Graph-based Structure Research Articles

Dynamic Routing and Knowledge Re-Learning for Data-Free Black-Box Attack.

Deep learning models have emerged as strong and efficient tools that can be applied to a broad spectrum of complex learning problems and many real-world applications. However, more and more works show that deep models are vulnerable to adversarial examples. Compared to vanilla attack settings, this paper advocates a more practical setting of data-free black-box attack, for which the attackers can completely not access the structures and parameters of the target model, as well as the intermediate features and any training data associated with the model. To tackle this task, previous methods generate transferable adversarial examples from a transparent substitute model to the target model. However, we found that these works have the limitations of taking static substitute model structure for different targets, only using hard synthesized examples once, and still relying on data statistics of the target model. This may potentially harm the performance of attacking the target model. To this end, we propose a novel Dynamic Routing and Knowledge Re-Learning framework (DraKe) to effectively learn a dynamic substitute model from the target model. Specifically, given synthesized training samples, a dynamic substitute structure learning strategy is proposed to adaptively generate optimal substitute model structure via a policy network according to different target models and tasks. To facilitate the substitute training, we present a graph-based structure information learning to capture the structural knowledge learned from the target model. For the inherent limitation that online data generation can only be learned once, a dynamic knowledge re-learning strategy is proposed to adjust the weights of optimization objectives and re-learn hard samples. Extensive experiments on four public image classification datasets and one face recognition benchmark are conducted to evaluate the efficacy of our Drake. We can obtain significant improvement compared with state-of-the-art competitors. More importantly, our DraKe consistently achieves attack superiority for different target models (e.g., residual networks, and vision transformers), showing great potential for complex real-world applications.

The spatio-temporal evolution of spatial structure and supply-demand relationships of the ecological network in the Yellow River Delta region of China

Rapid urbanization has caused unregulated transformation of land use/land cover, further exacerbating the imbalance in ecological supply-demand relationships. Constructing the ecological supply-demand network by integrating natural ecological processes and human ecological demands into a unified spatio-temporal framework and considering their internal dynamic flow characteristics is an effective approach to address this issue. However, there is currently a lack of evaluation methods for the coordination of ecological network supply-demand relationships, which could provide essential guidance and reference. This article takes Yellow River Delta region (YRDR) as an example and utilizes the integrated traditional ecological network paradigm and ecosystem services theory to simulate the ecological supply-demand network for the years 2000, 2010, and 2020 in the study area. Then, a novel evaluation index system for the ecological network supply-demand relationships was proposed from the perspectives of intensity, capacity, and equilibrium. Based on this, combined with a graph-based network structure evaluation system, it reveals the spatio-temporal evolution process of the spatial structure and supply-demand relationship of the ecological network in the study area. The study results showed that: (1) The quantity of elements in the ecological supply network and the demand network show opposite trends, leading to an exacerbation of the spatial imbalance in the arrangement of ecological supply-demand network elements. (2) Although the structural changes in the ecological supply network were significantly greater than those in the ecological demand network, the overall structure of the ecological supply-demand network gradually improved. (3) However, under the coupled influence of imbalanced change in the spatial distribution and connectivity of ecological elements, the overall coordination of the ecological supply-demand relationships within the YRDR's ecological network is gradually deteriorating. (4) The proposed optimization strategy, "Two zones, Three belts, Four corridors", for the ecological security pattern in YRDR can accommodate the ecological process cycle and human ecological demand. This study can provide scientific references for "harmonious coexistence between humans and nature" ecological protection and sustainable development planning under the background of rapid urbanization.

Segmentation of 3D Point Clouds of Heritage Buildings Using Edge Detection and Supervoxel-Based Topology.

