Implicit Relationships Research Articles

SAX-STGCN: Dynamic Spatio-Temporal Graph Convolutional Networks for Traffic Flow Prediction

Accurate, timely, and reliable traffic flow prediction is essential for an intelligent transportation system due to the complex spatio-temporal correlation of traffic flow. The prediction model based on graph convolution neural network (GCN) has become mainstream in recent years. However, most of the prediction models based on GCN only use an adjacency matrix to characterize the spatial correlation of traffic flow, ignoring the complex and dynamic relationship between road network adjacent nodes and missing the hidden connection between the global nodes of the road network. This paper proposes a SAX-STGCN network for traffic flow prediction to solve the above problems. The SAX-STGCN model uses symbolic approximation (sax) to obtain the similarity of the historical data of the predicted node in the previous period, including adjacent nodes and non-adjacent nodes, forming a similarity matrix to replace the original adjacency matrix composed of 0 and 1, which is defined as a global sax-correlation matrix to characterize the correlation between nodes in the road network and capture the implicit spatial relationships in the road network. Then, based on the dynamic global sax-correlation matrix, GCN is used to capture the spatial correlation of traffic flow, and GRU is used to capture the temporal correlation of traffic flow. The prediction accuracy is better than the baseline and has long-term prediction ability.

Multilevel Deformable Attention-Aggregated Networks for Change Detection in Bitemporal Remote Sensing Imagery

Deep learning (DL) approaches based on convolutional encoder–decoder networks have shown promising results in bitemporal change detection. However, their performance is limited by insufficient contextual information aggregation because they cannot fully capture the implicit contextual dependency relationships among feature maps at different levels. Moreover, harvesting long-range contextual information typically incurs high computational complexity. To circumvent these challenges, we propose multilevel deformable attention-aggregated networks (MLDANets) to effectively learn long-range dependencies across multiple levels of bitemporal convolutional features for multiscale context aggregation. Specifically, a multilevel change-aware deformable attention (MCDA) module consisting of linear projections with learnable parameters is built based on multihead self-attention (SA) with a deformable sampling strategy. It is applied in the skip connections of an encoder–decoder network taking a bitemporal deep feature hypersequence (BDFH) as input. MCDA can progressively address a set of informative sampling locations in multilevel feature maps for each query element in the BDFH. Simultaneously, MCDA learns to characterize beneficial information from different spatial and feature subspaces of BDFH using multiple attention heads for change perception. As a result, contextual dependencies across multiple levels of bitemporal feature maps can be adaptively aggregated via attention weights to generate multilevel discriminative change-aware representations. Experiments on very-high-resolution (VHR) datasets verify that MLDANets outperform state-of-the-art change detection approaches with dramatically faster training convergence and high computational efficiency.

Skeleton-Based ST-GCN for Human Action Recognition With Extended Skeleton Graph and Partitioning Strategy

Skeleton-based Graph Convolutional Networks (GCN) for human action and interaction recognition have received considerable attention of researchers due to its compact and view-invariant nature of skeleton data. However, the static skeleton graph topology in conventional GCNs does not reflect the implicit relationships of non-adjacent joints, which contain vital latent information for a skeleton pose in an action sequence. Moreover, traditional tri-categorical node partitioning strategy discards much of the motion dependencies along temporal dimension for non-physically connected edges. We propose an extended skeleton graph topology along with extended partitioning strategy to extract much of the non-adjacent joint relational information in the model for robust discriminative features. Extended skeleton graph represents joints as vertices and weighted edges represent intrinsic and extrinsic relationships between physically connected and non-physically connected joints respectively. Furthermore, extended partitioning strategy divides the input graph for GCN as five-categorical fixed-length tensor to encompass maximal motion dependencies. Finally, the extended skeleton graph and partitioning strategy are realized by adopting Spatio-Temporal Graph Convolutional Network (ST-GCN). The experiments carried out over three large scale datasets NTU-RGB+D, NTU-RGB+D 120 and Kinetics-Skeleton show improved performance over conventional state-of-the-art ST-GCNs.

