Unsupervised Learning Problem Research Articles

A unified deep semi-supervised graph learning scheme based on nodes re-weighting and manifold regularization

In recent years, semi-supervised learning on graphs has gained importance in many fields and applications. The goal is to use both partially labeled data (labeled examples) and a large amount of unlabeled data to build more effective predictive models. Deep Graph Neural Networks (GNNs) are very useful in both unsupervised and semi-supervised learning problems. As a special class of GNNs, Graph Convolutional Networks (GCNs) aim to obtain data representation through graph-based node smoothing and layer-wise neural network transformations. However, GCNs have some weaknesses when applied to semi-supervised graph learning: (1) it ignores the manifold structure implicitly encoded by the graph; (2) it uses a fixed neighborhood graph and focuses only on the convolution of a graph, but pays little attention to graph construction; (3) it rarely considers the problem of topological imbalance.To overcome the above shortcomings, in this paper, we propose a novel semi-supervised learning method called Re-weight Nodes and Graph Learning Convolutional Network with Manifold Regularization (ReNode-GLCNMR). Our proposed method simultaneously integrates graph learning and graph convolution into a unified network architecture, which also enforces label smoothing through an unsupervised loss term. At the same time, it addresses the problem of imbalance in graph topology by adaptively reweighting the influence of labeled nodes based on their distances to the class boundaries. Experiments on 8 benchmark datasets show that ReNode-GLCNMR significantly outperforms the state-of-the-art semi-supervised GNN methods.11The code is available at https://github.com/BiJingjun/ReNode-GLCNMR.

Koopman-Based Spectral Clustering of Directed and Time-Evolving Graphs

While spectral clustering algorithms for undirected graphs are well established and have been successfully applied to unsupervised machine learning problems ranging from image segmentation and genome sequencing to signal processing and social network analysis, clustering directed graphs remains notoriously difficult. Two of the main challenges are that the eigenvalues and eigenvectors of graph Laplacians associated with directed graphs are in general complex-valued and that there is no universally accepted definition of clusters in directed graphs. We first exploit relationships between the graph Laplacian and transfer operators and in particular between clusters in undirected graphs and metastable sets in stochastic dynamical systems and then use a generalization of the notion of metastability to derive clustering algorithms for directed and time-evolving graphs. The resulting clusters can be interpreted as coherent sets, which play an important role in the analysis of transport and mixing processes in fluid flows.Graphic

Boosted-oriented probabilistic smoothing-spline clustering of series

Fuzzy clustering methods allow the objects to belong to several clusters simultaneously, with different degrees of membership. However, a factor that influences the performance of fuzzy algorithms is the value of fuzzifier parameter. In this paper, we propose a fuzzy clustering procedure for data (time) series that does not depend on the definition of a fuzzifier parameter. It comes from two approaches, theoretically motivated for unsupervised and supervised classification cases, respectively. The first is the Probabilistic Distance clustering procedure. The second is the well known Boosting philosophy. Our idea is to adopt a boosting prospective for unsupervised learning problems, in particular we face with non hierarchical clustering problems. The global performance of the proposed method is investigated by various experiments.

Out‐of‐bag stability estimation for k‐means clustering

AbstractClustering data is a challenging problem in unsupervised learning where there is no gold standard. Results depend on several factors, such as the selection of a clustering method, measures of dissimilarity, parameters, and the determination of the number of reliable groupings. Stability has become a valuable surrogate to performance and robustness that can provide insight to an investigator on the quality of a clustering, and guidance on subsequent cluster prioritization. This work develops a framework for stability measurements that is based on resampling and OB estimation. Bootstrapping methods for cluster stability can be prone to overfitting in a setting that is analogous to poor delineation of test and training sets in supervised learning. Stability that relies on OB items from a resampling overcomes these issues and does not depend on a reference clustering for comparisons. Furthermore, OB stability can provide estimates at the level of the item, cluster, and as an overall summary, which has good interpretive value. This framework is extended to develop stability estimates for determining the number of clusters (model selection) through contrasts between stability estimates on clustered data, and stability estimates of clustered reference data with no signal. These contrasts form stability profiles that can be used to identify the largest differences in stability and do not require a direct threshold on stability values, which tend to be data specific. These approaches can be implemented using the R package bootcluster that is available on the Comprehensive R Archive Network.

Quantum spectral clustering algorithm for unsupervised learning

Clustering is one of the most crucial problems in unsupervised learning, and the well-known $k$-means clustering algorithm has been shown to be implementable on a quantum computer with a significant speedup. However, many clustering problems cannot be solved by $k$-means, and a powerful method called spectral clustering is introduced to solve these problems. In this work, we propose a circuit design to implement spectral clustering on a quantum processor with a substantial speedup, by initializing the processor into a maximally entangled state and encoding the data information into an efficiently-simulatable Hamiltonian. Compared with the established quantum $k$-means algorithms, our method does not require a quantum random access memory or a quantum adiabatic process. It relies on an appropriate embedding of quantum phase estimation into Grover's search to gain the quantum speedup. Simulations demonstrate that our method is effective in solving clustering problems and will serve as an important supplement to quantum $k$-means for unsupervised learning.