This paper presents a novel segmentation algorithm specially developed for applications in 3D point clouds with high variability and noise, particularly suitable for heritage building 3D data. The method can be categorized within the segmentation procedures based on edge detection. In addition, it uses a graph-based topological structure generated from the supervoxelization of the 3D point clouds, which is used to make the closure of the edge points and to define the different segments. The algorithm provides a valuable tool for generating results that can be used in subsequent classification tasks and broader computer applications dealing with 3D point clouds. One of the characteristics of this segmentation method is that it is unsupervised, which makes it particularly advantageous for heritage applications where labelled data is scarce. It is also easily adaptable to different edge point detection and supervoxelization algorithms. Finally, the results show that the 3D data can be segmented into different architectural elements, which is important for further classification or recognition. Extensive testing on real data from historic buildings demonstrated the effectiveness of the method. The results show superior performance compared to three other segmentation methods, both globally and in the segmentation of planar and curved zones of historic buildings.

Efficient Static Vulnerability Analysis for JavaScript with Multiversion Dependency Graphs

While static analysis tools that rely on Code Property Graphs (CPGs) to detect security vulnerabilities have proven effective, deciding how much information to include in the graphs remains a challenge. Including less information can lead to a more scalable analysis but at the cost of reduced effectiveness in identifying vulnerability patterns, potentially resulting in classification errors. Conversely, more information in the graph allows for a more effective analysis but may affect scalability. For example, scalability issues have been recently highlighted in ODGen, the state-of-the-art CPG-based tool for detecting Node.js vulnerabilities. This paper examines a new point in the design space of CPGs for JavaScript vulnerability detection. We introduce the Multiversion Dependency Graph (MDG), a novel graph-based data structure that captures the state evolution of objects and their properties during program execution. Compared to the graphs used by ODGen, MDGs are significantly simpler without losing key information needed for vulnerability detection. We implemented Graph.js, a new MDG-based static vulnerability scanner specialized in analyzing npm packages and detecting taint-style and prototype pollution vulnerabilities. Our evaluation shows that Graph.js outperforms ODGen by significantly reducing both the false negatives and the analysis time. Additionally, we have identified 49 previously undiscovered vulnerabilities in npm packages.

Efficient suppression algorithms for preserving trajectory privacy

Concerns about the security of trajectory data have surfaced amid rising public awareness of privacy protection. In this study, we focus on how to protect personal information effectively when adversaries have access to a portion of users’ trajectory data. To resolve this issue, we design a graph-based index structure to store users’ trajectory data and develop a novel subtrajectory upper bound estimation method and corresponding pruning strategies. Using a graph-based index structure and pruning strategies, we propose a global suppression algorithm and a local suppression algorithm to prevent personal information from being extracted from the original trajectory data. Experimental results show that the maintenance cost of the graph-based index structure is low when performing global and local suppression, and that the pruning strategies effectively eliminate unnecessary computation of non-upper-bound subtrajectories. Therefore, the execution times of the proposed algorithms are far shorter than those of existing algorithms.

OFIDA: Object-focused image data augmentation with attention-driven graph convolutional networks.

Image data augmentation plays a crucial role in data augmentation (DA) by increasing the quantity and diversity of labeled training data. However, existing methods have limitations. Notably, techniques like image manipulation, erasing, and mixing can distort images, compromising data quality. Accurate representation of objects without confusion is a challenge in methods like auto augment and feature augmentation. Preserving fine details and spatial relationships also proves difficult in certain techniques, as seen in deep generative models. To address these limitations, we propose OFIDA, an object-focused image data augmentation algorithm. OFIDA implements one-to-many enhancements that not only preserve essential target regions but also elevate the authenticity of simulating real-world settings and data distributions. Specifically, OFIDA utilizes a graph-based structure and object detection to streamline augmentation. Specifically, by leveraging graph properties like connectivity and hierarchy, it captures object essence and context for improved comprehension in real-world scenarios. Then, we introduce DynamicFocusNet, a novel object detection algorithm built on the graph framework. DynamicFocusNet merges dynamic graph convolutions and attention mechanisms to flexibly adjust receptive fields. Finally, the detected target images are extracted to facilitate one-to-many data augmentation. Experimental results validate the superiority of our OFIDA method over state-of-the-art methods across six benchmark datasets.

