Search

Re-identification (Re-ID) is a process that seeks to identify concern individuals from successive non-overlapping photographs. The area of computer vision has recently seen an uptick in the amount of attention focused on deep neural networks, especially given the popularity of smart monitoring systems and the development of sophisticated learning algorithms. We classified existing Re-ID technologies into closed-world and open-world contexts based on the used components. The closed-world scenario has been commonly used under a variety of data analysis hypotheses, and it brought precise results when applied to a variety of datasets utilizing deep learning techniques. We began with a comprehensive overview of closed-world person Re-ID considering deep metric learning, an extensive representation of features learning, ranking optimization, and in-depth analysis. Due to the accomplishment of performance in the packed scenario, the Re-ID focuses research has lately turned to a bare environment setting, which brings new issues. This setting is more akin to what we'd find in real-world circumstances. We summarized the unsupervised Re-ID literature as well as current research trends and proposed future studies. [ABSTRACT FROM AUTHOR]

Traditional high-dimensional electroencephalography (EEG) features (spectral or temporal) may not always attain satisfactory results in cognitive workload estimation. In contrast, deep representation learning (DRL) transforms high-dimensional data into cluster-friendly low-dimensional feature space. Therefore, this paper proposes an Integrated Spatio-Temporal Deep Clustering (ISTDC) model that uses DRL followed by a clustering method to achieve better clustering performance. The proposed model is illustrated using four Algorithms and Variational Bayesian Gaussian Mixture Model (VBGMM) clustering method. Temporal and spatial Variational Auto Encoder (VAE) models (mentioned in Algorithm 2 and Algorithm 3) learn temporal and spatial latent features from sequence-wise EEG signals and scalp topographical maps using the Long short-term memory and Convolutional Neural Network models. The concatenated spatio-temporal latent feature (mentioned in Algorithm 4) is passed to the VBGMM clustering method to efficiently estimate workload levels of n -back task. For the 0-back vs. 2-back task, the proposed model achieves the maximum mean clustering accuracy of 98.0%, and it improves by 11.0% over the state-of-the-art method. The results also indicate that the proposed multimodal approach outperforms temporal and spatial latent feature-based unimodal models in workload assessment. [ABSTRACT FROM AUTHOR]

The task of automatically extracting large homogeneous datasets of medical images based on detailed criteria and/or semantic similarity can be challenging because the acquisition and storage of medical images in clinical practice is not fully standardised and can be prone to errors, which are often made unintentionally by medical professionals during manual input. In this paper, we propose an algorithm for learning cluster-oriented representations of medical images by fusing images with partially observable DICOM tags. Pairwise relations are modelled by thresholding the Gower distance measure which is calculated using eight DICOM tags. We trained the models using 30,000 images, and we tested them using a disjoint test set consisting of 8000 images, gathered retrospectively from the PACS repository of the Clinical Hospital Centre Rijeka in 2017. We compare our method against the standard and deep unsupervised clustering algorithms, as well as the popular semi-supervised algorithms combined with the most commonly used feature descriptors. Our model achieves an NMI score of 0.584 with respect to the anatomic region, and an NMI score of 0.793 with respect to the modality. The results suggest that DICOM data can be used to generate pairwise constraints that can help improve medical images clustering, even when using only a small number of constraints.

• A document clustering framework that leverages contextualized vectors is proposed. • Informative representations for documents are extracted from pre-trained models. • A partial optimization and centroid update is proposed in the clustering module. • The proposed method outperforms the baselines in several datasets for clustering. • The effect of clustering method and embeddings are explored in various experiments. Document representation is central to modern natural language processing systems including document clustering. Empirical experiments in recent studies provide strong evidence that unsupervised language models can learn context-aware representations in the given documents and advance several NLP benchmark results. However, existing clustering approaches focus on the dimensionality reduction and do not exploit these informative representations. In this paper, we propose a conceptually simple but experimentally effective clustering framework called Advanced Document Clustering (ADC). In contrast to previous clustering methods, ADC is designed to leverage syntactically and semantically meaningful features through feature-extraction and clustering modules in the framework. We first extract features from pre-trained language models and initialize cluster centroids to spread out uniformly. In the clustering module of ADC, the semantic similarity can be measured using the cosine similarity and centroids update while assigning centroids to a mini-batch input. Also, we utilize cross entropy loss partially, as the self-training scheme can be biased when parameters in the model are inaccurate. As a result, ADC can take advantages of contextualized representations while mitigating the limitations introduced by high-dimensional vectors. In numerous experiments with four datasets, the proposed ADC outperforms other existing approaches. In particular, experiments on categorizing news corpus with fake news demonstrated the effectiveness of our method for contextualized representations. [ABSTRACT FROM AUTHOR]

Short-term traffic prediction (STTP) is one of the most critical capabilities in Intelligent Transportation Systems (ITS), which can be used to support driving decisions, alleviate traffic congestion and improve transportation efficiency. However, STTP of large-scale road networks remains challenging due to the difficulties of effectively modeling the diverse traffic patterns by high-dimensional time series. Therefore, this paper proposes a framework that involves a deep clustering method for STTP in large-scale road networks. The deep clustering method is employed to supervise the representation learning in a visualized way from the large unlabeled dataset. More specifically, to fully exploit the traffic periodicity, the raw series is first divided into a number of sub-series for triplet generation. The convolutional neural networks (CNNs) with triplet loss are utilized to extract the features of shape by transforming the series into visual images. The shape-based representations are then used to cluster road segments into groups. Thereafter, a model sharing strategy is further proposed to build recurrent NNs-based predictions through group-based models (GBMs). GBM is built for a type of traffic patterns, instead of one road segment exclusively or all road segments uniformly. Our framework can not only significantly reduce the number of prediction models, but also improve their generalization by virtue of being trained on more diverse examples. Furthermore, the proposed framework over a selected road network in Beijing is evaluated. Experiment results show that the deep clustering method can effectively cluster the road segments and GBM can achieve comparable prediction accuracy against the IBM with less number of prediction models. [ABSTRACT FROM AUTHOR]