Mechanical intelligence for learning embodied sensor-object relationships

Intelligence involves processing sensory experiences into representations useful for prediction. Understanding sensory experiences and building these contextual representations without prior knowledge of sensor models and environment is a challenging unsupervised learning problem. Current machine learning methods process new sensory data using prior knowledge defined by either domain knowledge or datasets. When datasets are not available, data acquisition is needed, though automating exploration in support of learning is still an unsolved problem. Here we develop a method that enables agents to efficiently collect data for learning a predictive sensor model—without requiring domain knowledge, human input, or previously existing data—using ergodicity to specify the data acquisition process. This approach is based entirely on data-driven sensor characteristics rather than predefined knowledge of the sensor model and its physical characteristics. We learn higher quality models with lower energy expenditure during exploration for data acquisition compared to competing approaches, including both random sampling and information maximization. In addition to applications in autonomy, our approach provides a potential model of how animals use their motor control to develop high quality models of their sensors (sight, sound, touch) before having knowledge of their sensor capabilities or their surrounding environment.

Wuthering heights: gauging fear at altitude in virtual reality

In this study we propose an approach to assess the fear of heights through a 3D virtual reality environment. We show that an immersive scenario provides a suitable infrastructure to such purpose, when supported by related behavioural and physiological measurements. Our approach is grounded in the principled framework of constructed emotions. This allows to shape fear detection as a case of categorical perception, which is amenable to be formalised as an unsupervised learning problem. Meanwhile, it paves the way for addressing meaningful physiological parameters for the assessment. Gauging fear of heights in individuals, beyond its theoretical relevance, is cogent for the early discernment of workers who are unsuited for operating at altitude and who may require to undergo specific training or, eventually, to be recruited for different positions.

Unsupervised Reinforcement Learning in Multiple Environments

Several recent works have been dedicated to unsupervised reinforcement learning in a single environment, in which a policy is first pre-trained with unsupervised interactions, and then fine-tuned towards the optimal policy for several downstream supervised tasks defined over the same environment. Along this line, we address the problem of unsupervised reinforcement learning in a class of multiple environments, in which the policy is pre-trained with interactions from the whole class, and then fine-tuned for several tasks in any environment of the class. Notably, the problem is inherently multi-objective as we can trade off the pre-training objective between environments in many ways. In this work, we foster an exploration strategy that is sensitive to the most adverse cases within the class. Hence, we cast the exploration problem as the maximization of the mean of a critical percentile of the state visitation entropy induced by the exploration strategy over the class of environments. Then, we present a policy gradient algorithm, alphaMEPOL, to optimize the introduced objective through mediated interactions with the class. Finally, we empirically demonstrate the ability of the algorithm in learning to explore challenging classes of continuous environments and we show that reinforcement learning greatly benefits from the pre-trained exploration strategy w.r.t. learning from scratch.

RepBin: Constraint-Based Graph Representation Learning for Metagenomic Binning

Mixed communities of organisms are found in many environments -- from the human gut to marine ecosystems -- and can have profound impact on human health and the environment. Metagenomics studies the genomic material of such communities through high-throughput sequencing that yields DNA subsequences for subsequent analysis. A fundamental problem in the standard workflow, called binning, is to discover clusters, of genomic subsequences, associated with the constituent organisms. Inherent noise in the subsequences, various biological constraints that need to be imposed on them and the skewed cluster size distribution exacerbate the difficulty of this unsupervised learning problem. In this paper, we present a new formulation using a graph where the nodes are subsequences and edges represent homophily information. In addition, we model biological constraints providing heterophilous signal about nodes that cannot be clustered together. We solve the binning problem by developing new algorithms for (i) graph representation learning that preserves both homophily relations and heterophily constraints (ii) constraint-based graph clustering method that addresses the problems of skewed cluster size distribution. Extensive experiments, on real and synthetic datasets, demonstrate that our approach, called RepBin, outperforms a wide variety of competing methods. Our constraint-based graph representation learning and clustering methods, that may be useful in other domains as well, advance the state-of-the-art in both metagenomics binning and graph representation learning.

A Hyperparameter-Free, Fast and Efficient Framework to Detect Clusters From Limited Samples Based on Ultra High-Dimensional Features.