Hybrid programming-model strategies for GPU offloading of electronic structure calculation kernels.

To address the challenge of performance portability and facilitate the implementation of electronic structure solvers, we developed the basic matrix library (BML) and Parallel, Rapid O(N), and Graph-based Recursive Electronic Structure Solver (PROGRESS) library. The BML implements linear algebra operations necessary for electronic structure kernels using a unified user interface for various matrix formats (dense and sparse) and architectures (CPUs and GPUs). Focusing on density functional theory and tight-binding models, PROGRESS implements several solvers for computing the single-particle density matrix and relies on BML. In this paper, we describe the general strategies used for these implementations on various computer architectures, using OpenMP target functionalities on GPUs, in conjunction with third-party libraries to handle performance critical numerical kernels. We demonstrate the portability of this approach and its performance in benchmark problems.

Graph-based geometric structure line parsing

Line drawings by artists capture the organization, relationships, and semantics of observable objects. To endow machines with similar capacities and improve the storage and processing of such drawings, we investigate uniform representation and learning of hetero-dimensional line drawings. For this, nonlinear curves are first approximated as paths, resulting in node detection and link prediction (LP). An end-to-end trainable neural network is then developed for parsing structural-semantic sketches from images based on such representations. In particular, node regression and feature extraction are performed using convolutional neural networks(CNNs), and the prediction of connections is achieved using a graph-based encoder–decoder architecture. We provide a topology-guided graph initialization strategy, which increases the accuracy of LP but also greatly minimizes the self-overlapping among predicted paths. We also develop a semi-automatic toolkit for annotation generation, as well as a dataset for understanding structural sketches—the first large parametric dataset consistent with human visual perception. Through comparative experiments and ablation studies, we demonstrate the advantages and effectiveness of our methods.

Modeling a domain wall network in BiFeO3 with stochastic geometry and entropy-based similarity measure

A compact and tractable two-dimensional model to generate the topological network structure of domain walls in BiFeO3 thin films is presented in this study. Our method combines the stochastic geometry parametric model of the centroidal Voronoi tessellation optimized using the von Neumann entropy, a novel information-theoretic tool for networks. The former permits the generation of image-based stochastic artificial samples of domain wall networks, from which the network structure is subsequently extracted and converted to the graph-based representation. The von Neumann entropy, which reflects information diffusion across multiple spatiotemporal scales in heterogeneous networks, plays a central role in defining a fitness function. It allows the use of the network as a whole rather than using a subset of network descriptors to search for optimal model parameters. The optimization of the parameters is carried out by a genetic algorithm through the maximization of the fitness function and results in the desired graph-based network connectivity structure. Ground truth empirical networks are defined, and a dataset of network connectivity structures of domain walls in BiFeO3 thin films is undertaken through manual annotation. Both a versatile tool for manual network annotation of noisy images and a new automatic network extraction method for high-quality images are developed.

Knowledge mapping analysis of Chinese traditional culture under the strategy of common wealth