Subspace clustering aims to separate the data into clusters under the hypothesis that the samples within the same cluster will lie in the same low-dimensional subspace. Due to the tough pairwise constraints, $k$ -subspace clustering is sensitive to outliers and initialization. In this letter, we present a novel deep architecture for $k$ -subspace clustering to address this issue, called as Deep Weighted $k$ -Subspace Clustering (DWSC). Specifically, our framework consists of autoencoder and weighted $k$ -subsapce network. We first use the autoencoder to non-linearly compress the samples into the low-dimensional latent space. In the weighted $k$ -subspace network, we feed the latent representation into the assignment network to output soft assignments which represent the probability of data belonging to the according subspace. Subsequently, the optimal $k$ subspaces are identified by minimizing the projection residuals of the latent representations to all subspaces, using the learned soft assignments as a weighting vector. Finally, we jointly optimize the representation learning and clustering in a unified framework. Experimental results show that our approach outperforms the state-of-the-art subspace clustering methods on two benchmark datasets. [ABSTRACT FROM AUTHOR]

We formulate a generic framework for blind source separation (BSS), which allows integrating data-driven spectro-temporal methods, such as deep clustering and deep attractor networks, with physically motivated probabilistic spatial methods, such as complex angular central Gaussian mixture models. The integrated model exploits the complementary strengths of the two approaches to BSS: the strong modeling power of neural networks, which, however, is based on supervised learning, and the ease of unsupervised learning of the spatial mixture models whose few parameters can be estimated on as little as a single segment of a real mixture of speech. Experiments are carried out on both artificially mixed speech and true recordings of speech mixtures. The experiments verify that the integrated models consistently outperform the individual components. We further extend the models to cope with noisy, reverberant speech and introduce a cross-domain teacher–student training where the mixture model serves as the teacher to provide training targets for the student neural network. [ABSTRACT FROM AUTHOR]

Speech enhancement and source separation systems based on deep learning have recently reached unprecedented levels of quality, to the point that performance is reaching a new ceiling. Most systems rely on estimating the magnitude of a target source by estimating a real-valued mask to be applied to a time-frequency representation of the mixture signal. A limiting factor in such approaches is a lack of phase estimation: the phase of the mixture is most often used when reconstructing the estimated time-domain signal. Here, we propose “magbook,” “phasebook,” and “combook,” three new types of layers based on discrete representations that can be used to estimate complex time-frequency masks. Magbook layers extend classical sigmoidal units and a recently introduced convex softmax activation for mask-based magnitude estimation. Phasebook layers use a similar structure to give an estimate of the phase mask without suffering from phase wrapping issues. Combook layers are an alternative to the magbook–phasebook combination that directly estimate complex masks. We present various training and inference schemes involving these representations, and explain in particular how to include them in an end-to-end learning framework. We also present an oracle study to assess upper bounds on performance for various types of masks using discrete phase representations. We evaluate the proposed methods on the wsj0-2mix dataset, a well-studied corpus for single-channel speaker-independent speaker separation, matching the performance of state-of-the-art mask-based approaches without requiring additional phase reconstruction steps. [ABSTRACT FROM AUTHOR]

The bang-bang relays of the multiple-boiler system (MBS) control, are characterized by complex limiter saturation functions and classified as fixed parameters. Their action signals cannot precisely control the nonlinear dynamic building heating demand over their entire range of operation. Moreover, in a mono-boiler system, the bang-bang controller endures increasing short cycling over partial load time due to the heating system being considered to have an oversized boiler at most times of running, thus promoting high energy consumption and fluctuating indoor thermal comfort. So, it is difficult to cope with uncertainties in outdoor environments and indoor heating load. Hence, this study formulates the MBS control problem as a dynamic Markov decision process and applies a deep clustering of reinforcement learning approach to obtain the optimal control policy through interaction with the environment based on multi-agent learning according to bang-bang action. With such an approach, adopting a new boiler sequencing control (BSC) strategy using deep clustering of reinforcement learning based on a bang-bang (DCRLBB) manner. The deep clustering is configured to break Lagrangian trajectory curves into piecewise segments to represent the RL agent's action policy. The agent's action policy signals are configured from the bang-bang reward formula based on trade-off implications to be more adjustable than traditional fixed parameters such as fuzzy bang-bang controller (FBBC). The agent of BSC significantly affects the energy performance of the MBS, whereas the other agent resizes boiler capacity by acting to adjust the boiler solenoid fuel valve. The comparison of results between the proposed strategy and conventional FBBC shows distinct differences in the superior response of DCRLBB under dynamic indoor/outdoor actual conditions and energy saving by more than 32% while maintaining the indoor thermal in the comfortable range. [Display omitted] • Proposed strategy based on DCRLBB is used to minimize short cycling on oversized boilers. • DCRLBB provides a viable approach to handling various cooperative policies of multi-agent • Clustering technique helps the well-fitting of the Lagrangian model to optimal agent policy. • Online learning for active decision-making enables the agent to predict the power demand of a multi-boiler. • The comparison of results for the DCRLBB shows energy saving by more than 32%. [ABSTRACT FROM AUTHOR]

Effective customer segmentation and communication of these findings to non-experts is a pressing task in the financial services sector, with the potential for widespread applications. This study employs a three-stage dimension reduction and clustering technique to segment a large, high-dimensional dataset, emphasizing explainability and intuitive visualization. We present the high-dimensional data and feature set using novel network-based visualization methods and identify the multi-stage process's optimal configuration. The approach segments 14,837 potential customers, each with 163 categorical and 143 numerical features. The first stage of the dimension reduction process employs deep neural network-based autoencoders. The second and third stage uses a non-neural network-based dimension reduction algorithm and clustering algorithm contingent on clustering performance. Subsequently, game theory-inspired Shapley values are computed for each feature to enhance explainability. The optimal approach involves an autoencoder, isometric mapping to three dimensions, and K-means clustering. Lastly, we derive investment portfolios for each segment to demonstrate an expert system application in financial investment advisory to underscore the importance of explainable segmentations. [ABSTRACT FROM AUTHOR]