Clustering is a challenging problem in machine learning in which one attempts to group N objects into K0 groups based on P features measured on each object. In this article, we examine the case where N ≪ P and K0 is not known. Clustering in such high dimensional, small sample size settings has numerous applications in biology, medicine, the social sciences, clinical trials, and other scientific and experimental fields. Whereas most existing clustering algorithms either require the number of clusters to be known a priori or are sensitive to the choice of tuning parameters, our method does not require the prior specification of K0 or any tuning parameters. This represents an important advantage for our method because training data are not available in the applications we consider (i.e., in unsupervised learning problems). Without training data, estimating K0 and other hyperparameters-and thus applying alternative clustering algorithms-can be difficult and lead to inaccurate results. Our method is based on a simple transformation of the Gram matrix and application of the strong law of large numbers to the transformed matrix. If the correlation between features decays as the number of features grows, we show that the transformed feature vectors concentrate tightly around their respective cluster expectations in a low-dimensional space. This result simplifies the detection and visualization of the unknown cluster configuration. We illustrate the algorithm by applying it to 32 benchmarked microarray datasets, each containing thousands of genomic features measured on a relatively small number of tissue samples. Compared to 21 other commonly used clustering methods, we find that the proposed algorithm is faster and twice as accurate in determining the "best" cluster configuration.

Unsupervised community detection in attributed networks based on mutual information maximization

Community detection is of great significance for understanding network functions and behaviors. With the successful application of deep learning in network-based analyses, recent studies have turned to utilizing graph convolutional networks (GCNs) to this problem due to their capability in capturing network attributes. Nevertheless, most existing GCN-based community detection approaches are semi-supervised and local structure-aware, even though community detection is an unsupervised learning problem essentially. In this paper, we develop a novel GCN method for unsupervised community detection under the framework of mutual information (MI) maximization, called UCDMI. Specifically, a novel MI maximization mechanism is developed to capture more fine-grained information of the global network structure in an unsupervised manner. Moreover, a new aggregation function is proposed for GCN to distinguish the importance between different neighboring nodes, which enables our method to identify more high-quality node representations and improve the community detection performance. Our extensive experiments demonstrate the effectiveness of our proposed UCDMI compared with several state-of-the-art community detection methods.

Learning image embeddings without labels

Can we automatically learn discriminative embedding features from images when human-annotated labels are absent? The problem of unsupervised embedded learning remains a significant and open challenge in image and vision community. A joint online deep embedded clustering and hard samples mining framework are proposed to improve the representation ability of embedded learning. In addition, to enhance the discriminability of feature representations, a structure-level pair-based loss is introduced to take full advantage of structure correlation between all the mined hard samples. Finally, the quantitative results of exhaustive experiments on three benchmarks show that our proposed method performs better than existing state-of-the-art methods.

A Novel and Scalable Spatio-Temporal Technique for Ocean Eddy Monitoring

Swirls of ocean currents known as ocean eddies are a crucial component of the ocean's dynamics. In addition to dominating the ocean's kinetic energy, eddies play a significant role in the transport of water, salt, heat, and nutrients. Therefore, understanding current and future eddy patterns is a central climate challenge to address future sustainability of marine ecosystems. The emergence of sea surface height observations from satellite radar altimeter has recently enabled researchers to track eddies at a global scale. The majority of studies that identify eddies from observational data employ highly parametrized connected component algorithms using expert filtered data, effectively making reproducibility and scalability challenging. In this paper, we frame the challenge of monitoring ocean eddies as an unsupervised learning problem. We present a novel change detection algorithm that automatically identifies and monitors eddies in sea surface height data based on heuristics derived from basic eddy properties. Our method is accurate, efficient, and scalable. To demonstrate its performance we analyze eddy activity in the Nordic Sea (60-80N and 20W-20E), an area that has received limited attention and has proven to be difficult to analyze using other methods.

Feature Selection as State-Space Search: An Empirical Study in Clustering Problems

In this paper we treat the problem of feature selection in unsupervised learning as a state-space search problem. We introduce three different heuristic functions and perform extensive experiments on datasets with tens, hundreds, and thousands of features. Namely, we test different search algorithms using the heuristic functions we introduce. Our results show that the heuristic search approach for feature selection in unsupervised learning problems can be far superior than traditional baselines such as PCA and random projections.