Flow past a porous plate of non-Newtonian fluids with implicit shear stress shear rate relationships

One of the simplest problems in the classical Navier–Stokes theory that illustrates the development of boundary layers is the asymptotic suction problem (Schlichting (1960)). This problem has been studied within the context of many different non-Newtonian fluid models and in this note we consider it within the context of a new class of constitutive relations that has been put into place. Depending on the values of the parameters, this class of constitutive relations includes models for both shear thinning and shear thickening fluids, and truly implicit shear stress shear rate relationships. For any set of the material parameters we are able to find exact solutions to the governing equations, study their stability, show that suction has a stabilizing effect, and delineate the development of boundary layers.

Exploring implicit relationships between pavement surface friction and vehicle crash severity using interpretable extreme gradient boosting method

Pavement friction has been identified as crucial in traffic safety. Since the Highway Safety Manual prediction algorithm is often based on crash frequency, the crash severity distribution might be assumed unchanged before and after the countermeasure. However, pavement surface treatments can improve the friction to different levels, by which crash severity outcomes may vary greatly. To explore the implicit effects of pavement friction on vehicle crash severity, this paper first validates the extreme gradient boosting model performance and then the Shapley additive explanations interaction values are employed to interpret individual features and the nonlinear interactions among predictors. Under various scenarios, the extreme gradient boosting (XGBoost) output probability is utilized to convert into dynamic crash severity distributions. Results also indicate that friction becomes more significant when the friction number is less than 38, and immediate corrective actions are needed when the friction number is below 20.

Using Graph Representation in Host-Based Intrusion Detection

Cybersecurity has become an important part of our daily lives. As an important part, there are many researches on intrusion detection based on host system call in recent years. Compared to sentences, a sequence of system calls has unique characteristics. It contains implicit pattern relationships that are less sensitive to the order of occurrence and that have less impact on the classification results when the frequency of system calls varies slightly. There are also various properties such as resource consumption, execution time, predefined rules, and empirical weights of system calls. Commonly used word embedding methods, such as Bow, TI-IDF, N-Gram, and Word2Vec, do not fully exploit such relationships in sequences as well as conveniently support attribute expansion. To solve these problems, we introduce Graph Representation based Intrusion Detection (GRID), an intrusion detection framework based on graph representation learning. It captures the potential relationships between system calls to learn better features, and it is applicable to a wide range of back-end classifiers. GRID utilizes a new sequence embedding method Graph Random State Embedding (GRSE) that uses graph structures to model a finite number of sequence items and represent the structural association relationships between them. A more efficient representation of sequence embeddings is generated by random walks, word embeddings, and graph pooling. Moreover, it can be easily extended to sequences with attributes. Our experimental results on the AFDA-LD dataset show that GRID has an average improvement of 2% using the GRSE embedding method comparing to others.

Global world (dis-)order? Analyzing the dynamic evolution of the micro-structure of multipolarism by means of an unsupervised neural network approach

We make use of an advanced artificial neural network (Auto-CM) to model the structure of the current world order as a data-driven reconstruction of the implicit relationships between countries and of their time evolution, as derived from a database of publicly observable socioeconomic and political variables. Building on previous research, we analyze 93 variables derived from dozens of key indicators for 128 countries and trace their evolution along a period of eight years, 2007–2014. We find evidence of an increasing structural instability that seems to signal a transition toward a new, as yet undetermined, multipolar world order.

Analysis of Relationship amongst Money Supply, Prices and Output in India

The paper studies the relationship between Money Supply, Prices and Output by creating a robust VAR Model incorporating the changing trend of the macro-economic variables. The observations were that rate of change of Prices and Money Supply have a direct relationship (in short and long run), rate of change of Price and Private Consumption have an inverse relationship (more significant in long run), and rate of change of Investment and Money Supply have an inverse relationship (significant in long run). The study provides a foundation for understanding of stability and shocks amongst the macro-economic variables which have further scope of exploration in terms of magnitude of absolute change and implicit relationships which exist amongst the variables in the Quantity Theory of Money.