Deep clustering is a popular unsupervised technique for feature representation learning. We recently proposed the chunk-based DeepEmoCluster framework for speech emotion recognition (SER) to adopt the concept of deep clustering as a novel semi-supervised learning (SSL) framework, which achieved improved recognition performances over conventional reconstruction-based approaches. However, the vanilla DeepEmoCluster lacks critical sentence-level temporal information that is useful for SER tasks. This study builds upon the DeepEmoCluster framework, creating a powerful SSL approach that leverages temporal information within a sentence. We propose two sentence-level temporal modeling alternatives using either the temporal-net or the triplet loss function, resulting in a novel temporal-enhanced DeepEmoCluster framework to capture essential temporal information. The key contribution to achieving this goal is the proposed sentence-level uniform sampling strategy, which preserves the original temporal order of the data for the clustering process. An extra network module (e.g., gated recurrent unit) is utilized for the temporal-net option to encode temporal information across the data chunks. Alternatively, we can impose additional temporal constraints by using the triplet loss function while training the DeepEmoCluster framework, which does not increase model complexity. Our experimental results based on the MSP-Podcast corpus demonstrate that the proposed temporal-enhanced framework significantly outperforms the vanilla DeepEmoCluster framework and other existing SSL approaches in regression tasks for the emotional attributes arousal, dominance, and valence. The improvements are observed in fully-supervised learning or SSL implementations. Further analyses validate the effectiveness of the proposed temporal modeling, showing (1) high temporal consistency in the cluster assignment, and (2) well-separated emotional patterns in the generated clusters. [Display omitted] • We propose the novel semi-supervised temporal-enhanced DeepEmoCluster framework. • The model achieves better performance compared to other existing SSL frameworks in SER. • The DeepEmoCluster framework leads to high temporal consistency in the clusters. • The DeepEmoCluster framework leads to well-separable emotional clusters. [ABSTRACT FROM AUTHOR]

Deep clustering has shown its promising capability in joint representation learning and clustering via deep neural networks. Despite the significant progress, the existing deep clustering works mostly utilize some distribution-based clustering loss, lacking the ability to unify representation learning and multi-scale structure learning. To address this, this paper presents a new deep clustering approach termed I mage c luster i ng with c ontrastive le arning and multi-scale G raph C onvolutional N etworks (IcicleGCN), which bridges the gap between convolutional neural network (CNN) and graph convolutional network (GCN) as well as the gap between contrastive learning and multi-scale structure learning for the deep clustering task. Our framework consists of four main modules, namely, the CNN-based backbone, the Instance Similarity Module (ISM), the Joint Cluster Structure Learning and Instance reconstruction Module (JC-SLIM), and the Multi-scale GCN module (M-GCN). Specifically, the backbone network with two weight-sharing views is utilized to learn the representations for the two augmented samples (from each image). The learned representations are then fed to ISM and JC-SLIM for joint instance-level and cluster-level contrastive learning, respectively, during which an auto-encoder in JC-SLIM is also pretrained to serve as a bridge to the M-GCN module. Further, to enforce multi-scale neighborhood structure learning, two streams of GCNs and the auto-encoder are simultaneously trained via (i) the layer-wise interaction with representation fusion and (ii) the joint self-adaptive learning. Experiments on multiple image datasets demonstrate the superior clustering performance of IcicleGCN over the state-of-the-art. The code is available at https://github.com/xuyuankun631/IcicleGCN. • This paper for the first time enables multi-scale structure learning for image clustering via GCNs. • Multi-scale structure learning and two levels of contrastive learning are jointly enforced. • A novel deep image clustering approach termed IcicleGCN is proposed. • Extensive experimental results confirm the superiority of IcicleGCN over the state-of-the-art. [ABSTRACT FROM AUTHOR]

The intelligent buildings provided various incentives to get highly inefficient energy-saving caused by the non-stationary building environments. In the presence of such dynamic excitation with higher levels of nonlinearity and coupling effect of temperature and humidity, the HVAC system transitions from underdamped to overdamped indoor conditions. This led to the promotion of highly inefficient energy use and fluctuating indoor thermal comfort. To address these concerns, this study develops a novel framework based on deep clustering of lagrangian trajectories for multi-task learning (DCLTML) and adding a pre-cooling coil in the air handling unit (AHU) to alleviate a coupling issue. The proposed DCLTML exhibits great overall control and is suitable for multi-objective optimisation based on cooperative multi-agent systems (CMAS). The framework of DCLTML is used greedy iterative training to get an optimal set of weights and tabulated as a layer for each clustering structure. Such layers can deal with the challenges of large space and its massive data. Then the layer weights of each cluster are tuned by the Quasi-Newton (QN) algorithm to make the action sequence of CMAS optimal. Such a policy of CMAS effectively manipulates the inputs of the AHU, where the agents of the AHU activate the natural ventilation and set chillers into an idle state when the outdoor temperature crosses the recommended value. So, it is reasonable to assess the impact potential of thermal mass and hybrid ventilation strategy in reducing cooling energy; accordingly, the assigning results of the proposed DCLTML show that its main cooling coil saves >40% compared to the conventional benchmarks. Besides significant energy savings and improving environmental comfort, the DCLTML exhibits superior high-speed response and robustness performance and eliminates fatigue and wear due to shuttering valves. The results show that the DCLTML algorithm is a promising new approach for controlling HVAC systems. It is more robust to environmental variations than traditional controllers, and it can learn to control the HVAC system in a way that minimises energy consumption. The DCLTML algorithm is still under development, but it can potentially revolutionise how HVAC systems are controlled. [Display omitted] • Clustering based DCLTML structure is used to tackle an extremely large state-action space. • DCLTML provides a viable approach to handling high-dimensional dataset and massive data. • Clustering technique generated to represent a Lagrangian formula for agent action. • The quasi-Newton algorithm is a well-fitting of Lagrangian formula to best agent action. • The results for the DCLTML show saving >40% of main cooling coil energy. [ABSTRACT FROM AUTHOR]

