Learning Embedded Representation of the Stock Correlation Matrix Using Graph Machine Learning

Graph machine-learning (ML) methods have recently attracted great attention and have made significant progress in graph applications. To date, most graph ML approaches have been evaluated on social networks, but they have not been comprehensively reviewed in the health informatics domain. Herein, a review of graph ML methods and their applications in the disease prediction domain based on electronic health data is presented in this study from two levels: node classification and link prediction. Commonly used graph ML approaches for these two levels are shallow embedding and graph neural networks (GNN). This study performs comprehensive research to identify articles that applied or proposed graph ML models on disease prediction using electronic health data. We considered journals and conferences from four digital library databases (i.e., PubMed, Scopus, ACM digital library, and IEEEXplore). Based on the identified articles, we review the present status of and trends in graph ML approaches for disease prediction using electronic health data. Even though GNN-based models have achieved outstanding results compared with the traditional ML methods in a wide range of disease prediction tasks, they still confront interpretability and dynamic graph challenges. Though the disease prediction field using ML techniques is still emerging, GNN-based models have the potential to be an excellent approach for disease prediction, which can be used in medical diagnosis, treatment, and the prognosis of diseases.

Amidst the shifting tides of the educational landscape, this research article embarks on a transformative journey delving into the fusion of theoretical principles and pragmatic implementations within the realm of Graph Machine Learning (GML), particularly accentuated within the sphere of nature, environment, and pollution technology. GML emerges as a potent and indispensable tool, adeptly leveraging the intrinsic interconnectedness embedded within environmental datasets. Its application extends far beyond mere analysis towards the profound ability to forecast ecological patterns, prescribe sustainable interventions, and tailor pollution mitigation strategies with precision and efficacy. This article does not merely scratch the surface of GML’s applications but dives deep into its tangible implementations, unraveling its potential to revolutionize environmental science and pollution technology. It endeavors to bridge the gap between theory and practice, weaving together relevant ecological theories and empirical evidence that underpin the theoretical foundations supporting GML’s practical utility in environmental domains. By synthesizing theoretical insights with real-world applications, this research elucidates the profound transformative potential of GML, paving the way for proactive and data-driven approaches toward addressing pressing environmental challenges. In essence, this harmonization of theory and application catalyzes advancing the adoption of GML in environmental science and pollution technology. It not only illuminates the path towards sustainable practices but also lays the groundwork for fostering a holistic understanding of our ecosystem. Through this integration, GML emerges as a beacon guiding us toward a future where environmental stewardship is informed by data-driven insights, leading to more effective and sustainable solutions for the benefit of our planet and future generations.

Understanding customer behavior and predicting customer churn remain vital and central strategic objectives for modern digital organizations. Customer interactions are becoming multi-touch, they are cross-channel, and dynamically interconnected. The traditional linear modeling approach does not meet these demands. The traditional model does not capture the complexity that is vital in digital customer journeys. These challenges should be addressed, and graph machine learning (GraphML) and graph neural networks (GNNs) have become effective frameworks for addressing these challenges. The frameworks are capable of modeling the behavior of the customers as a network of entities, transitions, and interactions. The paper explores how graph machine learning can be applied to map the journeys of the customer and predict churn with higher accuracy, as well as deeper behavioral insight than conventional models. This paper will draw insights from peer-reviewed sources. It develops a conceptual, training graph model to detect churn signals, a multi-layered framework for constructing customer journey graphs, and operationalizing predictions into retention strategies that are actionable. The findings depict that graph-based modeling captures network dependencies. It captures cross-channel transitions and the behavior patterns that traditional machine learning does not incorporate. The findings indicate that GraphML offers a better method for describing customers as interconnected nodes within a relational structure. It enables more accurate churn prediction and a comprehensive mapping of the real-world customer journeys. The paper will conclude with a managerial implication, suggestions for future research, and limitations.

