, which in turn silences key genes involved in myeloid differentiation and tumor suppressor function via so-called epigenetic switching, whereas PRC1 represses an overlapping and distinct set of genes, including novel BC tumor suppressors. On the basis of these observations, we developed an integrated model of BC that facilitated the identification of combinatorial therapies capable of reversing BC reprogramming (decitabine+PRC1 inhibitors), novel PRC-silenced tumor suppressor genes (NR4A2), and gene expression signatures predictive of disease progression and drug resistance in CP.

Auxiliary Graph for Attribute Graph Clustering.

Attribute graph clustering algorithms that include topological structural information into node characteristics for building robust representations have proven to have promising efficacy in a variety of applications. However, the presented topological structure emphasizes local links between linked nodes but fails to convey relationships between nodes that are not directly linked, limiting the potential for future clustering performance improvement. To solve this issue, we offer the Auxiliary Graph for Attribute Graph Clustering technique (AGAGC). Specifically, we construct an additional graph as a supervisor based on the node attribute. The additional graph can serve as an auxiliary supervisor that aids the present one. To generate a trustworthy auxiliary graph, we offer a noise-filtering approach. Under the supervision of both the pre-defined graph and an auxiliary graph, a more effective clustering model is trained. Additionally, the embeddings of multiple layers are merged to improve the discriminative power of representations. We offer a clustering module for a self-supervisor to make the learned representation more clustering-aware. Finally, our model is trained using a triplet loss. Experiments are done on four available benchmark datasets, and the findings demonstrate that the proposed model outperforms or is comparable to state-of-the-art graph clustering models.

Adaptive Weighted Graph Fusion Incomplete Multi-View Subspace Clustering.

With the enormous amount of multi-source data produced by various sensors and feature extraction approaches, multi-view clustering (MVC) has attracted developing research attention and is widely exploited in data analysis. Most of the existing multi-view clustering methods hold on the assumption that all of the views are complete. However, in many real scenarios, multi-view data are often incomplete for many reasons, e.g., hardware failure or incomplete data collection. In this paper, we propose an adaptive weighted graph fusion incomplete multi-view subspace clustering (AWGF-IMSC) method to solve the incomplete multi-view clustering problem. Firstly, to eliminate the noise existing in the original space, we transform complete original data into latent representations which contribute to better graph construction for each view. Then, we incorporate feature extraction and incomplete graph fusion into a unified framework, whereas two processes can negotiate with each other, serving for graph learning tasks. A sparse regularization is imposed on the complete graph to make it more robust to the view-inconsistency. Besides, the importance of different views is automatically learned, further guiding the construction of the complete graph. An effective iterative algorithm is proposed to solve the resulting optimization problem with convergence. Compared with the existing state-of-the-art methods, the experiment results on several real-world datasets demonstrate the effectiveness and advancement of our proposed method.

An Efficient and Robust Unsupervised Anomaly Detection Method Using Ensemble Random Projection in Surveillance Videos.

Video anomaly detection is widely applied in modern society, which is achieved by sensors such as surveillance cameras. This paper learns anomalies by exploiting videos under the fully unsupervised setting. To avoid massive computation caused by back-prorogation in existing methods, we propose a novel efficient three-stage unsupervised anomaly detection method. In the first stage, we adopt random projection instead of autoencoder or its variants in previous works. Then we formulate the optimization goal as a least-square regression problem which has a closed-form solution, leading to less computational cost. The discriminative reconstruction losses of normal and abnormal events encourage us to roughly estimate normality that can be further sifted in the second stage with one-class support vector machine. In the third stage, to eliminate the instability caused by random parameter initializations, ensemble technology is performed to combine multiple anomaly detectors' scores. To the best of our knowledge, it is the first time that unsupervised ensemble technology is introduced to video anomaly detection research. As demonstrated by the experimental results on several video anomaly detection benchmark datasets, our algorithm robustly surpasses the recent unsupervised methods and performs even better than some supervised approaches. In addition, we achieve comparable performance contrast with the state-of-the-art unsupervised method with much less running time, indicating the effectiveness, efficiency, and robustness of our proposed approach.

ParaBTM: A Parallel Processing Framework for Biomedical Text Mining on Supercomputers.