Document clustering plays an important role in text mining and machine learning. Recently, various studies have indicated that a hybrid representation contributes to document clustering tasks. Adaptively adjusting representation learning to make each representation of documents mutually reinforced remains an open problem. In this study, we propose a deep document clustering model, DCAHR, to improve the document clustering performance by adaptively learning a hybrid representation. Specifically, an adaptive representation enhanced network (AREN) is designed using a composite Gated Linear Units (GLU) mechanism to adaptively capture the consistent information of semantic and structure representations, and utilize it to adaptively adjust each representation learning for studying the corresponding enhanced representations. A hybrid representation was then obtained by normalizing the two enhanced representations. Furthermore, an adaptive joint objective function was developed to learn document partitioning and supervise the entire model by adaptively learning the weights of each objective function. Experimental results on nine commonly used real-world datasets demonstrated the outstanding performance of the proposed DCAHR model. • A novel deep document clustering model is proposed by adaptively learning a hybrid representation. • An adaptive representation enhanced network (AREN) is designed. • The AREN learns semantic and structural representations simultaneously. • The AREN adaptively adjusts and mutually reinforces the two representations. • An adaptive joint objective function is designed to supervise the proposed model. [ABSTRACT FROM AUTHOR]

Recently, deep clustering utilizing Graph Neural Networks has shown good performance in the graph clustering. However, the structure information of graph was underused in existing deep clustering methods. Particularly, the lack of concern on mining different types structure information simultaneously. To tackle with the problem, this paper proposes a G raph C lustering Network with S tructure E mbedding E nhanced (GC-SEE) which extracts nodes importance-based and attributes importance-based structure information via a feature attention fusion graph convolution module and a graph attention encoder module respectively. Additionally, it captures different orders-based structure information through multi-scale feature fusion. Finally, a self-supervised learning module has been designed to integrate different types structure information and guide the updates of the GC-SEE. The comprehensive experiments on benchmark datasets commonly used demonstrate the superiority of the GC-SEE. The results showcase the effectiveness of the GC-SEE in exploiting multiple types of structure for deep clustering. • The focus on structural information learned by GCN and GAE varies. • We propose a Graph Clustering Network with Structure Embedding Enhanced (GC-SEE). • GC-SEE integrates different types structure information to enrich the embedding. • A self-supervised loss is designed to achieve clear boundaries and high accuracy. • GC-SEE outperforms the methods using single structure information. [ABSTRACT FROM AUTHOR]

• A structural deep clustering algorithm was employed to classify a total of 2526 breast cancer samples. • This study offers an intriguing classification of breast cancer patients using solely somatic mutation data. • Each subtype of breast cancer was found to be linked with distinct prognosis, biological features, immunogenomic landscape, as well as immunotherapy and chemotherapy implications. Breast cancer is among of the most malignant tumor that occurs in women and is one of the leading causes of death from gynecologic malignancy worldwide. The high degree of heterogeneity that characterizes breast cancer makes it challenging to devise effective therapeutic strategies. Accumulating evidence highlights the crucial role of stratifying breast cancer patients into clinically significant subtypes to achieve better prognoses and treatments. The structural deep clustering network is a graph convolutional network-based clustering algorithm that integrates structural information and has achieved state-of-the-art performance in various applications. In this study, we employed structural deep clustering network to integrate somatic mutation profiles for stratifying 2526 breast cancer patients from the Memorial Sloan Kettering Cancer Center into two clinically differentiable subtypes. Breast cancer patients in cluster 1 exhibited better prognosis than breast cancer patients in cluster 2, and the difference between them was statistically significant. The immunogenomic landscape further demonstrated that cluster 1 was associated with remarkable infiltration of the tumor infiltrating lymphocytes. The clustering subtype could be used to evaluate the therapeutic benefit of immunotherapy and chemotherapy in breast cancer patients. Furthermore, our approach effectively classified patients from eight different cancer types, demonstrating its generalizability. Our study represents a step towards a generic methodology for classifying cancer patients using only somatic mutation data and structural deep clustering network approaches. Employing structural deep clustering network to identify breast cancer subtypes is promising and can inform the development of more accurate and personalized therapies. [ABSTRACT FROM AUTHOR]

Amid the ever-expanding landscape of financial data, the importance of predicting potential risks through artificial intelligence methodologies has steadily risen. To achieve prudent financial data management, this manuscript delves into the domain of intelligent financial risk forecasting within the scope of system design. It presents a data model based on the variational encoder (VAE) enhanced with an attention mechanism meticulously tailored for forecasting a company's financial peril. The framework called the ATT-VAE embarks on its journey by encoding and enhancing multidimensional data through VAE. It then employs the attention mechanism to enrich the outputs of the VAE network, thereby demonstrating the apex of the model's clustering capabilities. In the experimentation, we implemented the model to a battery of training tests using diverse public datasets with multimodal features like AWA and CUB and verified with the local finance dataset. The results conspicuously highlight the model's commendable performance in comparison to publicly available datasets, surpassing numerous deep clustering networks at this juncture. In the realm of financial data, the ATT-VAE model, as presented within this treatise, achieves a clustering accuracy index exceeding 0.7, a feat demonstrably superior to its counterparts in the realm of deep clustering networks. The method outlined herein provides algorithmic foundations and serves as a pivotal reference for the prospective domain of intelligent financial data governance and scrutiny., Competing Interests: The author declares that he has no competing interests., (©2024 Huang and Lu.)

Machine learning applications are widespread due to straightforward supervised learning of known data labels. Many data samples in real-world scenarios, including medicine, are unlabeled because data annotation can be time-consuming and error-prone. The application and evaluation of unsupervised clustering methods are not trivial and are limited to traditional methods (e.g., k-means) when clinicians demand deeper insights into patient data beyond classification accuracy. The contribution of this paper is three-fold: 1) to introduce a patient stratification strategy based on a clinical variable instead of a diagnostic label, 2) to evaluate clustering performance using within-cluster homogeneity and between-cluster statistical difference, and 3) to compare widely used traditional clustering algorithms (e.g., k-means) with a state-of-the-art deep learning solution for clustering tabular data. The deep clustering method achieves superior within-cluster homogeneity and between-cluster separation compared to k-means and identifies three statistically distinct and clinically interpretable high blood pressure patient clusters. The proposed clustering strategy and evaluation metrics will facilitate the stratification of large patient cohorts in health science research without requiring explicit diagnostic labels.