AI’s 10 to Watch, 2022

Exploring genetic influences on adverse outcome pathways using heuristic simulation and graph data science

Privacy and interpretability are two important ingredients for achieving trustworthy machine learning. We study the interplay of these two aspects in graph machine learning through graph reconstruction attacks. The goal of the adversary here is to reconstruct the graph structure of the training data given access to model explanations. Based on the different kinds of auxiliary information available to the adversary, we propose several graph reconstruction attacks. We show that additional knowledge of post-hoc feature explanations substantially increases the success rate of these attacks. Further, we investigate in detail the differences between attack performance with respect to three different classes of explanation methods for graph neural networks: gradient-based, perturbation-based, and surrogate model-based methods. While gradient-based explanations reveal the most in terms of the graph structure, we find that these explanations do not always score high in utility. For the other two classes of explanations, privacy leakage increases with an increase in explanation utility. Finally, we propose a defense based on a randomized response mechanism for releasing the explanations, which substantially reduces the attack success rate. Our code is available at https://github.com/iyempissy/graph-stealing-attacks-with-explanation.

Transparency and accountability have become major concerns for black-box machine learning (ML) models. Proper explanations for the model behavior increase model transparency and help researchers develop more accountable models. Graph neural networks (GNN) have recently shown superior performance in many graph ML problems than traditional methods, and explaining them has attracted increased interest. However, GNN explanation for link prediction (LP) is lacking in the literature. LP is an essential GNN task and corresponds to web applications like recommendation and sponsored search on web. Given existing GNN explanation methods only address node/graph-level tasks, we propose Path-based GNN Explanation for heterogeneous Link prediction (PaGE-Link) that generates explanations with connection interpretability, enjoys model scalability, and handles graph heterogeneity. Qualitatively, PaGE-Link can generate explanations as paths connecting a node pair, which naturally captures connections between the two nodes and easily transfer to human-interpretable explanations. Quantitatively, explanations generated by PaGE-Link improve AUC for recommendation on citation and user-item graphs by 9 - 35% and are chosen as better by 78.79% of responses in human evaluation.

Rapid prediction of environmental chemistry properties is critical for the green and sustainable development of the chemical industry and drug discovery.

This study investigates the potential of graph-based machine learning to enhance the detection and understanding of trading by insiders, encompassing both legal and potentially illicit activities. Given the complex and interconnected nature of financial markets, insider trading poses a significant challenge to market fairness and resource allocation efficiency. This research employs Graph Convolutional Networks to model insider trading within a network constructed from shared board memberships among S&P 500 companies. By integrating both historical stock prices and board interlocks, GCNs offer a comprehensive perspective to uncover subtle interactions leading to insider trading behaviours. The study aims to demonstrate the superiority of GCNs over traditional models on predicting the tendency of insider trading by conducting rigorous ablation studies and benchmark comparisons, which assess the impact of feature removal, edge deletion, and edge rewiring on model performance. These results demonstrate the potential benefits of graph-based methods for financial applications such as market regulation and monitoring.

Graph neural networks (GNNs) based on message passing have achieved remarkable performance in graph machine learning. By combining it with the power of pseudo labeling, one can further push forward the performance on the task of semi-supervised node classification. However, most existing works assume that the training node labels are purely noise-free, while this strong assumption usually does not hold in practice. GNNs will overfit the noisy training labels and the adverse effects of mislabeled nodes can be exaggerated by being propagated to the remaining nodes through the graph structure, exacerbating the model failure. Worse still, the noisy pseudo labels could also largely undermine the model's reliability without special treatment. In this paper, we revisit the role of (1) message passing and (2) pseudo labels in the studied problem and try to address two denoising subproblems from the model architecture and algorithm perspective, respectively. Specifically, we first develop a label-noise robust GNN that discards the coupled message-passing scheme. Despite its simple architecture, this learning backbone prevents overfitting to noisy labels and also inherently avoids the noise propagation issue. Moreover, we propose a novel reliable graph pseudo labeling algorithm that can effectively leverage the knowledge of unlabeled nodes while mitigating the adverse effects of noisy pseudo labels. Based on those novel designs, we can attain exceptional effectiveness and efficiency in solving the studied problem. We conduct extensive experiments on benchmark datasets for semi-supervised node classification with different levels of label noise and show new state-of-the-art performance. The code is available at https://github.com/DND-NET/DND-NET.