A prevailing way of extracting valuable information from biomedical literature is to apply text mining methods on unstructured texts. However, the massive amount of literature that needs to be analyzed poses a big data challenge to the processing efficiency of text mining. In this paper, we address this challenge by introducing parallel processing on a supercomputer. We developed paraBTM, a runnable framework that enables parallel text mining on the Tianhe-2 supercomputer. It employs a low-cost yet effective load balancing strategy to maximize the efficiency of parallel processing. We evaluated the performance of paraBTM on several datasets, utilizing three types of named entity recognition tasks as demonstration. Results show that, in most cases, the processing efficiency can be greatly improved with parallel processing, and the proposed load balancing strategy is simple and effective. In addition, our framework can be readily applied to other tasks of biomedical text mining besides NER.

[Chemical constituents from Crotalaria sessiliflora L.].

In this study, we isolated and purified the extracts of the whole plant of Crotalaria sessiliflora L. by column chromatographic.Twelve compounds were isolated and identified as followings: sessiliflorin B(1), quercetin (2), kaempferol (3), soyasapogenol B(4), fernenol (5), neoechinulin A(6), ethyl 4-hydroxybenzoate (7), ethyl caffeate (8), 5,7-dihydroxychromone(9), crotadihydrofuran A(10), butesuperin B(11) and aurantiamide acetate(12).Compound 1 is a new compound, compound 3-12 were isolated from the plant for the first time.

Herbal prescription Chang'an II repairs intestinal mucosal barrier in rats with post-inflammation irritable bowel syndrome.

AIM: The herbal prescription Chang'an II is derived from a classical TCM formula Tong-Xie-Yao-Fang for the treatment of liver-qi stagnation and spleen deficiency syndrome of irritable bowel syndrome (IBS). In this study we investigated the effects of Chang'an II on the intestinal mucosal immune barrier in a rat post-inflammation IBS (PI-IBS) model. METHODS: A rat model of PI-IBS was established using a multi-stimulation paradigm including early postnatal sibling deprivation, bondage and intrarectal administration of TNBS. Four weeks after TNBS administration, the rats were treated with Chang'an II (2.85, 5.71 and 11.42 g · kg(-1) · d(-1), ig) for 14 d. Intestinal sensitivity was assessed based on the abdominal withdrawal reflex (AWR) scores and fecal water content. Open field test and two-bottle sucrose intake test were used to evaluate the behavioral changes. CD4(+) and CD8(+) cells were counted and IL-1ß and IL-4 levels were measured in intestinal mucosa. Transmission electron microscopy was used to evaluate ultrastructural changes of the intestinal mucosal barrier. RESULTS: PI-IBS model rats showed significantly increased AWR reactivity and fecal water content, and decreased locomotor activity and sucrose intake. Chang'an II treatment not only reduced AWR reactivity and fecal water content, but also suppressed the anxiety and depressive behaviors. Ultrastructural study revealed that the gut mucosal barrier function was severely damaged in PI-IBS model rats, whereas Chang'an II treatment relieved intestinal mucosal inflammation and repaired the gut mucosal barrier. Furthermore, PI-IBS model rats showed a significantly reduced CD4(+)/CD8(+) cell ratio in lamina propria and submucosa, and increased IL-1ß and reduced IL-4 expression in intestinal mucosa, whereas Chang'an II treatment reversed PI-IBS-induced changes in CD4(+)/CD8(+) cell ratio and expression of IL-1ß and IL-4. CONCLUSION: Chang'an II treatment protects the intestinal mucosa against PI-IBS through anti-inflammatory, immunomodulatory and anti-anxiety effects.

[Rats hyperuricemia model established by lipid emulsion simulating irregular of diet].