Abstract With the continuous improvement of people’s living standards, their attention to the spiritual world and culture, especially traditional culture, is gradually increasing. With the help of the graph-based data structure of a knowledge graph, the connections between entities in conventional culture can be more intuitively displayed in the form of nodes and edges. Traditional knowledge is represented by the Trans series model in this paper, and the attention mechanism is integrated into the Graph Attention Network Model (GAT). The descriptive information of entities and relationships was introduced for pre-training initialization, and a Graph Attention Network Connection Prediction Model (BFGAT) was proposed to integrate the descriptive information and structural features and the BFGAT model was optimized by using information filtering operations. After the experimental study of the performance of the modified model, it is found that the score of the BFGAT model is 0.594 under the Hits@10 evaluation index in the dataset WN188RR, which is 0.01 and 0.018 higher than the two variant models, respectively, and the performance is better than that of other models. In the analysis of the ability of the knowledge graph to disseminate Chinese traditional culture, the number of searches for “Confucius” in Chinese traditional culture in the Americas is higher. The search frequency of Confucius has increased significantly from 2013 to 2014, from 830,000 times in 2013 to 2.95 million times in 2014, and the search frequency of Confucius in the Americas has reached 7.59 million times in 2023. It can be seen that the knowledge graph technology based on the knowledge graph attention network link prediction model has higher accuracy for keyword recognition and is also conducive to the wide dissemination of traditional Chinese culture.

APGVAE: Adaptive disentangled representation learning with the graph-based structure information

Neural networks are used to learn task-oriented high-level representations in an end-to-end manner by building a multi-layer neural network. Generation models have developed rapidly with the emergence of deep neural networks. But it still has problems with the insufficient authenticity of generated images, the deficiency of diversity, consistency, and unexplainability in the generation process. Disentangled representation is an effective method to learn a high-level feature representation and realize the interpretability of deep neural networks. We propose a general disentangled representation learning network with variational autoencoder network as the basic framework for the image generation process. The graph-based structure of the priors is embedded in the last module of the deep encoder network to build the feature spaces by the class, task-oriented, and task-unrelated information respectively. Meanwhile the priors should be adaptively modified with the task relevance of a generated image. And the semi-supervised learning is further involved in the disentangled representation network framework to reduce the requirements of label and extend the majority of feature space under the task-unrelated feature assumption. Experimental results show that the proposed method is efficient for various types of images and has a good potential for further research and development.

Enforcing legal information extraction through context-aware techniques: The ASKE approach

To cope with the growing volume, complexity, and articulation of legal documents as well as to foster digital justice and digital law, increasing effort is being devoted to legal knowledge extraction and digital transformation processes. In this paper, we present the ASKE (Automated System for Knowledge Extraction) approach to legal knowledge extraction, based on a combination of context-aware embedding models and zero-shot learning techniques into a three-phase extraction cycle, which is executed a number of times (called generations) to progressively extract concepts representative of the different meanings of terminology used in legal documents chunks. A graph-based data structure called ASKE Conceptual Graph is initially populated through a data preparation step, and it is continuously enriched at each ASKE generation with results of document chunk classification, new extracted terminology, and newly derived concepts. A quantitative evaluation of ASKE knowledge extraction and document classification is provided by considering the EurLex dataset. Furthermore, we present the results of applying ASKE to a real case-study of Italian case law decisions with qualitative feedback from legal experts in the framework of an ongoing national research project.

Generalization of Randic ́ Index of the Non-commuting Graph for Some Finite Groups

Randić index is one of the classical graph-based molecular structure descriptors in the field of mathematical chemistry. The Randić index of a graph is calculated by summing the reciprocals of the square root of the product of the degrees of two adjacent vertices in the graph. Meanwhile, the non-commuting graph is the graph of vertex set whose vertices are non-central elements and two distinct vertices are joined by an edge if and only if they do not commute. In this paper, the general formula of the Randić index of the non-commuting graph associated to three types of finite groups are presented. The groups involved are the dihedral groups, the generalized quaternion groups, and the quasi-dihedral groups. Some examples of the Randić index of the non-commuting graph related to a certain order of these groups are also given based on the main results.