Recursive Gauss-Helmert model with equality constraints applied to the efficient system calibration of a 3D laser scanner

Abstract Sensors for environmental perception are nowadays applied in numerous vehicles and are expected to be used in even higher quantities for future autonomous driving. This leads to an increasing amount of observation data that must be processed reliably and accurately very quickly. For this purpose, recursive approaches are particularly suitable in terms of their efficiency when powerful CPUs and GPUs are uneconomical, too large, or too heavy for certain applications. If explicit functional relationships between the available observations and the requested parameters are used to process and adjust the observation data, complementary approaches exist. The situation is different for implicit relationships, which could not be considered recursively for a long time but only in the context of batch adjustments. In this contribution, a recursive Gauss-Helmert model is presented that can handle explicit and implicit equations and thus allows high flexibility. This recursive estimator is based on a Kalman filter for implicit measurement equations, which has already been used for georeferencing kinematic multi-sensor systems (MSS) in urban environments. Furthermore, different methods for introducing additional information using constraints and the resulting added value are shown. Practical application of the methodology is given by an example for the calibration of a laser scanner for a MSS.

Eulogising the Dead: A Systemic Functional Exploration of Tributes Delivered by Ghanaian Presidents

Presidents are considered as caring and sensitive fathers of their nations whose words of condolences are usually needed in times of grief. One medium through which they express their condolences is through the delivery of tributes to eulogize the dead and sympathize with bereaved families. There is therefore the need to examine how presidents employ the faculties of language to eulogize the dead in their tributes. Premised on the transitivity framework postulated by Halliday and Matthiessen, the present study investigates the use of transitivity patterns in Ghanaian presidential tributes to unearth various experiences and also unveil the implicit relationships that exist between politicians and traditional rulers. The sample for the study is composed of four tributes delivered by Akuffo Addo, Mahama, Kufour and Rawlings to the late Queen Mother of the Ashanti Kingdom, Nana Afia Kobi Serwaa Apem II. The data were manually coded using the consensual coding strategy. The results indicate a preponderant use of relational processes to identify the unique qualities possessed by the late Queen Mother which further unveil the varied relationships between her and the presidents. Other process types such as material, mental, verbal and behavioural processes are used to construe the positive actions carried out by the late Queen Mother during her lifetime, activate the minds and emotions of mourners and sympathizers regarding the loss, posthumously project the late Queen Mother as a legend who needs to be modelled after and present a collective purgation of emotions of pain and agony. The study concludes that, despite the apolitical status of traditional rulers, they still have a latent but cordial relationship with politicians.
 
 Citation:Mwinwelle, P., Duah, I. and Ernest Mensah, S. (2021). Eulogising the Dead: A Systemic Functional Exploration of Tributes Delivered by Ghanaian Presidents. International Journal of Technology and Management Research (IJTMR), Vol. 6 (2): Pp.38-58.
 Received: April 15, 2020Accepted: September 1, 2021

Global Land Surface Dry/Wet Conditions Mining Based on Spatial‐Temporal Association Rules

AbstractSevere drought and wetness would have serious impacts on human society and natural environment. Discovering the implicit relationships among severely dry/wet conditions is of great significance for meteorological disaster early warning and risk management strategy formulation. In this study, we propose an improved spatial‐temporal association rule mining algorithm to mine the global dry/wet associations. A modified time‐lagged association rule mining algorithm is first developed to mine unit‐to‐unit dry/wet association rules. Then, a density‐based clustering algorithm is employed to identify associated dry/wet zones. The gridded Standardized Precipitation Evapotranspiration Index (SPEI) data sets at 12‐ and 3‐month time scales were used to characterize annual and seasonal dry/wet conditions. Analysis results show that there are strong dry/wet associations between many regions. Some of the discovered association rules reflect similar associations with known phenomena, indicating the effectiveness and feasibility of the method. Other rules were previously unknown and can provide new knowledge for this research field. Several predictions can be made according to the discovered rules that western Oman and southeastern Saudi Arabia would suffer severe or extreme drought in 2021 and northwestern China would suffer severe or extreme drought in 2025 with probabilities exceeding 77%; northwestern China, northern Argentina, and western Australia would experience severe or extreme wetness in 2021, 2022, and 2023, respectively, with probabilities exceeding 85%. The algorithm provides a new perspective for analyzing dry/wet associations and the discovered associations can be used as theoretical bases for local governments to take precautions to mitigate the potential impacts of severe drought or wetness.