Due to the irregular of diet and overfeeding greasy and surfeit flavor closely associated with hyperuricemia disease, the lipid emulsion containing high cholesterol was used to model. To obtain a more stable and sustained animal model for the efficacy evaluation of traditional Chinese herbs, we observed the influence on the serum uric acid of rat induced by the lipid emulsion compared with high purine diet. 36 SD male rats were randomized to the normal control group, high purine diet group and lipid emulsion group respectively. The general behavior, body weight and daily food intake of rats were observed. The orbital blood was taken to separate into the serum and 24 hours urine was collected. The serum indexes such as UA, BUN, Cr, ALT, AST, TC, TG, LDL-c were determined every 2 weeks, and XOD, ADA enzyme activity were determined at the 4th week. The urine indexes such as UA, Cr and Cua/Ccr were determined at the 4th week. After stopping modeling, the serum UA were determined two weeks and four weeks later respectively. At the 2nd week, the body weight and daily food intake of rats in the lipid emulsion group reduced significantly, and the level of serum UA, BUN, Cr, TC, LDL-c, ATL, AST raised significantly meanwhile TG reduced. At the 4th week, the serum UA in high purine diet group did not raise, and the serum XOD raised obviously while ADA did not; the serum UA in lipid emulsion group was higher significantly, and the serum XOD and ADA raised while Cua/Ccr reduced obviously. At the 6th weeks, the serum UA in both the high purine diet group and lipid emulsion group raised obviously. After stopping modeling, the serum UA in lipid emulsion group still maintained a high level at the 2nd week and back to the normal level at the 4th week. Compared with high purine diet, the hyperuricemia model induced by lipid emulsion forms earlierand more stable. It maybe has great value to study the pharmacodynamics of traditional Chinese medicine treatment to hyperuricemia disease. Its mechanism may be related to increasing XOD and ADA enzyme activity which can promote uric acid synthesis, meanwhile inhibiting of uric acid excretion.

[Effect of improper diets on blood viscosity in SD rats in high-salt and fat diet and alcohol abuse simulation model].

OBJECTIVE: To observe the effect of composite factors, like long-term high-salt & fat diet and alcohol abuse on blood viscosity and blood pressure in rats, and compare with a model induced by high molecular dextran, in order to build a chronic hyperviscosity aminal model which is similar to human hyperviscosity in clinic and lay a foundation for efficacy evaluation on traditional Chinese medicines. METHOD: Male SD rats were randomly divided into the normal group, the high molecular dextran (HMD) group and the high salt & fat and alcohol (HSFA) group. The HMD group was given normal diet and water for 23 day and then 10% HMD through tail vein for 5 days. The HSFA group was fed with high salt and high fat diets every day and alcohol for 20 h x d(-1) for 13 weeks. After the modeling, whole blood viscosity and plasma viscosity were measured in the 5th, 8th and 11th week. Blood pressure was measured in the 5d, 7h, and 10th week. Red cell count (RBC) and hematocrit (HCT) were measured in the 11th week. PAgT, Fb, ET-1, NO, PGI, TXA2 contents of the normal group and the HSFA group were measured in the 13th week, and IECa21 content was measured with flow cytometry. Result: After the modeling, the HMD group was in good conditions with glossy hairs and active behaviors. The HSFA group was depressed with withered hairs and less activities. During the 5th-11th weeks, the HMD group and the HSFA group showed higher values in high and low shear whole blood viscosity (WBV) than the normal control group. The plasma viscosity (PV) of HMD rats was significantly increased only in the 5th week, and that of HSFA rats significantly increased in the 8"' and 11th week, particularly in the 11'h week. In the 111h week, the HSFA group showed significant increases in RBC and HCT. After the modeling, the blood pressure of HMD rats showed no significant changes, but the blood pressure of HSFA rats significantly increased during 7' and 101h weeks, particularly in the 10"' week. In the 13th week, PAgT, IECa2+, Fb, ET-1 of HSFA rats significantly increased, but with decreases in NO and PGI2. CONCLUSION: Long-term high salt & fat and alcohol diets can cause abnormal blood viscosity in rats. WBV significantly increased since the 5th week in rats, and PV increased since the 8th week. The mechanism for increasing BV may be: (1) increases in RBC, HCT, and IECa2+, (2) PAgT increase, (3) Fb content increase, or (4) TXA2/PGI2, ET-1/NO imbalance. Although the modeling time with the method is longer than that with the HMD method, the model is more stable and moderate, and could lead to abnormal increases in WBV and PV; Whereas the HMD method only induced transient increase in plasma viscosity and abnormal increase in SBP. The model is more similar to traditional Chinese medicine syndromes and pathogenesis, with higher value for studies on efficacy of traditional Chinese medicines.

Category Alignment Mechanism for Few-Shot Image Classification.