CAM-NAS: An Efficient and Interpretable Neural Architecture Search Model Based on Class Activation Mapping

Artificial intelligence (AI) has made rapid progress in recent years, but as the complexity of AI models and the need to deploy them on multiple platforms gradually increases, the design of network model structures for specific platforms becomes more difficult. A neural network architecture search (NAS) serves as a solution to help experts discover new network structures that are suitable for different tasks and platforms. However, traditional NAS algorithms often consume time and many computational resources, especially when dealing with complex tasks and large-scale models, and the search process can become exceptionally time-consuming and difficult to interpret. In this paper, we propose a class activation graph-based neural structure search method (CAM-NAS) to address these problems. Compared with traditional NAS algorithms, CAM-NAS does not require full training of submodels, which greatly improves the search efficiency. Meanwhile, CAM-NAS uses the class activation graph technique, which makes the searched models have better interpretability. In our experiments, we tested CAM-NAS on an NVIDIA RTX 3090 graphics card and showed that it can evaluate a submodel in only 0.08 seconds, which is much faster than traditional NAS methods. In this study, we experimentally evaluated CAM-NAS using the CIFAR-10 and CIFAR-100 datasets as benchmarks. The experimental results show that CAM-NAS achieves very good results. This not only proves the efficiency of CAM-NAS, but also demonstrates its powerful performance in image classification tasks.

GRIP: Graph Representation of Immune Repertoire Using Graph Neural Network and Transformer

The immune repertoire is a collection of immune recep-tors that has emerged as an important biomarker for both diagnostic and therapeutic of cancer patients. In terms of deep learning, analyzing immune repertoire is a challeng-ing multiple-instance learning problem in which the im-mune repertoire of an individual is a bag, and the immune receptor is an instance. Although several deep learning methods for immune repertoire analysis are introduced, they consider the immune repertoire as a set-like struc-ture that doesn’t take account of the nature of the im-mune response. When the immune response occurs, mu-tations are introduced to the immune receptor sequence sequentially to optimize the immune response against the pathogens that enter our body. As a result, immune receptors for the specific pathogen have the lineage of evolution; thus, immune repertoire is better represented as a graph-like structure. In this work, we present our novel method graph representation of immune repertoire (GRIP), which analyzes the immune repertoire as a hier-archical graph structure and utilize the collection of graph neural network followed by graph pooling and transformer to efficiently represents the immune reper-toire as an embedding vector. We show that GRIP predict the survival probability of cancer patients better than the set-based methods and graph-based structure is critical for performance. Also, GRIP provides interpretable re-sults, which prove that GRIP adequately use the progno-sis-related immune receptor and give further possibility to use the GRIP as the novel biomarker searching tool

Semantic Clone Detection Based on Code Feature Fusion Learning

Code clones are duplicated code snippets that significantly threaten software maintenance and the public corpora of code representation learning. Traditionally, code context and its structure information abstract syntax tree (AST), control flow graph (CFG) are typical representations of source code, and context-based models and structure-based models contributed significantly to the development of code clone detection. In this paper, we present a hybrid embedding model for code clone detection (HEM-CCD), a fusion method of token sequential information and graph-based structure information. We insert tokens’ global context information encoded by a bi-directional recurrent neural network into the AST-based graph for comprehensive code semantic representation. Then, feeding the graph into a gated graph neural network we generate code semantic vectors for similarity evaluation. We have implemented our model on two public clone datasets (BigCloneBench and GoogleCodeJam), and the results indicate that HEM-CCD outperforms several state-of-the-art approaches.

Synchronizing Many Filesystems in Near Linear Time

Finding a provably correct subquadratic synchronization algorithm for many filesystem replicas is one of the main theoretical problems in operational transformation (OT) and conflict-free replicated data types (CRDT) frameworks. Based on the algebraic theory of filesystems, which incorporates non-commutative filesystem commands natively, we developed and built a proof-of-concept implementation of an algorithm suite which synchronizes an arbitrary number of replicas. The result is provably correct, and the synchronized system is created in linear space and time after an initial sorting phase. It works by identifying conflicting command pairs and requesting one of the commands to be removed. The method can be guided to reach any of the theoretically possible synchronized states. The algorithm also allows asynchronous usage. After the client sends a synchronization request, the local replica remains available for further modifications. When the synchronization instructions arrive, they can be merged with the changes made since the synchronization request. The suite also works on filesystems with a directed acyclic graph-based path structure in place of the traditional tree-like arrangement. Consequently, our algorithms apply to filesystems with hard or soft links as long as the links create no loops.