A Multi-Perspective malware detection approach through behavioral fusion of API call sequence

The widespread development of the malware industry is considered the main threat to our e-society. Therefore, malware analysis should also be enriched with smart heuristic tools that recognize malicious behaviors effectively. Although the generated API calling graph representation for malicious processes encodes worthwhile information about their malicious behavior, it is pragmatically inconvenient to generate a behavior graph for each process. Therefore, we experimented with creating generic behavioral graph models that describe malicious and non-malicious processes. These behavioral models relied on the fusion of statistical, contextual, and graph mining features that capture explicit and implicit relationships between API functions in the calling sequence. Our generated behavioral models proved the behavioral contrast between malicious and non-malicious calling sequences. According to that distinction, we built different relational perspective models that characterize processes’ behaviors. To prove our approach novelty, we experimented with our approach over Windows and Android platforms. Our experimentations demonstrated that our proposed system identified unseen malicious samples with high accuracy with low false-positive. In terms of detection accuracy, our model returns an average accuracy of 0.997 and 0.977 to the unseen Windows and Android malware testing samples, respectively. Moreover, we proposed a new indexing method for APIs based on their contextual similarities. We also suggested a new expressive, a visualized form that renders the API calling sequence. Consequently, we introduced a confidence metric to our model classification decision. Furthermore, we developed a behavioral heuristic that effectively identified malicious API call sequences that were deceptive or mimicry.

ACMF: An Attention Collaborative Extended Matrix Factorization Based Model for MOOC course service via a heterogeneous view

The spouting development of Massive Open Online Courses (MOOC) has enabled any learner to obtain abundant resource anytime and anywhere, offered a large-scale and open-access opportunity for learners to gain knowledge with great convenience. However, the surge of both learners and courses is engendering a sequence of issues, such as learner cognitive overload and messy course resources. The existing course resource service is not superior enough to suffice the dynamic and diverse demands of different individuals. Simultaneously, MOOC data suffers from sparsity and imbalance conundrums. To address those issues, we construct a heterogeneous information network (HIN) on MOOC to capture the plentiful heterogeneity between multiple entities and then propose an Attention Collaborative Extended Matrix Factorization Based Model, short as ACMF, for personalized recommendation service on MOOC course. In addition to learner and course subjects, we consider other four types of entities (e.g., knowledge, teacher, university, video) to give in-depth insights into why a learner is likely to enroll one course. Graph Convolutional Network with a novel node sampling algorithm based meta-path is utilized to learn the preserving structures and semantics in the MOOC HIN. Attention Mechanism is leveraged to automatically generate the final joint representation embedding. ACMF incorporates both explicit and implicit relationships via a heterogeneous view, integrates both learner and course embeddings via a specific semantic space. The ultimate personalized course recommendation list can be further adjusted based on learner-related or course-related through two tuning parameters. Experiments results on two real-world datasets collected from XuetangX and Junyi demonstrate the robustness and superiorities of ACMF.

The Role of Value-Semantic Oppositions in Revealing the Deep Literary and Psychological Characteristics of the Narrator’s Image in “Poor Liza” by N.M. Karamzin

Value-semantic oppositions in the novel by N.M. Karamzin, seen as meaningful text dominants in situations of "existential fullness", are discussed in the article from the point of view of an interdisciplinary approach of literary criticism, text theory, psychology and linguistics. The article describes the process of literary and psychological meaning generation, coupled with the narrator’s inherent properties, as well as implicit literary and psychological textual relationships with inherent properties, it also singles out the contextual ways of these characteristics, reflecting motives, attitudes, generalized worldview judgments (generalizations). The author shows the dialog nature of the narrator’s literary and psychological attitude to the realities denoted by various parts of the opposition in the situation of “existential fullness”. The role of lexical markers corresponding to the depicted integral emotional state of the narrator in the privision of active reader's attention and attracting him to the deep literary and psychological characteristics is revealed.

