Robust k -Means-Type Clustering for Noisy Data.

Data clustering is a fundamental machine learning task that seeks to categorize a dataset into homogeneous groups. However, real data usually contain noise, which poses significant challenges to clustering algorithms. In this article, motivated by how the k -means algorithm is derived from a Gaussian mixture model (GMM), we propose a robust k -means-type algorithm, named k -means-type clustering based on t -distribution (KMTD), by assuming that the data points are drawn from a special multivariate t -mixture model (TMM). Compared to the Gaussian distribution, the t -distribution has a fatter tail. The proposed algorithm is more robust to noise. Like the k -means algorithm, the proposed algorithm is simpler than those based on a full TMM. Both synthetic and actual data are used to illustrate the proposed algorithm's performance and efficiency. The experimental results demonstrated that the proposed algorithm operates more quickly than other sophisticated algorithms and, in most cases, achieves higher accuracy than the other algorithms.

IL-37 Modulates Myocardial Calcium Handling via the p-STAT3/SERCA2a Axis in HF-Related Engineered Human Heart Tissue.

Interleukin-37 (IL-37) is a potent anti-inflammatory cytokine belonging to the IL-1 family. This study investigates the regulatory mechanism and reparative effects of IL-37 on HF-related human induced pluripotent stem cells derived cardiomyocytes (hiPSC-CMs) and engineered human heart tissue subjected to hypoxia and H2O2 treatment. The contractile force and Ca2+ conduction capacity of the tissue are assessed using a stretching platform and high-resolution fluorescence imaging system. This investigation reveals that IL-37 treatment significantly enhances cell viability, calcium transient levels, contractile force, and Ca2+ conduction capacity in HF-related hiPSC-CMs and engineered human heart tissue. Notably, IL-37 facilitates the upregulation of sarcoplasmic reticulum calcium ATPase 2a (SERCA2a) through enhancing nuclear p-STAT3 levels. This effect is mediated by the binding of p-STAT3 to the SERCA2a promoter, providing a novel insight on the reparative potential of IL-37 in HF. IL-37 demonstrates its ability to enhance systolic function by modulating myocardial calcium handling via the p-STAT3/SERCA2a axis in HF-related engineered human heart tissue (as shown in schematic diagram).

Versatile Weight Attack via Flipping Limited Bits.

Many attack paradigms against deep neural networks have been well studied, such as the backdoor attack in the training stage and the adversarial attack in the inference stage. In this article, we study a novel attack paradigm, the bit-flip based weight attack, which directly modifies weight bits of the attacked model in the deployment stage. To meet various attack scenarios, we propose a general formulation including terms to achieve effectiveness and stealthiness goals and a constraint on the number of bit-flips. Furthermore, benefitting from this extensible and flexible formulation, we present two cases with different malicious purposes, i.e., single sample attack (SSA) and triggered samples attack (TSA). SSA which aims at misclassifying a specific sample into a target class is a binary optimization with determining the state of the binary bits (0 or 1); TSA which is to misclassify the samples embedded with a specific trigger is a mixed integer programming (MIP) with flipped bits and a learnable trigger. Utilizing the latest technique in integer programming, we equivalently reformulate them as continuous optimization problems, whose approximate solutions can be effectively and efficiently obtained by the alternating direction method of multipliers (ADMM) method. Extensive experiments demonstrate the superiority of our methods.

On the Effectiveness of Adversarial Training Against Backdoor Attacks.

Although adversarial training (AT) is regarded as a potential defense against backdoor attacks, AT and its variants have only yielded unsatisfactory results or have even inversely strengthened backdoor attacks. The large discrepancy between expectations and reality motivates us to thoroughly evaluate the effectiveness of AT against backdoor attacks across various settings for AT and backdoor attacks. We find that the type and budget of perturbations used in AT are important, and AT with common perturbations is only effective for certain backdoor trigger patterns. Based on these empirical findings, we present some practical suggestions for backdoor defense, including relaxed adversarial perturbation and composite AT. This work not only boosts our confidence in AT's ability to defend against backdoor attacks but also provides some important insights for future research.

Adversarial Examples Generation for Deep Product Quantization Networks on Image Retrieval.