Deep clustering generally obtains promising performance by learning deep feature representations. However, there are two limitations: (1) end-to-end deep density clustering needs to be explored; (2) prior information is ignored to guide the learning process. To overcome these limitations, we propose a novel semi-supervised deep density clustering (SDDC). Specifically, a convolutional autoencoder is applied to learn embedded features, and semi-supervised density peaks clustering is designed to identify stable cluster centers. Meanwhile, prior information is introduced to instruct the preferable clustering process. By integrating prior information, a joint clustering loss is directly built on embedded features to perform feature representation and cluster assignment simultaneously. Extensive experiments validate the power of SDDC for initializing and the effectiveness on clustering tasks. • This is the first end-to-end framework to extend deep clustering to density-based clustering. • The semi-supervised density-based clustering algorithm aims to stably identify cluster centers. • The experiment illustrate that the proposed framework achieves competitive clustering results on high-dimensional datasets. [ABSTRACT FROM AUTHOR]

Deep clustering plays an important role in data analysis, and with the prevalence of graph data nowadays, various deep clustering models on graph are constantly proposed. However, due to the lack of more adequate clustering guidance, the discriminability of feature representation learned from these models for the clustering task is limited. Therefore, for the purpose of enabling the output of these models to be more cluster-oriented, we propose a Deep Clustering Optimization Method for Graph Neural Networks (DCOM-GNN), which can be attached to the original model architecture conveniently. For DCOM-GNN, it contains two components, one is the inter-cluster distance optimization module, whose role is to further adjust the distance between clusters of the original model output rationally. Another one is the intra-cluster distance optimization module, which aims to improve the cohesiveness of the original model output. Comprehensive experiments show that the performance of various deep clustering models on graph can be significantly improved after adding DCOM-GNN. [ABSTRACT FROM AUTHOR]

As an indispensable branch of unsupervised learning, deep clustering is rapidly emerging along with the growth of deep neural networks. Recently, contrastive learning paradigm has been combined with deep clustering to achieve more competitive performance. However, previous works mostly employ random augmentations to construct sample pairs for contrastive clustering. Different augmentations of a sample are treated as positive sample pairs, which may result in false positives and ignore the semantic variations of different samples. To address these limitations, we present a novel end-to-end contrastive clustering framework termed Contrastive Clustering with Effective Sample pairs construction (CCES), which obtains more semantic information by jointly leveraging an effective data augmentation method ContrastiveCrop and constructing positive sample pairs based on nearest-neighbor mining. Specifically, we augment original samples by adopting ContrastiveCrop, which explicitly reduces false positives and enlarges the variance of samples. Further, with the extracted feature representations, we provide a strategy to construct positive sample pairs via a sample and its nearest neighbor for instance-wise and cluster-wise contrastive learning. Experimental results on four challenging datasets demonstrate the effectiveness of CCES for clustering, which surpasses the state-of-the-art deep clustering methods. • We propose Contrastive Clustering with Effective Sample pairs construction (CCES). • CCES constructs sample pairs via the nearest neighbor in contrastive clustering. • CCES adopts data augmentation strategy ContrastiveCrop for contrastive clustering. • CCES obtains the superior performance than state-of-the-art clustering method. [ABSTRACT FROM AUTHOR]

Modern industrial facilities generate large volumes of raw sensor data during the production process. This data is used to monitor and control the processes and can be analyzed to detect and predict process abnormalities. Typically, the data has to be annotated by experts in order to be used in predictive modeling. However, manual annotation of large amounts of data can be difficult in industrial settings. In this paper, we propose SensorSCAN, a novel method for unsupervised fault detection and diagnosis, designed for industrial chemical process monitoring. We demonstrate our model's performance on two publicly available datasets of the Tennessee Eastman Process with various faults. The results show that our method significantly outperforms existing approaches (+0.2-0.3 TPR for a fixed FPR) and effectively detects most of the process faults without expert annotation. Moreover, we show that the model fine-tuned on a small fraction of labeled data nearly reaches the performance of a SOTA model trained on the full dataset. We also demonstrate that our method is suitable for real-world applications where the number of faults is not known in advance. The code is available at https://github.com/AIRI-Institute/sensorscan. [ABSTRACT FROM AUTHOR]

• A time series segmentation method (APCA) was presented for extracting stable rock-machine interaction data. • A novel deep clustering model (SSAE-FCM) focus on operational parameters was proposed. • A surrounding rock classification suitable for TBM was obtained. • The superiority of the proposed method and the importance of the selected features were discussed. Accurate, continuous real-time perception of surrounding rock conditions on the tunnel boring machine (TBM) tunnel face provides particularly important prior information to ensure safe, economic, and efficient tunneling. This study presents an intelligent method for assessing the TBM surrounding rock conditions using a combination of time series segmentation and a deep clustering model (SSAE-FCM). In data preprocessing, fifteen operational parameters were selected as input features. And an adaptive piecewise constant approximation method was employed for the segmentation and compression of stable phase data. Then, a total of 69,214 data samples that can characterize the stable rock-machine interaction were obtained to construct a dataset. Thereafter, the SSAE-FCM model was established and trained to classify the surrounding rock conditions into four clusters. The distribution result of the main operational parameters and composite indexes (TPI, FPI, and SE) within these clusters demonstrated that the classification results are compatible with rock-machine interaction in engineering practice. Additionally, the rock mass assessment performance of the classification results was compared to BQ classification in terms of various rock mass parameters (UCS, Jv), fault sections, and collapse sections. These results indicate that the proposed classification method accurately reflects the stability and boreability of rock mass and is more sensitive to unfavorable geology than the BQ classification. From the three clustering evaluation indexes (SC, CHI, and DBI) perspective, SSAE-FCM was superior to the other clustering models (BIRCH + k-means++, FCM, and SSAE + FCM). In conclusion, the proposed intelligent method for TBM surrounding rock classification is reasonable and applicable for the evaluation of rock mass quality, as well as accurate and sensitive in real-time perception. [ABSTRACT FROM AUTHOR]

