Graph neural networks (GNNs) find applications in various domains such as computational biology, natural language processing, and computer security. Owing to their popularity, there is an increasing need explain GNN predictions since GNNs are black-box machine learning models. One way address this counterfactual reasoning where the objective change prediction by minimal changes input graph. Existing methods for explanation of limited instance-specific local reasoning. This approach has two...

There has been an increased interest in discovering heuristics for combinatorial problems on graphs through machine learning. While existing techniques have primarily focused obtaining high-quality solutions, scalability to billion-sized not adequately addressed. In addition, the impact of budget-constraint, which is necessary many practical scenarios, remains be studied. this paper, we propose a framework called GCOMB bridge these gaps. trains Graph Convolutional Network (GCN) using novel...

Graph clustering is a fundamental and challenging task in the field of graph mining where objective to group nodes into clusters taking consideration topology graph. It has several applications diverse domains spanning social network analysis, recommender systems, computer vision, bioinformatics. In this work, we propose novel method, DGCluster, which primarily optimizes modularity using neural networks scales linearly with size. Our method does not require number be specified as part input...

Graph neural networks (GNNs) are powerful graph-based deep-learning models that have gained significant attention and demonstrated remarkable performance in various domains, including natural language processing, drug discovery, recommendation systems. However, combining feature information combinatorial graph structures has led to complex non-linear GNN models. Consequently, this increased the challenges of understanding workings GNNs underlying reasons behind their predictions. To address...

Graph Neural Networks (GNNs), a generalization of deep neural networks on graph data have been widely used in various domains, ranging from drug discovery to recommender systems. However, GNNs such applications are limited when there few available samples. Meta-learning has an important framework address the lack samples machine learning, and recent years, researchers started apply meta-learning GNNs. In this work, we provide comprehensive survey different metalearning approaches involving...

Graph Neural Networks (GNNs) have been extensively used in various real-world applications. However, the predictive uncertainty of GNNs stemming from diverse sources such as inherent randomness data and model training errors can lead to unstable erroneous predictions. Therefore, identifying, quantifying, utilizing are essential enhance performance for downstream tasks well reliability GNN This survey aims provide a comprehensive overview perspective with an emphasis on its integration graph...

In several domains, the flow of data is governed by an underlying network. Reduction delays in end-to-end important network optimization task. Reduced enable shorter travel times for vehicles road networks, faster information social and increased rate packets communication networks. While techniques delay minimization have been proposed, they fail to provide any noticeable reduction individual flows. Furthermore, treat all nodes as equally important, which often not case real-world this...

K-cores are maximal induced subgraphs where all vertices have degree at least k. These dense patterns applications in community detection, network visualization and protein function prediction. However, k-cores can be quite unstable to modifications, which motivates the question: How resilient is k-core structure of a network, such as Web or Facebook, edge deletions? We investigate this question from an algorithmic perspective. More specifically, we study problem computing small set edges...

Financial market analysis has focused primarily on extracting signals from accounting, stock price, and other numerical "hard" data reported in P&L statements or earnings per share reports. Yet, it is well-known that decision-makers routinely use "soft" text-based documents interpret the hard they narrate. Recent advances computational methods for analyzing unstructured soft at scale offer possibilities understanding financial behavior could improve investments equity. A critical ubiquitous...

Among various distance functions for graphs, graph and subgraph edit distances (GED SED respectively) are two of the most popular expressive measures. Unfortunately, exact computations both NP-hard. To overcome this computational bottleneck, neural approaches to learn predict in polynomial time have received much interest. While considerable progress has been made, there exist limitations that need be addressed. First, efficacy an approximate function lies not only its approximation...

Graph Neural Networks (GNNs) have been a powerful tool for node classification tasks in complex networks. However, their decision-making processes remain black-box to users, making it challenging understand the reasoning behind predictions. Counterfactual explanations (CFE) shown promise enhancing interpretability of machine learning models. Prior approaches compute CFE GNNS often are learning-based that require training additional graphs. In this paper, we propose semivalue-based,...

Combinatorial Optimization (CO) problems over graphs appear routinely in many applications such as optimizing traffic, viral marketing social networks, and matching for job allocation. Due to their combinatorial nature, these are often NP-hard. Existing approximation algorithms heuristics rely on the search space find solutions become time-consuming when this is large. In paper, we design a neural method called COMBHelper reduce thus improve efficiency of traditional CO based node selection....

Recent advancements in Artificial Intelligence (AI) and machine learning have demonstrated transformative capabilities across diverse domains. This progress extends to the field of patent analysis innovation, where AI-based tools present opportunities streamline enhance important tasks cycle such as classification, retrieval, valuation prediction. not only accelerates efficiency researchers applicants but also opens new avenues for technological innovation discovery. Our survey provides a...

Graph neural networks (GNNs) find applications in various domains such as computational biology, natural language processing, and computer security. Owing to their popularity, there is an increasing need explain GNN predictions since GNNs are black-box machine learning models. One way address this issue involves using counterfactual reasoning where the objective alter prediction by minimal changes input graph. Existing methods for explanation of limited instance-specific local reasoning....

Online social networks have become major battlegrounds for political campaigns, viral marketing, and the dissemination of news. As a consequence, "bad actors" are increasingly exploiting these platforms, which is key challenge their administrators, businesses society in general. The spread fake news classical example abuse by bad actors. While some advocated stricter policies to control misinformation networks, this often happens detriment democratic organic structure. In paper, we aim limit...

Reduction of end-to-end network delay is an optimization task with applications in multiple domains. Low delays enable improved information flow social networks, quick spread ideas collaboration low travel times for vehicles on road and increased rate packets the case communication networks. Delay reduction can be achieved by both improving propagation capabilities individual nodes adding additional edges network. One main challenges such design problems that effects local changes are not...

Adversarial attacks on Graph Neural Networks (GNNs) reveal their security vulnerabilities, limiting adoption in safety-critical applications. However, existing attack strategies rely the knowledge of either GNN model being used or predictive task attacked. Is this necessary? For example, a graph may be for multiple downstream tasks unknown to practical attacker. It is thus important test vulnerability GNNs adversarial perturbations and task-agnostic setting. In work, we study problem show...