Deep product quantization networks (DPQNs) have been successfully used in image retrieval tasks, due to their powerful feature extraction ability and high efficiency of encoding high-dimensional visual features. Recent studies show that deep neural networks (DNNs) are vulnerable to input with small and maliciously designed perturbations (a.k.a., adversarial examples) for classification. However, little effort has been devoted to investigating how adversarial examples affect DPQNs, which raises the potential safety hazard when deploying DPQNs in a commercial search engine. To this end, we propose an adversarial example generation framework by generating adversarial query images for DPQN-based retrieval systems. Unlike the adversarial generation for the classic image classification task that heavily relies on ground-truth labels, we alternatively perturb the probability distribution of centroids assignments for a clean query, then we can induce effective non-targeted attacks on DPQNs in white-box and black-box settings. Moreover, we further extend the non-targeted attack to a targeted attack by a novel sample space averaging scheme ([Formula: see text]AS), whose theoretical guarantee is also obtained. Extensive experiments show that our methods can create adversarial examples to successfully mislead the target DPQNs. Besides, we found that our methods both significantly degrade the retrieval performance under a wide variety of experimental settings. The source code is available at https://github.com/Kira0096/PQAG.

Establishment of induced pluripotent stem cell lines from an ARVC patient carrying a heterozygous variant in LAMA2 gene.

Using peripheral blood mononuclear cell (PBMC) reprogramming technology, a human-induced pluripotent stem cell (iPSC) line was produced from a patient who presents classic clinical features of arrhythmogenic right ventricular cardiomyopathy (ARVA) and carries a de novo laminin subunit alpha 2 (LAMA2) heterozygous mutation (NM_000426.3: c.8842G > A, p.G2948S). This mutation was not inherited from his parents, who also present normal cardiac phenotype. This iPSC line demonstrates a normal karyotype. Pluripotency and differentiation capacity has been confirmed in vitro. This cell line can help in efforts to understand the pathogenic mechanism between LAMA2 mutation and ARVC.

Yap1 modulates cardiomyocyte hypertrophy via impaired mitochondrial biogenesis in response to chronic mechanical stress overload.

Rationale: Chronic pressure overload is a major trigger of cardiac pathological hypertrophy that eventually leads to heart disease and heart failure. Understanding the mechanisms governing hypertrophy is the key to develop therapeutic strategies for heart diseases. Methods: We built chronic pressure overload mice model by abdominal aortic constriction (AAC) to explore the features of Yes-associated protein 1 (YAP1). Then AAV-cTNT-Cre was applied to Yap1F/F mice to induce mosaic depletion of YAP1. Myh6CreERT2; H11CAG-LSL-YAP1 mice were involved to establish YAP1 overexpression model by Tomaxifen injection. ATAC-seq and bioChIP-seq were used to explore the potential targets of YAP1, which were verified by a series of luciferase reporter assays. Dnm1l and Mfn1 were re-expressed in AAC mice by AAV-cTNT-Dnm1l and AAV-cTNT-Mfn1. Finally, Verteprofin was used to inhibit YAP1 to rescue cardiac hypertrophy. Results: We found that pathological hypertrophy was accompanied with the activation of YAP1. Cardiomyocyte-specific deletion of Yap1 attenuated AAC-induced hypertrophy. Overexpression of YAP1 was sufficient to phenocopy AAC-induced hypertrophy. YAP1 activation resulted in the perturbation of mitochondria ultrastructure and function, which was associated with the repression of mitochondria dynamics regulators Dnm1l and Mfn1. Mitochondrial-related genes Dnm1l and Mfn1, are significantly targeted by TEAD1/YAP complex. Overexpression of Dnm1l and Mfn1 synergistically rescued YAP1-induced mitochondrial damages and cardiac hypertrophy. Pharmacological repression of YAP1 by verteporfin attenuated mitochondrial damages and pathological hypertrophy in AAC-treated mice. Interestingly, YAP1-induced mitochondria damages also led to increased reactive oxidative species, DNA damages, and the suppression of cardiomyocyte proliferation. Conclusion: Together, these data uncovered YAP signaling as a therapeutic target for pressure overload-induced heart diseases and cautioned the efforts to induce cardiomyocyte regeneration by activating YAP.

Backdoor Learning: A Survey.

Backdoor attack intends to embed hidden backdoors into deep neural networks (DNNs), so that the attacked models perform well on benign samples, whereas their predictions will be maliciously changed if the hidden backdoor is activated by attacker-specified triggers. This threat could happen when the training process is not fully controlled, such as training on third-party datasets or adopting third-party models, which poses a new and realistic threat. Although backdoor learning is an emerging and rapidly growing research area, there is still no comprehensive and timely review of it. In this article, we present the first comprehensive survey of this realm. We summarize and categorize existing backdoor attacks and defenses based on their characteristics, and provide a unified framework for analyzing poisoning-based backdoor attacks. Besides, we also analyze the relation between backdoor attacks and relevant fields (i.e., adversarial attacks and data poisoning), and summarize widely adopted benchmark datasets. Finally, we briefly outline certain future research directions relying upon reviewed works. A curated list of backdoor-related resources is also available at https://github.com/THUYimingLi/backdoor-learning-resources.