Single-cell RNA sequencing (ScRNA-seq) technology reveals gene expression information at the cellular level. The critical tasks in ScRNA-seq data analysis are clustering and dimensionality reduction. Recent deep clustering algorithms are used to optimize the two tasks jointly, and their variations, graph-based deep clustering algorithms, are used to capture and preserve topological information in the process. However, the existing graph-based deep clustering algorithms ignore the distribution information of nodes when constructing cell graphs which leads to incomplete information in the embedding representation; and graph convolutional networks (GCN), which are most commonly used, often suffer from over-smoothing that leads to high sample similarity in the embedding representation and then poor clustering performance. Here, the dual-GCN-based deep clustering with Triplet contrast (scDGDC) is proposed for dimensionality reduction and clustering of scRNA-seq data. Two critical components are dual-GCN-based encoder for capturing more comprehensive topological information and triplet contrast for reducing GCN over-smoothing. The two components improve the dimensionality reduction and clustering performance of scDGDC in terms of information acquisition and model optimization, respectively. The experiments on eight real ScRNA-seq datasets showed that scDGDC achieves excellent performance for both clustering and dimensionality reduction tasks and is high robustness to parameters. [Display omitted] • Dual-GCN-based encoder captures distance and distribution information so embedding representation preserves comprehensive information. • Triplet contrast reduces over-smoothing of GCN so avoids high similarity of cells in embedding space. • Proposed scDGDC shows excellent performance on clustering and dimensionality reduction and strong robustness to parameters. [ABSTRACT FROM AUTHOR]

[Display omitted] • The noise increment affects the quality of identifying the ERP components. • The designed deep clustering methods can learn the most powerful features of ERP. • The time window of the ERP can be reliably qualified using the proposed method. • Proposed ensemble learning provides a complementary understanding of ERP data. Cluster analysis of spatio-temporal event-related potential (ERP) data is a promising tool for exploring the measurement time window of ERPs. However, even after preprocessing, the remaining noise can result in uncertain cluster maps followed by unreliable time windows while clustering via conventional clustering methods. We designed an ensemble deep clustering pipeline to determine a reliable time window for the ERP of interest from temporal concatenated grand average ERP data. The proposed pipeline includes semi-supervised deep clustering methods initialized by consensus clustering and unsupervised deep clustering methods with end-to-end architectures. Ensemble clustering from those deep clusterings was used by the designed adaptive time window determination to estimate the time window. After applying simulated and real ERP data, our method successfully obtained the time window for identifying the P3 components (as the interest of both ERP studies) while additional noise (e.g., adding 20 dB to −5 dB white Gaussian noise) was added to the prepared data. Compared to the state-of-the-art clustering methods, a superior clustering performance was yielded from both ERP data. Furthermore, more stable and precise time windows were elicited as the noise increased. Our study provides a complementary understanding of identifying the cognitive process using deep clustering analysis to the existing studies. Our finding suggests that deep clustering can be used to identify the ERP of interest when the data is imperfect after preprocessing. [ABSTRACT FROM AUTHOR]

In medical image datasets, discrete labels are often used to describe a continuous spectrum of conditions, making unsupervised image stratification a challenging task. In this work, we propose VAESim, an architecture for image stratification based on a conditional variational autoencoder. VAESim learns a set of prototypical vectors during training, each associated with a cluster in a continuous latent space. We perform a soft assignment of each data sample to the clusters and reconstruct the sample based on a similarity measure between the sample embedding and the prototypical vectors. To update the prototypical embeddings, we use an exponential moving average of the most similar representations between actual prototypes and samples in the batch size. We test our approach on the MNIST handwritten digit dataset and the PneumoniaMNIST medical benchmark dataset, where we show that our method outperforms baselines in terms of kNN accuracy (up to + 15 % improvement in performance) and performs at par with classification models trained in a fully supervised way. Our model also outperforms current end-to-end models for unsupervised stratification. [Display omitted] • Discover prototypes and clusters in image data with VAE's new prototypes matrix. • Improve VAE's latent space for clustering, sampling, and downstream tasks. • Novel and simple approach, outperforms other deep clustering algorithms • Self-supervised training enables data generation, reconstruction, and learning • The model is benchmarked on famous images dataset, including a medical dataset. [ABSTRACT FROM AUTHOR]

The main challenge of an intrusion detection system (IDS) is detecting novelties (i.e., zero-day attacks) in addition to generating a report about the known attacks (i.e., classifying known attacks). Another challenge of an IDS is the adaptation to the detected novelties. For this purpose, it needs sufficient labeled samples of the new attacks. On the other hand, the labeling procedure is a time-consuming task for the security expert teams. This paper proposes a DL-based IDS framework adaptable to new attacks, consisting of different phases. The first phase uses deep learning-based open set recognition methods to identify unknown samples (i.e., new attacks), and make a report from different known attacks simultaneously. Then, the novel samples are clustered by combining the deep model and clustering algorithms. These clusters are the main key to making the labeling procedure more practical and reducing the time and effort of the expert knowledge team. Finally, we use the labeled groups to update the model and make it adaptable to new traffic behaviors. To evaluate the proposed framework, we compare our proposed DOC++ alongside different algorithms (including DOC, OpenMax, AutoSVM, and CROSR) in the open set recognition phase of the framework and use both the CIC-IDS2017 and CSE-CIC-IDS2018 data sets for the evaluation. Our results show that DOC++ achieves the best performance among the other open set recognition approaches. Besides, the completeness and homogeneity of the clustering phase prove that the generated labeled groups are good enough for the supervised labeling and updating phases. [ABSTRACT FROM AUTHOR]