Recent years have witnessed the remarkable success of applying Graph Machine Learning (GML) to node/graph classification and link prediction. However, edge classification task that enjoys numerous real-world applications such as social network analysis and cybersecurity, has not seen significant advancement with the progress of GML. To address this gap, our study pioneers a comprehensive approach to edge classification. We identify a novel 'Topological Imbalance Issue,' which arises from the skewed distribution of edges across different classes, affecting the local subgraph of each edge and harming the performance of edge classifications. Inspired by recent node-level studies observing performance discrepancies with varying local structural patterns, we aim to investigate if the topological imbalanced edge classification tasks can also be mitigated by characterizing the local class distribution variance. Thus, we introduce Topological Entropy (TE), a novel topological-based metric that measures the topological imbalance for each edge. Our empirical studies confirm that TE effectively measures local class distribution variance, and indicate that prioritizing edges with high TE values can help address the issue of topological imbalance. Inspired by this observation, we develop two strategies - Topological Reweighting and TE Wedge-based Mixup - to adaptively focus training on (synthetic) edges based on their TEs. While topological reweighting directly manipulates training edge weights according to TE, our wedge-based mixup interpolates synthetic edges between high TE wedges. To further enhance performance, we integrate these strategies into a novel topological imbalance strategy for edge classification: TopoEdge. Extensive experiments on real-world datasets demonstrate the efficacy of our proposed strategies. Our code and data are available at https://github.com/XueqiC/TopoEdge. Additionally, our curated datasets and designed experimental settings establish a new benchmark for future edge classification research, particularly in addressing imbalance issues.

Matrix representations and interdependency on L-fuzzy covering-based approximation operators

Functional brain connectivity from fMRI studies has become an important tool in studying functional interactions in the human brain as a complex network. Most recently, research has started focusing on whole brain functional networks at the voxel-level, where fMRI time-signals at each voxel are correlated with every other voxel in the brain to determine their functional connectivity. For a typical 4mm isotropic voxel resolution, this results in connectivity networks with more than twenty thousand nodes and over 400 million links. These cannot be effectively visualized or interactively explored using node-link representations, and due to their size are challenging to show as correlation matrix bitmaps. In this paper, we present a number of methods for the visualization and interactive visual analysis of this new high resolution brain network data, both in its matrix representation as well as in its anatomical context. We have implemented these methods in a GPU raycasting framework that enables real-time interaction, such as network probing and volume deformation, as well as real-time filtering. The techniques are integrated in a visual analysis application in which the different views are coupled, supporting linked interaction. Furthermore, we allow visual comparison of different brain networks with side-by-side and difference visualization. We have evaluated our approach via case studies with domain scientists at two different university medical centers.

Sparse matrix vector multiplication (SpMV) is one of the most common operations in scientific and high-performance applications, and is often responsible for the application performance bottleneck. While the sparse matrix representation has a significant impact on the resulting application performance, choosing the right representation typically relies on expert knowledge and trial and error. This paper provides the first comprehensive study on the impact of sparse matrix representations on two emerging many-core architectures: the Intel's Knights Landing (KNL) XeonPhi and the ARM-based FT-2000Plus (FTP). Our large-scale experiments involved over 9,500 distinct profiling runs performed on 956 sparse datasets and five mainstream SpMV representations. We show that the best sparse matrix representation depends on the underlying architecture and the program input. To help developers to choose the optimal matrix representation, we employ machine learning to develop a predictive model. Our model is first trained offline using a set of training examples. The learned model can be used to predict the best matrix representation for any unseen input for a given architecture. We show that our model delivers on average 95% and 91% of the best available performance on KNL and FTP respectively, and it achieves this with no runtime profiling overhead.

We present a rigorous mathematical proof of concept demonstrating the theoretical feasibility of quantum linear regression through variational Hamiltonian formulation. Our framework systematically transforms classical linear regression into a quantum optimization problem by encoding regression weights as expectation values of Pauli-Z operators and reformulating the loss function as a quantum Hamiltonian. We provide explicit mathematical constructions, matrix representations, and convergence analysis to establish the theoretical foundation for quantum machine learning applications. Unlike previous quantum linear regression approaches that focus on exponential speedups through quantum linear systems solvers, our variational approach is designed for near-term quantum devices and naturally incorporates regularization through quantum parameter constraints. The mathematical framework demonstrates computational equivalence between classical and quantum approaches while revealing inherent advantages in constrained optimization scenarios. This work provides the theoretical groundwork for future empirical applications to financial time series prediction, including potential implementation on Thai SET Index forecasting using data from Yahoo Finance and quantum computing platforms. Our key contributions include: (1) complete mathematical mapping from classical to quantum regression formulation, (2) explicit Hamiltonian constructions with matrix representations, (3) theoretical convergence analysis for quantum gradient descent, and (4) identification of quantumspecific regularization properties that distinguish this approach from classical methods.