Application of hiPSC as a Drug Tester Via Mimicking a Personalized Mini Heart.

Human induced pluripotent stem cells (hIPSC) have been used to produce almost all types of human cells currently, which makes them into several potential applications with replicated patient-specific genotype. Furthermore, hIPSC derived cardiomyocytes assembled engineering heart tissue can be established to achieve multiple functional evaluations by tissue engineering technology. This short review summarized the current advanced applications based on the hIPSC derived heart tissue in molecular mechanisms elucidating and high throughput drug screening.

Mitochondrial dysfunction by TFAM depletion disrupts self-renewal and lineage differentiation of human PSCs by affecting cell proliferation and YAP response.

Genetic mitochondrial dysfunction is frequently associated with various embryonic developmental defects. However, how mitochondria contribute to early development and cell fate determination is poorly studied, especially in humans. Using human pluripotent stem cells (hPSCs), we established a Dox-induced knockout model with mitochondrial dysfunction and evaluated the effect of mitochondrial dysfunction on human pluripotency maintenance and lineage differentiation. The nucleus-encoded gene TFAM (transcription factor A, mitochondrial), essential for mitochondrial gene transcription and mitochondrial DNA replication, is targeted to construct the mitochondrial dysfunction model. The hPSCs with TFAM depletion exhibit the decrease of mtDNA level and oxidative respiration efficiency, representing a typical mitochondrial dysfunction phenotype. Mitochondrial dysfunction leads to impaired self-renewal in hPSCs due to proliferation arrest. Although the mitochondrial dysfunction does not affect pluripotent gene expression, it results in a severe defect in lineage differentiation. Further study in mesoderm differentiation reveals that mitochondrial dysfunction causes proliferation disability and YAP nuclear translocalization and thus together blocks mesoderm lineage differentiation. These findings provide new insights into understanding the mitochondrial function in human pluripotency maintenance and mesoderm differentiation.

S2A: Scale-Attention-Aware Networks for Video Super-Resolution.

Convolutional Neural Networks (CNNs) have been widely used in video super-resolution (VSR). Most existing VSR methods focus on how to utilize the information of multiple frames, while neglecting the feature correlations of the intermediate features, thus limiting the feature expression of the models. To address this problem, we propose a novel SAA network, that is, Scale-and-Attention-Aware Networks, to apply different attention to different temporal-length streams, while further exploring both spatial and channel attention on separate streams with a newly proposed Criss-Cross Channel Attention Module (C3AM). Experiments on public VSR datasets demonstrate the superiority of our method over other state-of-the-art methods in terms of both quantitative and qualitative metrics.

Smartwatch User Authentication by Sensing Tapping Rhythms and Using One-Class DBSCAN.

As important sensors in smart sensing systems, smartwatches are becoming more and more popular. Authentication can help protect the security and privacy of users. In addition to the classic authentication methods, behavioral factors can be used as robust measures for this purpose. This study proposes a lightweight authentication method for smartwatches based on edge computing, which identifies users by their tapping rhythms. Based on the DBSCAN clustering algorithm, a new classification method called One-Class DBSCAN is presented. It first seeks core objects and then leverages them to perform user authentication. We conducted extensive experiments on 6110 real data samples collected from more than 600 users. The results show that our method achieved the lowest Equal Error Rate (EER) of only 0.92%, which was lower than those of other state-of-the-art methods. In addition, a statistical method for detecting the security level of a tapping rhythm is proposed. It can prevent users from setting a simple tapping rhythm password, and thus improve the security of smartwatches.

Attenuation of Cardiomyocyte Hypertrophy via Depletion Myh7 using CASAAV.