Detecting vulnerability in source code using CNN and LSTM network

Automated vulnerability detection has become a research hot spot because it is beneficial for improving software quality and security. The code metric (CM) is one class of important representations of vulnerability in source code. The implicit relationships among different metric attributes have not been sufficiently considered in traditional vulnerability detection based on CMs. In this paper, in view of the local perception capability of convolutional neural network (CNN) and the time-series prediction capability of long short-term memory (LSTM), we propose VulExplore, a compound neural network model for vulnerability detection that consists of a CNN for feature extraction and an LSTM network for deep representation. Moreover, to further indicate the vulnerability features in the source code, we reconstruct a CM dataset that includes two additional important attributes: maintainability index and average number of vulnerabilities committed per line. Our proposed numerical method can obtain both false-negative rate (FNR) and false-positive rate (FPR) under 20% and, meanwhile, achieve recall and precision over 80%, respectively.

Correlations between apparent diffusion coefficient values of WB-DWI and clinical parameters in multiple myeloma

BackgroundWhole-body diffusion-weighted imaging (WB-DWI) is a method for evaluating bone marrow infiltration in multiple myeloma (MM). This study seeks to elucidate the correlation between the apparent diffusion coefficient (ADC) value and some selected clinical parameters.MethodsA total of 101 Chinese patients with MM who had undergone WB-DWI from May 2017 to May 2019 were enrolled in this study. The ADC values of the MM lesions and the clinical parameters were quantified at the first (baseline) visit and after four-course induction chemotherapy. Multiple linear regression and logistic analyses were carried out to find the implicit inherent relationships within the patients’ data.ResultsThe paired Wilcoxon test showed that the ADC values at the baseline visit (ADC0) were significantly lower than the values after four-course induction chemotherapy (ADC4 c) (p < 0.001), including different therapeutic responses. The Revised International Staging System (RISS) stage, type of MM, and β2-microglobulin (β2-MG) were predictors of clinically significant increases or decreases in the ADC values (p < 0.05). Multiple linear regression showed that the ADC0 was negatively associated with β2-MG (p < 0.001) and immunoglobulin heavy chain gene rearrangement (p = 0.012), while the RISS Stage III (p = 0.001), type IgG λ (p = 0.005), and albumin were negatively associated with ADC4 c (p = 0.010). The impacts of the therapeutic response were associated with ADC0 and immunoglobulin heavy chain gene rearrangement (p < 0.001).ConclusionThe ADC values of WB-DWI may be associated with clinical parameters of MM including the fluorescence in situ hybridization result, and may be useful in the prognosis of patients with MM.Trial Registration: ChiCTR2000029587

Multi-view Dynamic Heterogeneous Information Network Embedding

Abstract Most existing heterogeneous information network (HIN) embedding methods focus on static environments while neglecting the evolving characteristic of real-world networks. Although several dynamic embedding methods have been proposed, they are merely designed for homogeneous networks and cannot be directly applied in heterogeneous environments. To tackle above challenges, we propose a novel framework for incorporating temporal information into HIN embedding, named multi-view dynamic HIN embedding (MDHNE), which can efficiently preserve evolution patterns of implicit relationships from different views in updating node vectors over time. We first transform HIN to a series of homogeneous networks corresponding to different views. Then our proposed MDHNE applies recurrent neural network (RNN) to incorporate evolving pattern of complex network structure and semantic relationships between nodes into latent embedding spaces, and thus the node vectors from multiple views can be learned and updated when HIN evolves over time. Moreover, we come up with an attention-based fusion mechanism, which can automatically infer weights of latent vectors corresponding to different views by minimizing the objective function specific for different mining tasks. Extensive experiments clearly demonstrate that our MDHNE model outperforms state-of-the-art baselines on three real-world dynamic datasets for different network mining tasks.