While humans can excel at image classification tasks by comparing a few images, existing metric-based few-shot classification methods are still not well adapted to novel tasks. Performance declines rapidly when encountering new patterns, as feature embeddings cannot effectively encode discriminative properties. Moreover, existing matching methods inadequately utilize support set samples, focusing only on comparing query samples to category prototypes without exploiting contrastive relationships across categories for discriminative features. In this work, we propose a method where query samples select their most category-representative features for matching, making feature embeddings adaptable and category-related. We introduce a category alignment mechanism (CAM) to align query image features with different categories. CAM ensures features chosen for matching are distinct and strongly correlated to intra-and inter-contrastive relationships within categories, making extracted features highly related to their respective categories. CAM is parameter-free, requires no extra training to adapt to new tasks, and adjusts features for matching when task categories change. We also implement a cross-validation-based feature selection technique for support samples, generating more discriminative category prototypes. We implement two versions of inductive and transductive inference and conduct extensive experiments on six datasets to demonstrate the effectiveness of our algorithm. The results indicate that our method consistently yields performance improvements on benchmark tasks and surpasses the current state-of-the-art methods.

Fast Continual Multi-View Clustering With Incomplete Views.

Multi-view clustering (MVC) has attracted broad attention due to its capacity to exploit consistent and complementary information across views. This paper focuses on a challenging issue in MVC called the incomplete continual data problem (ICDP). Specifically, most existing algorithms assume that views are available in advance and overlook the scenarios where data observations of views are accumulated over time. Due to privacy considerations or memory limitations, previous views cannot be stored in these situations. Some works have proposed ways to handle this problem, but all of them fail to address incomplete views. Such an incomplete continual data problem (ICDP) in MVC is difficult to solve since incomplete information with continual data increases the difficulty of extracting consistent and complementary knowledge among views. We propose Fast Continual Multi-View Clustering with Incomplete Views (FCMVC-IV) to address this issue. Specifically, the method maintains a scalable consensus coefficient matrix and updates its knowledge with the incoming incomplete view rather than storing and recomputing all the data matrices. Considering that the given views are incomplete, the newly collected view might contain samples that have yet to appear; two indicator matrices and a rotation matrix are developed to match matrices with different dimensions. In addition, we design a three-step iterative algorithm to solve the resultant problem with linear complexity and proven convergence. Comprehensive experiments conducted on various datasets demonstrate the superiority of FCMVC-IV over the competing approaches. The code is publicly available at https://github.com/wanxinhang/FCMVC-IV.

Scalable and Structural Multi-View Graph Clustering With Adaptive Anchor Fusion.

Anchor graph has been recently proposed to accelerate multi-view graph clustering and widely applied in various large-scale applications. Different from capturing full instance relationships, these methods choose small portion anchors among each view, construct single-view anchor graphs and combine them into the unified graph. Despite its efficiency, we observe that: (i) Existing mechanism adopts a separable two-step procedure-anchor graph construction and individual graph fusion, which may degrade the clustering performance. (ii)These methods determine the number of selected anchors to be equal among all the views, which may destruct the data distribution diversity. A more flexible multi-view anchor graph fusion framework with diverse magnitudes is desired to enhance the representation ability. (iii) During the latter fusion process, current anchor graph fusion framework follows simple linearly-combined style while the intrinsic clustering structures are ignored. To address these issues, we propose a novel scalable and flexible anchor graph fusion framework for multi-view graph clustering method in this paper. Specially, the anchor graph construction and graph alignment are jointly optimized in our unified framework to boost clustering quality. Moreover, we present a novel structural alignment regularization to adaptively fuse multiple anchor graphs with different magnitudes. In addition, our proposed method inherits the linear complexity of existing anchor strategies respecting to the sample number, which is time-economical for large-scale data. Experiments conducted on various benchmark datasets demonstrate the superiority and effectiveness of the newly proposed anchor graph fusion framework against the existing state-of-the-arts over the clustering performance promotion and time expenditure. Our code is publicly available at https://github.com/wangsiwei2010/SMVAGC-SF.

Differentiated Anchor Quantity Assisted Incomplete Multiview Clustering Without Number-Tuning.

Incomplete multiview clustering (IMVC) generally requires the number of anchors to be the same in all views. Also, this number needs to be tuned with extra manual efforts. This not only degenerates the diversity of multiview data but also limits the model's scalability. For generating differentiated numbers of anchors without tuning, in this article we devise a novel framework named DAQINT. To be specific, the most perfect solution is to jointly find the optimal number of anchors that belongs to respective view. Regretfully, it is extremely time consuming. In view of this, we choose to first offer a set of anchor numbers for each view, and then integrate their contributions by adaptive weighting to approximate the optimal number. In particular, these offered numbers are all predefined and do not require any tuning. Through adaptively weighting them, we hold that this equivalently makes each view enjoy a different number of anchors. Accordingly, the bipartite graphs generated on all views are with diverse scales. Besides exploring multiview features more deeply, they also balance the importance between views. Then, to fuse these multiscale bipartite graphs, we design a combination strategy that owns linear computation and storage overheads. Afterward, to solve the resulting optimization problem, we also carefully develop a three-step iterative algorithm with linear complexities and demonstrated convergence. Experiments on the multiple public datasets validate the superiority of DAQINT against several advanced IMVC methods, such as on Mfeat, DAQINT surpasses the competitors like MKC, EEIMVC, FLSD, DSIMVC, IMVC-CBG, and DCP by 36.65%, 6.33%, 48.53%, 22.46%, 15.06%, and 32.04%, respectively, in ACC.