Myh7 is a classic biomarker for cardiac remodeling and a potential target to attenuate cardiomyocyte (CM) hypertrophy. This study aimed to identify the dominant function of Myh7 after birth and determine whether its removal would affect CM maturation or contribute to reversal of pathological hypertrophy phenotypes. The CASAAV (CRISPR/Cas9-AAV9-based somatic mutagenesis) technique was used to deplete Myh6 and Myh7, and an AAV dosage of 5 × 109 vg/g was used to generate a mosaic CM depletion model to explore the function of Myh7 in adulthood. CM hypertrophy was induced by transverse aortic constriction (TAC) in Rosa26Cas9-P2A-GFP mice at postnatal day 28 (PND28). Heart function was measured by echocardiography. Isolated CMs and in situ imaging were used to analyze the structure and morphology of CM. We discovered that CASAAV successfully silenced Myh6 and Myh7 in CMs, and early depletion of Myh7 led to mild adulthood lethality. However, the Myh7 PND28-knockout mice had normal heart phenotype and function, with normal cellular size and normal organization of sarcomeres and T-tubules. The TAC mice also received AAV-Myh7-Cre to produce Myh7-knockout CMs, which were also of normal size, and echocardiography demonstrated a reversal of cardiac hypertrophy. In conclusion, Myh7 has a role during the maturation period but rarely functions in adulthood. Thus, the therapeutic time should exceed the period of maturation. These results confirm Myh7 as a potential therapeutic target and indicate that its inhibition could help reverse CM hypertrophy.

TNT: An Interpretable Tree-Network-Tree Learning Framework using Knowledge Distillation.

Deep Neural Networks (DNNs) usually work in an end-to-end manner. This makes the trained DNNs easy to use, but they remain an ambiguous decision process for every test case. Unfortunately, the interpretability of decisions is crucial in some scenarios, such as medical or financial data mining and decision-making. In this paper, we propose a Tree-Network-Tree (TNT) learning framework for explainable decision-making, where the knowledge is alternately transferred between the tree model and DNNs. Specifically, the proposed TNT learning framework exerts the advantages of different models at different stages: (1) a novel James-Stein Decision Tree (JSDT) is proposed to generate better knowledge representations for DNNs, especially when the input data are in low-frequency or low-quality; (2) the DNNs output high-performing prediction result from the knowledge embedding inputs and behave as a teacher model for the following tree model; and (3) a novel distillable Gradient Boosted Decision Tree (dGBDT) is proposed to learn interpretable trees from the soft labels and make a comparable prediction as DNNs do. Extensive experiments on various machine learning tasks demonstrated the effectiveness of the proposed method.

Establishment of induced pluripotent stem cell lines from a family of an ARVC patient receiving heart transplantation in infant age carrying compound heterozygous mutations in DSP gene.

The mutations of desmoplakin (DSP) lead to arrhythmia right ventricular cardiomyopathy (ARVC) which is a kind of progressive cardiomyopathy characterized by arrhythmia, heart failure and sudden cardiac death. The human induced pluripotent stem (iPS) cell line HUBUi001-A was generated from a patient carrying the compound DSP heterozygous mutations (c.104G > T p.G35V; c.5617C > T p.R1873C), which were inherited from his parents (HUBUi002-A, HUBUi003- A), who presented a normal phenotype. We have derived the iPSC cell lines through peripheral blood mononuclear (PBMCs) cell reprogramming technology. Pluripotency and differentiation capacity have been confirmed by RT-PCR, immunocytochemistry and teratoma experiment. These cell lines can help us understand the pathogenic mechanism and screening potential therapeutic options.

Establishment of induced pluripotent stem cell line CSUASOi003- A from an autosomal recessive retinitis pigmentosa patient carrying compound heterozygous mutations in CRB1 gene.

Crumbs homologue 1 (CRB1) mutations have been found in retinitis pigmentosa (RP) patients lead to severe retinal dystrophies. The human induced pluripotent stem (iPS) cell line CSUASOi003-A derived from peripheral blood mononuclear cells (PBMCs) of a patient carrying two heterozygous mutations (2249G>A p.G750D and c.2809G>A p.A937T) in CRB1 gene was generated by non-integrative reprogramming technology. Pluripotency and differentiation capacity were assessed by immunocytochemistry and quantitative polymerase chain reaction (qPCR). The RP patient-specific iPS cell line provide a powerful model for evaluating the pathological phenotypes of the disease.

CNN-MHSA: A Convolutional Neural Network and multi-head self-attention combined approach for detecting phishing websites.

Increasing phishing sites today have posed great threats due to their terribly imperceptible hazard. They expect users to mistake them as legitimate ones so as to steal user information and properties without notice. The conventional way to mitigate such threats is to set up blacklists. However, it cannot detect one-time Uniform Resource Locators (URL) that have not appeared in the list. As an improvement, deep learning methods are applied to increase detection accuracy and reduce the misjudgment ratio. However, some of them only focus on the characters in URLs but ignore the relationships between characters, which results in that the detection accuracy still needs to be improved. Considering the multi-head self-attention (MHSA) can learn the inner structures of URLs, in this paper, we propose CNN-MHSA, a Convolutional Neural Network (CNN) and the MHSA combined approach for highly-precise. To achieve this goal, CNN-MHSA first takes a URL string as the input data and feeds it into a mature CNN model so as to extract its features. In the meanwhile, MHSA is applied to exploit characters' relationships in the URL so as to calculate the corresponding weights for the CNN learned features. Finally, CNN-MHSA can produce highly-precise detection result for a URL object by integrating its features and their weights. The thorough experiments on a dataset collected in real environment demonstrate that our method achieves 99.84% accuracy, which outperforms the classical method CNN-LSTM and at least 6.25% higher than other similar methods on average.