• A new deep learning-based clustering method is proposed to deal with the large number of daily load curves. • A novel customer selection framework is proposed to identify the potential good candidates for demand response programs. • The proposed clustering method outperforms the classical K-means and two state-of-the-art deep clustering methods. • The customers selected by the proposed framework contribute to 59 % of total peak reduction with only accounting for 23 % of population. Demand response (DR) is regarded as a promising solution to the problem of renewable energy integration, while it remains one of the key barriers for DR to target the right customers, i.e. the customers with high potential in peak reduction. Exploring the electricity consumption behaviors of customers can help DR program practitioners evaluate customers' peak reduction potential. Clustering has been the most common approach to describe customers' consumption behaviors in the literature. However, the existing clustering methods suffer from either unscalability or poor performance when handling a large number of daily load curves. To overcome these issues, a new deep learning-based clustering method is proposed in this paper, which integrates both intra-cluster compactness and inter-cluster separation into the objective function of clustering for consideration. Based on the clustering results, a new customer selection framework is developed to identify the potential good candidates for DR programs, which takes the stability of consumption behaviors and the characteristics at peak hours into account. Case studies on a London dataset demonstrate the superiority of the proposed clustering method over classical K-means and two state-of-the-art deep clustering methods in terms of clustering validity indexes. Furthermore, the simulations performed on an actual price-based pilot show that the customers selected by the proposed customer selection framework contribute to 59 % of total peak reduction with only accounting for 23 % of population. [ABSTRACT FROM AUTHOR]

• A novel deep evolutionary clustering (ADSMTDE) to overcome clustering drawbacks. • Improving the auto-encoder by applying sparsity constraint and manifold learning. • To enhance clustering, evolutionary algorithm is adopted to optimize solutions. • Employing an incremental clustering technique to perform clustering dynamically. • ADSMTDE is competitive and superior to over the latest deep clustering methods. Most deep clustering methods despite utilizing complex networks to learn better from data, use a shallow clustering method. These methods have difficulty in finding good clusters due to the lack of ability to handle between local search and global search to prevent premature convergence. In other words, they do not consider different aspects of the search and it causes them to get stuck in the local optimum. In addition, the majority of existing deep clustering approaches perform clustering with the knowledge of the number of clusters, which is not practical in most real scenarios where such information is not available. To address these problems, this paper presents a novel automatic deep sparse clustering approach based on an evolutionary algorithm called Multi-Trial Vector-based Differential Evolution (MTDE). Sparse auto-encoder is first applied to extract embedded features. Manifold learning is then adopted to obtain representation and extract the spatial structure of features. Afterward, MTDE clustering is performed without prior information on the number of clusters to find the optimal clustering solution. The proposed approach was evaluated on various datasets, including images and time-series. The results demonstrate that the proposed method improved MTDE by 18.94% on average and compared to the most recent deep clustering algorithms, is consistently among the top three in the majority of datasets. [ABSTRACT FROM AUTHOR]

The superior performance of deep learning in different domains has sparked significant interest in its applicability to 3D computer graphics. Deep learning has become the dominant technical architecture in current 3D mesh segmentation. However, learning-based 3D segmentation methods usually rely on high-quality training datasets, which are not readily available in practical applications. How to segment 3D meshes without exhaustive label annotations remains a challenging problem, especially in the context of deep learning. As a subset of unsupervised learning methods, self-supervised learning offers a promising learning paradigm for unlabeled 3D mesh segmentation. In this paper, we introduce a self-supervised clustering-Based network specifically for the segmentation of label-free 3D meshes. Our self-supervised clustering-based 3D mesh segmentation network (SCMS-Net) employs a two-branch architecture to learn effective feature representation. The two branches are unified into an end-to-end framework using a self-supervised strategy. Finally, the label predictions of the parts are generated by iterative clustering. We conducted ablation studies and comparative experiments on a standard benchmark to demonstrate the effectiveness of our approach. • This is the first work to explore the self-supervised clustering methods for mesh segmentation. • We use a two-branch network to integrate geometric and structural information. • The segmentation loss function consists of the clustering and reconstruction loss. [ABSTRACT FROM AUTHOR]

• We propose a deep statistical modeling to image data representation. • We construct a cluster assignment module based on the EM algorithm. • The model jointly encodes image features and estimates GMM in the feature space. • Experiments demonstrate the superiority of the proposed DEM-Net-S and DEM-Net-C. Unsupervised learning, such as unsupervised image segmentation and clustering, are fundamental tasks in image representation learning. In this paper, we design a deep expectation-maximization (DEM) network for unsupervised image segmentation and clustering. It is based on the statistical modeling of image in its latent feature space by Gaussian mixture model (GMM), implemented in a novel deep learning framework. Specifically, in the unsupervised setting, we design an auto-encoder network and an EM module over the image latent features, for jointly learning the image latent features and GMM model of the latent features in a single framework. To construct the EM-module, we unfold the iterative operations of EM algorithm and the online EM algorithm in fixed steps to be differentiable network blocks, plugged into the network to estimate the GMM parameters of the image latent features. The proposed network parameters can be end-to-end optimized using losses based on log-likelihood of GMM, entropy of Gaussian component assignment probabilities and image reconstruction error. Extensive experiments confirm that our proposed networks achieve favorable results compared with several state-of-the-art methods in unsupervised image segmentation and clustering. [ABSTRACT FROM AUTHOR]

• This paper jointly exploits strong and weak augmentations for unsupervised image clustering. • Two levels of contrastive learning are conducted on three streams of weak/strong augmented views. • A novel end-to-end deep clustering approach termed SACC is proposed. • Extensive experimental results confirm the superiority of SACC over the state-of-the-art. Deep clustering has attracted increasing attention in recent years due to its capability of joint representation learning and clustering via deep neural networks. In its latest developments, the contrastive learning has emerged as an effective technique to substantially enhance the deep clustering performance. However, the existing contrastive learning based deep clustering algorithms mostly focus on some carefully-designed augmentations (often with limited transformations to preserve the structure), referred to as weak augmentations, but cannot go beyond the weak augmentations to explore the more opportunities in stronger augmentations (with more aggressive transformations or even severe distortions). In this paper, we present an end-to-end deep clustering approach termed S trongly A ugmented C ontrastive C lustering (SACC), which extends the conventional two-augmentation-view paradigm to multiple views and jointly leverages strong and weak augmentations for strengthened deep clustering. Particularly, we utilize a backbone network with triply-shared weights, where a strongly augmented view and two weakly augmented views are incorporated. Based on the representations produced by the backbone, the weak-weak view pair and the strong-weak view pairs are simultaneously exploited for the instance-level contrastive learning (via an instance projector) and the cluster-level contrastive learning (via a cluster projector), which, together with the backbone, can be jointly optimized in a purely unsupervised manner. Experimental results on five challenging image datasets have shown the superiority of our SACC approach over the state-of-the-art. The code is available at https://github.com/dengxiaozhi/SACC. [ABSTRACT FROM AUTHOR]