An optimized graph-based structure for single-cell RNA-seq cell-type classification based on non-linear dimension reduction

BackgroundIt is now possible to analyze cellular heterogeneity at the single-cell level thanks to the rapid developments in single-cell sequencing technologies. The clustering of cells is a fundamental and common step in heterogeneity analysis. Even so, accurate cell clustering remains a challenge due to the high levels of noise, the high dimensions, and the high sparsity of data.ResultsHere, we present SCEA, a clustering approach for scRNA-seq data. Using two consecutive units, an encoder based on MLP and a graph attention auto-encoder, to obtain cell embedding and gene embedding, SCEA can simultaneously achieve cell low-dimensional representation and clustering performing various examinations to obtain the optimal value for each parameter, the presented result is in its most optimal form. To evaluate the performance of SCEA, we performed it on several real scRNA-seq datasets for clustering and visualization analysis.ConclusionsThe experimental results show that SCEA generally outperforms several popular single-cell analysis methods. As a result of using all available datasets, SCEA, in average, improves clustering accuracy by 4.4% in ARI Parameters over the well-known method scGAC. Also, the accuracy improvement of 11.65% is achieved by SCEA, compared to the Seurat model.

CCPNet136: automated detection of schizophrenia using carbon chain pattern and iterative TQWT technique with EEG signals

Objective. Schizophrenia (SZ) is a severe, chronic psychiatric-cognitive disorder. The primary objective of this work is to present a handcrafted model using state-of-the-art technique to detect SZ accurately with EEG signals. Approach. In our proposed work, the features are generated using a histogram-based generator and an iterative decomposition model. The graph-based molecular structure of the carbon chain is employed to generate low-level features. Hence, the developed feature generation model is called the carbon chain pattern (CCP). An iterative tunable q-factor wavelet transform (ITQWT) technique is implemented in the feature extraction phase to generate various sub-bands of the EEG signal. The CCP was applied to the generated sub-bands to obtain several feature vectors. The clinically significant features were selected using iterative neighborhood component analysis (INCA). The selected features were then classified using the k nearest neighbor (kNN) with a 10-fold cross-validation strategy. Finally, the iterative weighted majority method was used to obtain the results in multiple channels. Main results. The presented CCP-ITQWT and INCA-based automated model achieved an accuracy of 95.84% and 99.20% using a single channel and majority voting method, respectively with kNN classifier. Significance. Our results highlight the success of the proposed CCP-ITQWT and INCA-based model in the automated detection of SZ using EEG signals.

MSEDDI: Multi-Scale Embedding for Predicting Drug-Drug Interaction Events.

A norm in modern medicine is to prescribe polypharmacy to treat disease. The core concern with the co-administration of drugs is that it may produce adverse drug-drug interaction (DDI), which can cause unexpected bodily injury. Therefore, it is essential to identify potential DDI. Most existing methods in silico only judge whether two drugs interact, ignoring the importance of interaction events to study the mechanism implied in combination drugs. In this work, we propose a deep learning framework named MSEDDI that comprehensively considers multi-scale embedding representations of the drug for predicting drug-drug interaction events. In MSEDDI, we design three-channel networks to process biomedical network-based knowledge graph embedding, SMILES sequence-based notation embedding, and molecular graph-based chemical structure embedding, respectively. Finally, we fuse three heterogeneous features from channel outputs through a self-attention mechanism and feed them to the linear layer predictor. In the experimental section, we evaluate the performance of all methods on two different prediction tasks on two datasets. The results show that MSEDDI outperforms other state-of-the-art baselines. Moreover, we also reveal the stable performance of our model in a broader sample set via case studies.