Spatio-Temporal association rule based deep annotation-free clustering (STAR-DAC) for unsupervised person re-identification

Multi-camera video surveillance environment has a variety of emerging research problems among, which person re-identification is the premier one. Unsupervised person re-identification has been explored less in literature than the supervised approach. Images acquired from the video surveillance systems are unlabeled, which denotes that it is naturally an unsupervised learning problem. The state-of-the-art unsupervised methods seek external annotations support such as incorporating transfer learning techniques, partial labeling of train images, etc., which makes them not purely unsupervised and unsuitable for practical real-world surveillance settings. Identity mismatch happens due to the similar costumes and complex environmental factors. To resolve this issue, we introduce a new framework named Spatio-Temporal Association Rule based Deep Annotation-free Clustering (STAR-DAC) which incrementally clusters the unlabeled person re-identification images based on visual features and performs cluster fine-tuning through the mined spatio-temporal association rules. STAR formulations leveraged upto 75% of images for reliable sample selection through cluster fine-tuning. STAR based fine-tune algorithm aims to attain ground-truth labels of an unlabeled dataset and eliminate cluster outliers to stabilize the evaluation. Experiments are performed on image and video-based benchmark person re-identification datasets such as DukeMTMC re-ID, Market1501, MSMT17, CUHK03, GRID and Dukevideo re-ID, iLIDSVid, ViPer respectively. Experimental results clearly show that the proposed STAR-DAC framework outperforms the state-of-the-art methods in case of large scale datasets with multiple cameras.

Probabilistic Tractable Models in Mixed Discrete-Continuous Domains

Abstract We study the problem of the unsupervised learning of graphical models in mixed discrete-continuous domains. The problem of unsupervised learning of such models in discrete domains alone is notoriously challenging, compounded by the fact that inference is computationally demanding. The situation is generally believed to be significantly worse in discrete-continuous domains: estimating the unknown probability distribution of given samples is often limited in practice to a handful of parametric forms, and in addition to that, computing conditional queries need to carefully handle low-probability regions in safety-critical applications. In recent years, the regime of tractable learning has emerged, which attempts to learn a graphical model that permits efficient inference. Most of the results in this regime are based on arithmetic circuits, for which inference is linear in the size of the obtained circuit. In this work, we show how, with minimal modifications, such regimes can be generalized by leveraging efficient density estimation schemes based on piecewise polynomial approximations. Our framework is realized on a recent computational abstraction that permits efficient inference for a range of queries in the underlying language. Our empirical results show that our approach is effective, and allows a study of the trade-off between the granularity of the learned model and its predictive power.

GAN Ensemble for Anomaly Detection

When formulated as an unsupervised learning problem, anomaly detection often requires a model to learn the distribution of normal data. Previous works modify Generative Adversarial Networks (GANs) by using encoder-decoders as generators and apply them to anomaly detection tasks. Previous studies indicate that GAN ensembles are often more stable than single GANs in image generation tasks. In this work, we propose to construct GAN ensembles for anomaly detection. In the proposed method, a group of generators interact with a group of discriminators, so every generator gets feedback from every discriminator, and vice versa. Compared to a single GAN, an ensemble of GANs can better model the distribution of normal data and thus better detect anomalies. We also make a theoretical analysis of GANs and GAN ensembles in the context of anomaly detection. The empirical study constructs ensembles based on four different types of detecting models, and the results show that the ensemble outperforms the single model for all four model types.

Improving penalized regression-based clustering model in big data

Clustering is the main procedure for data mining with a wide application such as gene analysis. Clustering is a method of separates (grouping) previously unclassified data on the basis of its features, and it is an unsupervised learning problem that divides that data into groups in such a way that it makes those data in the same group more similar to each other compared to in other groups. Penalized regression-based clustering is an extension of the “Sum Of Norms” clustering model. In this paper, the nature-inspired algorithm is employed to improve the penalized regression-based clustering to better estimation. The real data application on gene expression data results suggests that our proposed improvement can bring significant improvement relative to others.

Circumventing spin-glass traps by microcanonical spontaneous symmetry breaking.

The planted p-spin interaction model is a paradigm of random-graph systems possessing both a ferromagnetic phase and a disordered phase with the latter splitting into many spin-glass states at low temperatures. Conventional simulated annealing dynamics is easily blocked by these low-energy spin-glass states. Here we demonstrate that actually this planted system is exponentially dominated by a microcanonical polarized phase at intermediate energy densities. There is a discontinuous microcanonical spontaneous symmetry breaking transition from the paramagnetic phase to the microcanonical polarized phase. This transition can serve as a mechanism to avoid all the spin-glass traps, and it is accelerated by the restart strategy of microcanonical random walk. We also propose an unsupervised learning problem on microcanonically sampled configurations for inferring the planted ground state.

Data Exploration by Representative Region Selection: Axioms and Convergence

We present a new type of unsupervised learning problem in which we find a small set of representative regions that approximates a larger data set. These regions may be presented to a practitioner along with additional information in order to help the practitioner explore the data set. An advantage of this approach is that it does not rely on cluster structure of the data. We formally define this problem, and we present axioms that should be satisfied by functions that measure the quality of representatives. We provide a quality function that satisfies all of these axioms. Using this quality function, we formulate two optimization problems for finding representatives. We provide convergence results for a general class of methods, and we show that these results apply to several specific methods, including methods derived from the solution of the optimization problems formulated in this paper. We provide an example of how representative regions may be used to explore a data set.