Adopting explicit and implicit social relations by SVD++ for recommendation system improvement

Recommender systems suffer a set of drawbacks such as sparsity. Social relations provide a useful source to overcome the sparsity problem. Previous studies have utilized social relations or rating feedback sources. However, they ignored integrating these sources. In this paper, the limitations of previous studies are overcome by exploiting four sources of information, namely: explicit social relationships, implicit social relationships, users’ ratings, and implicit feedback information. Firstly, implicit social relationships are extracted through the source allocation index algorithm to establish new relations among users. Secondly, the similarity method is applied to find the similarity between each pair of users who have explicit or implicit social relations. Then, users’ ratings and implicit rating feedback sources are extracted via a user-item matrix. Furthermore, all sources are integrated into the singular value decomposition plus (SVD++) method. Finally, missing predictions are computed. The proposed method is implemented on three real-world datasets: Last.Fm, FilmTrust, and Ciao. Experimental results reveal that the proposed model is superior to other studies such as SVD, SVD++, EU-SVD++, SocReg, and EISR in terms of accuracy, where the improvement of the proposed method is about 0.03% for MAE and 0.01% for RMSE when dimension value (d) = 10.

Implicit Emotion Relationship Mining Based on Optimal and Majority Synthesis From Multimodal Data Prediction

Emotion is precious, useful for many applications such as public opinion detection and psychological disease prediction. Emotion recognition on multimodal data has attracted extensive attention. Although modifying model structure or multimodal feature fusion methods have contributed a lot to emotion recognition, little attention is paid to mining implicit emotion relationship. In this article, implicit emotion relationship consists of emotion distribution, confusion, and transfer. Emotion distribution allows multiple emotions in one sample, while confusion and transfer imply the prediction confusion and bias. In order to mine implicit emotion relationship in multimodal data, this article employs three image and two text classification models to recognize emotions, respectively. Two prediction emotion synthesis methods (optimal prediction emotion synthesis and majority prediction emotion synthesis) are proposed to synthesize the outputs of multiple models. Based on the results of two emotion synthesis methods, emotion distribution on samples is obtained. Emotion confusion and transfer among different emotion samples are analyzed by relative entropy and Jensen–Shannon divergence. Implicit emotion relationship mining has potential not only in the interpretation of model performance, but also in guiding the development of emotion recognition as prior knowledge. Finally, we take topic scenario as an instance to mine implicit emotion relationships.

Developing a unified centralized transport flow control system

Currently, there is an increase in information for data mining in transport systems, the main reason is the increase in the number of heterogeneous sources. The relevance of the topic lies in the need to collect, process, aggregate, and model large volumes of unstructured information that cannot be effectively processed by traditional methods. With the increasing flow of vehicles, its diversity, there is a need to optimize the processes of transportation and logistics, increase the system safety of road traffic. The creation of an information knowledge base will help to solve a number of important problems, including: the efficiency of road use, reduction of toxic emissions, control and unloading of traffic flows, reduction in the number of accidents, and prompt notification of services.The idea of developing a unified centralized traffic control system is described. To collect, store and process heterogeneous information, it is proposed to use a cloud infrastructure with split computation. For the purpose of high-quality processing and aggregation of heterogeneous information, it is recommended to investigate hidden dependencies in the data, build and analyze various aggregation options and interpret them in relation to specific tasks.The system should connect all participants in ground traffic, collect dissimilar materials that can be obtained from their devices and a variety of sensors, and also automate the management and decision-making in transport systems. Unstructured information must be correctly interpreted, categorized, and consistently labeled to identify implicit relationships between data.The scientific novelty of the research consists in the formation of the functions of the system being developed, the description of the main aspects, requirements, interfaces, models and methods for aggregating heterogeneous data.The results of the work can be used not only for analyzing big data in the field of transport, but also in other directions when solving problems of processing heterogeneous information.