• We propose a novel modular deep clustering framework combining feature extractor, clustering layer, sample selecting module and classifier. • With this framework, the clustering model get evolved with the assistance of the neural network classifier in an unsupervised manner. • Fuzzy theory is introduced into clustering to model the fuzzy membership of samples which provides a natural measure for sample confidence. • Off-the-shelf neural network model is adopted as the self-supervised classifier to generate the target distribution instead of the simple nonlinear transformation function in existing deep clustering studies. Deep clustering outperforms conventional clustering by mutually promoting representation learning and cluster assignment. However, most existing deep clustering methods suffer from two major drawbacks. Firstly, most cluster assignment methods are highly dependent on the intermediate target distribution generated by a handcrafted nonlinear mapping function. Secondly, the clustering results can be easily guided towards wrong direction by the misassigned samples in each cluster. The existing deep clustering methods are incapable of discriminating such samples. These facts largely limit the possible performance that deep clustering methods can reach. To address these issues, a novel Self-Supervised Clustering (SSC) framework is constructed, which boosts the clustering performance by classification in an unsupervised manner. Fuzzy theory is used to score the membership of each sample to the clusters in terms of probability in each training epoch, which evaluates the intermediate clustering result certainty of each sample. The most reliable samples can be selected with the help of a sample selection method according to the membership and enhanced by data augmentation method. These augmented data are employed to fine-tune an off-the-shelf deep network classifier with the labels provided by the clustering in a self-supervised way. The classification results of the original dataset are used as the target distribution to guide the training process of the deep clustering model. The proposed framework can efficiently discriminate sample outliers and generate better target distribution with the assistance of the powerful classifier. Extensive experiments indicate that the proposed framework remarkably outperforms state-of-the-art deep clustering methods on four benchmark datasets. [ABSTRACT FROM AUTHOR]

Background: Physiological motion, such as respiratory motion, has become a limiting factor in the spatial resolution of positron emission tomography (PET) imaging as the resolution of PET detectors continue to improve. Motion-induced misregistration between PET and CT images can also cause attenuation correction artifacts. Respiratory gating can be used to freeze the motion and to reduce motion induced artifacts., Purpose: In this study, we propose a robust data-driven approach using an unsupervised deep clustering network that employs an autoencoder (AE) to extract latent features for respiratory gating., Methods: We first divide list-mode PET data into short-time frames. The short-time frame images are reconstructed without attenuation, scatter, or randoms correction to avoid attenuation mismatch artifacts and to reduce image reconstruction time. The deep AE is then trained using reconstructed short-time frame images to extract latent features for respiratory gating. No additional data are required for the AE training. K-means clustering is subsequently used to perform respiratory gating based on the latent features extracted by the deep AE. The effectiveness of our proposed Deep Clustering method was evaluated using physical phantom and real patient datasets. The performance was compared against phase gating based on an external signal (External) and image based principal component analysis (PCA) with K-means clustering (Image PCA)., Results: The proposed method produced gated images with higher contrast and sharper myocardium boundaries than those obtained using the External gating method and Image PCA. Quantitatively, the gated images generated by the proposed Deep Clustering method showed larger center of mass (COM) displacement and higher lesion contrast than those obtained using the other two methods., Conclusions: The effectiveness of our proposed method was validated using physical phantom and real patient data. The results showed our proposed framework could provide superior gating than the conventional External method and Image PCA., (© 2023 The Authors. Medical Physics published by Wiley Periodicals LLC on behalf of American Association of Physicists in Medicine.)

Recently, deep clustering has been extensively employed for various data mining tasks, and it can be divided into auto-encoder (AE)-based and graph neural networks (GNN)-based methods. However, existing AE-based methods fall short in effectively extracting structural information, while GNN suffer from smoothing and heterophily. Although methods that combine AE and GNN achieve impressive performance, there remains an inadequate balance between preserving the raw structure and exploring the underlying structure. Accordingly, we propose a novel network named Structure-Aware Deep Clustering network (SADC). Firstly, we compute the cumulative influence of non-adjacent nodes at multiple depths and, thus, enhance the adjacency matrix. Secondly, an enhanced graph auto-encoder is designed. Thirdly, the latent space of AE is endowed with the ability to perceive the raw structure during the learning process. Besides, we design self-supervised mechanisms to achieve co-optimization of node representation learning and topology learning. A new loss function is designed to preserve the inherent structure while also allowing for exploration of latent data structure. Extensive experiments on six benchmark datasets validate that our method outperforms state-of-the-art methods., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2023 Elsevier Ltd. All rights reserved.)

Attention deficit hyperactivity disorder (ADHD) is one prevalent neurodevelopmental disorder with childhood onset, however, there is no clear correspondence established between clinical ADHD subtypes and primary medications. Identifying objective and reliable neuroimaging markers for categorizing ADHD biotypes may lead to more individualized, biotype-guided treatment. Here we proposed graph convolutional network plus deep clustering for ADHD biotype detection using functional network connectivity (FNC), resulting in two biotypes based on 1069 ADHD patients selected from Adolescent Brain and Cognitive Development (ABCD) study, which were well replicated on independent ADHD adolescents undergoing longitudinal medication treatment (n=130). Interestingly, in addition to differences in cognitive performance and hyperactivity/impulsivity symptoms, biotype 1 treated with methylphenidate demonstrated significantly better recovery than biotype 2 treated with atomoxetine (p<0.05, FDR corrected). This imaging-driven, biotype-guided approach holds promise for facilitating personalized treatment of ADHD, exploring possible boundaries through innovative deep learning algorithms aimed at improving medication treatment effectiveness., Competing Interests: Additional Declarations: There is NO Competing Interest.