TFMKC: Tuning-Free Multiple Kernel Clustering Coupled With Diverse Partition Fusion.

Clustering is a popular research pipeline in unsupervised learning to find potential groupings. As a representative paradigm in multiple kernel clustering (MKC), late fusion-based models learn a consistent partition across multiple base kernels. Despite their promising performance, a common concern is the limited representation capacity caused by the inflexible fusion mechanism. Concretely, the representations are constrained by truncated- k Eigen-decomposition (EVD) without fully exploiting potential information. An intuitive idea to alleviate this concern is to generate a set of augmented partitions and then select the optimal partition by fine-tuning. However, this is overlimited by: 1) introducing undesired hyperparameters and dataset-related consequences; 2) neglecting rich information across diverse partitions; and 3) expensive parameter-tuning costs. To address these problems, we propose transforming the challenging problem of directly determining the optimal partition (optimal parameter) into a diverse partition fusion (parameter ensemble) problem. We design a novel flexible fusion mechanism called tuning-free multiple kernel clustering coupled with diverse partition fusion (TFMKC) by reweighting diverse partitions through optimization, achieving an optimal consensus partition by integrating diverse and complementary information rather than traditional fine-tuning, and distinguishing our work from existing methods. Extensive experiments verify that TFMKC achieves competitive effectiveness and efficiency over comparison baselines. The code can be accessed at https://github.com/ZJP/TFMKC.

Regularized Instance Weighting Multiview Clustering via Late Fusion Alignment.

Multiview clustering has become a prominent research topic in data analysis, with wide-ranging applications across various fields. However, the existing late fusion multiview clustering (LFMVC) methods still exhibit some limitations, including variable importance and contributions and a heightened sensitivity to noise and outliers during the alignment process. To tackle these challenges, we propose a novel regularized instance weighting multiview clustering via late fusion alignment (R-IWLF-MVC), which considers the instance importance from various views, enabling information integration to be more effective. Specifically, we assign each sample an importance attribute to enable the learning process to focus more on the key sample nodes and avoid being influenced by noise or outliers, while laying the groundwork for the fusion of different views. In addition, we continue to employ late fusion alignment to integrate base clustering from various views and introduce a new regularization term with prior knowledge to ensure that the learning process does not deviate too much from the expected results. After that, we design a three-step alternating optimization strategy with proven convergence for the resultant problem. Our proposed approach has been extensively evaluated on multiple real-world datasets, demonstrating its superiority to state-of-the-art methods.

On the Consistency and Large-Scale Extension of Multiple Kernel Clustering.

Existing multiple kernel clustering (MKC) algorithms have two ubiquitous problems. From the theoretical perspective, most MKC algorithms lack sufficient theoretical analysis, especially the consistency of learned parameters, such as the kernel weights. From the practical perspective, the high complexity makes MKC unable to handle large-scale datasets. This paper tries to address the above two issues. We first make a consistency analysis of an influential MKC method named Simple Multiple Kernel k-Means (SimpleMKKM). Specifically, suppose that ∧γn are the kernel weights learned by SimpleMKKM from the training samples. We also define the expected version of SimpleMKKM and denote its solution as γ*. We establish an upper bound of ||∧γn-γ*||∞ in the order of ~O(1/√n), where n is the sample number. Based on this result, we also derive its excess clustering risk calculated by a standard clustering loss function. For the large-scale extension, we replace the eigen decomposition of SimpleMKKM with singular value decomposition (SVD). Consequently, the complexity can be decreased to O(n) such that SimpleMKKM can be implemented on large-scale datasets. We then deduce several theoretical results to verify the approximation ability of the proposed SVD-based method. The results of comprehensive experiments demonstrate the superiority of the proposed method.