Insights Into the Pathogenesis of Catecholaminergic Polymorphic Ventricular Tachycardia From Engineered Human Heart Tissue.

BACKGROUND: Modeling of human arrhythmias with induced pluripotent stem cell-derived cardiomyocytes has focused on single-cell phenotypes. However, arrhythmias are the emergent properties of cells assembled into tissues, and the impact of inherited arrhythmia mutations on tissue-level properties of human heart tissue has not been reported. METHODS: Here, we report an optogenetically based, human engineered tissue model of catecholaminergic polymorphic ventricular tachycardia (CPVT), an inherited arrhythmia caused by mutation of the cardiac ryanodine channel and triggered by exercise. We developed a human induced pluripotent stem cell-derived cardiomyocyte-based platform to study the tissue-level properties of engineered human myocardium. We investigated pathogenic mechanisms in CPVT by combining this novel platform with genome editing. RESULTS: In our model, CPVT tissues were vulnerable to developing reentrant rhythms when stimulated by rapid pacing and catecholamine, recapitulating hallmark features of the disease. These conditions elevated diastolic Ca2+ levels and increased temporal and spatial dispersion of Ca2+ wave speed, creating a vulnerable arrhythmia substrate. Using Cas9 genome editing, we pinpointed a single catecholamine-driven phosphorylation event, ryanodine receptor-serine 2814 phosphorylation by Ca2+/calmodulin-dependent protein kinase II, that is required to unmask the arrhythmic potential of CPVT tissues. CONCLUSIONS: Our study illuminates the molecular and cellular pathogenesis of CPVT and reveals a critical role of calmodulin-dependent protein kinase II-dependent reentry in the tissue-scale mechanism of this disease. We anticipate that this approach will be useful for modeling other inherited and acquired cardiac arrhythmias.

A Novel Consistent Random Forest Framework: Bernoulli Random Forests.

Random forests (RFs) are recognized as one type of ensemble learning method and are effective for the most classification and regression tasks. Despite their impressive empirical performance, the theory of RFs has yet been fully proved. Several theoretically guaranteed RF variants have been presented, but their poor practical performance has been criticized. In this paper, a novel RF framework is proposed, named Bernoulli RFs (BRFs), with the aim of solving the RF dilemma between theoretical consistency and empirical performance. BRF uses two independent Bernoulli distributions to simplify the tree construction, in contrast to the RFs proposed by Breiman. The two Bernoulli distributions are separately used to control the splitting feature and splitting point selection processes of tree construction. Consequently, theoretical consistency is ensured in BRF, i.e., the convergence of learning performance to optimum will be guaranteed when infinite data are given. Importantly, our proposed BRF is consistent for both classification and regression. The best empirical performance is achieved by BRF when it is compared with state-of-the-art theoretical/consistent RFs. This advance in RF research toward closing the gap between theory and practice is verified by the theoretical and experimental studies in this paper.

Image quality assessment based on inter-patch and intra-patch similarity.

In this paper, we propose a full-reference (FR) image quality assessment (IQA) scheme, which evaluates image fidelity from two aspects: the inter-patch similarity and the intra-patch similarity. The scheme is performed in a patch-wise fashion so that a quality map can be obtained. On one hand, we investigate the disparity between one image patch and its adjacent ones. This disparity is visually described by an inter-patch feature, where the hybrid effect of luminance masking and contrast masking is taken into account. The inter-patch similarity is further measured by modifying the normalized correlation coefficient (NCC). On the other hand, we also attach importance to the impact of image contents within one patch on the IQA problem. For the intra-patch feature, we consider image curvature as an important complement of image gradient. According to local image contents, the intra-patch similarity is measured by adaptively comparing image curvature and gradient. Besides, a nonlinear integration of the inter-patch and intra-patch similarity is presented to obtain an overall score of image quality. The experiments conducted on six publicly available image databases show that our scheme achieves better performance in comparison with several state-of-the-art schemes.