Search

Action quality assessment is a significant research domain in computer vision, aimed at evaluating the accuracy of human movement and providing feedback and guidance for training and rehabilitation. However, the uneven nature of the data, which has a significant impact on the labels with less samples, is not taken into consideration by the generally used approaches in this field. To address this issue, we propose using kernel density estimation (KDE) to recalculate the label density and weight the loss function by the reciprocal of the square root of each label density. Additionally, we divide the entire motion into three sub-stages, including the takeoff, aerial movement, and entry for diving, and connect the three stages using an across-staged temporal reasoning module (ASTRM). Our approach achieves a performance of 0.9222 Spearman correlation coefficient (ρ ) and 0.3304 (× 100) Relative ℓ 2 -distance (R - ℓ 2 ) on the FineDiving dataset, demonstrating competitiveness compared to other methods. Furthermore, numerous comprehensive ablation experiments validate the effectiveness of the methods and modules we adopted. [ABSTRACT FROM AUTHOR]

Question Answering Over Temporal Knowledge Graphs is an important topic in question answering, which aims to find an entity or timestamp to answer temporal reasoning questions from temporal knowledge graphs. Answering complex questions remains a major challenge for question answering over temporal knowledge graphs because it is associated with complex temporal reasoning. The performance of the existing state-of-the-art model falls short when the question contains constraints (e.g., 'before/after', 'first/last' and 'during') that require complex temporal reasoning based on multiple relevant facts. In this paper, we propose an improving reasoning method called the Complex Temporal Reasoning Network, which improves the complex temporal reasoning for temporal reasoning questions. For each question, we capture implicit temporal features and relation representation and then integrate them to generate implicit temporal relation representation. The experimental results on the CRONQUESTIONS dataset demonstrate that our method significantly outperforms all baselines. In particular, we demonstrate the effectiveness of our method on complex questions. The source code of CTRN will be available at https://github.com/2399240664/CTRN. [ABSTRACT FROM AUTHOR]

In this paper we advocate the use of multi-dimensional modal logics as a framework for knowledge representation and, in particular, for representing spatio-temporal information. We construct a two-dimensional logic capable of describing topological relationships that change over time. This logic, called PSTL (Propositional Spatio-Temporal Logic) is the Cartesian product of the well-known temporal logic PTL and the modal logic S4u, which is the Lewis system S4 augmented with the universal modality. Although it is an open problem whether the full PSTL is decidable, we show that it contains decidable fragments into which various temporal extensions (both point-based and interval based) of the spatial logic RCC-8 can be embedded. We consider known decidability and complexity results that are relevant to computation with multi-dimensional formalisms and discuss possible directions for further research. [ABSTRACT FROM AUTHOR]

Knowledge graph embedding (KGE) is a method designed to predict missing relations between entities in a knowledge graph (KG), which has garnered much attention in recent years due to the incompleteness of KGs. However, existing KGE models have limitations in dealing with heterogeneous KGs and relation direction prediction. To address this issue, a novel KGE model called BHRAM is proposed. It is based on a bidirectional and heterogeneous relational attention mechanism. Specifically, BHRAM comprises three primary components, namely entity aggregation, relation aggregation and triplet prediction. The entity aggregation module divides the adjacency matrix into original and reverse relation adjacency matrices, using graph convolution to aggregate node features and subsequently form entity embedding representations. The relation aggregation module leverages bidirectional relations for feature extraction, learns the weight information of diverse paths independently and generates embedding representations of relation paths through an aggregation function. Finally, the triplet prediction module utilizes a score function for probabilistic predictions. To validate the superiority of BHRAM, comprehensive experiments were conducted on four well-known datasets, including baseline comparisons, relation classification tasks and ablation study. The results demonstrate that BHRAM significantly outperforms the other baselines on the FB15k-237, Kinship and UMLS datasets, while achieving similar or better performance than the baselines on the WN18RR dataset. These findings indicate that BHRAM can serve as a robust and effective model for addressing the heterogeneity in KGs. [ABSTRACT FROM AUTHOR]

Temporal knowledge graph completion (TKGC) focuses on inferring missing facts from temporal knowledge graphs (TKGs) and has been widely studied. While previous models based on graph neural networks (GNNs) have shown noteworthy outcomes, they tend to focus on designing complex modules to learn contextual representations. These complex solutions require a large number of parameters and heavy memory consumption. Additionally, existing TKGC approaches focus on exploiting static feature representation for entities and relationships, which fail to effectively capture the semantic information of contexts. In this paper, we propose a global and local information-aware relational graph convolutional neural network (GLARGCN) model to address these issues. First, we design a sampler, which captures significant neighbors by combining global historical event frequencies with local temporal relative displacements and requires no additional learnable parameters. We then employ a time-aware encoder to model timestamps, relations, and entities uniformly. We perform a graph convolution operation to learn a global graph representation. Finally, our method predicts missing entities using a scoring function. We evaluate the model on four benchmark datasets and one specific dataset with unseen timestamps. The experimental results demonstrate that our proposed GLARGCN model not only outperforms contemporary models but also shows robust performance in scenarios with unseen timestamps. [ABSTRACT FROM AUTHOR]

Bayesian Networks (BNs) are probabilistic graphical models used to represent variables and their conditional dependencies, making them highly valuable in a wide range of fields, such as radiology, agriculture, neuroscience, construction management, medicine, and engineering systems, among many others. Despite their widespread application, the reusability of BNs presented in papers that describe their application to real-world tasks has not been thoroughly examined. In this paper, we perform a structured survey on the reusability of BNs using the PRISMA methodology, analyzing 147 papers from various domains. Our results indicate that only 18% of the papers provide sufficient information to enable the reusability of the described BNs. This creates significant challenges for other researchers attempting to reuse these models, especially since many BNs are developed using expert knowledge elicitation. Additionally, direct requests to authors for reusable BNs yielded positive results in only 12% of cases. These findings underscore the importance of improving reusability and reproducibility practices within the BN research community, a need that is equally relevant across the broader field of Artificial Intelligence. [ABSTRACT FROM AUTHOR]

Few-shot action recognition aims to learn a model that can be easily adapted to identify novel action classifications using only a few labeled samples. Recent methods primarily focus on visual features and fail to fully utilize the available classification title of the video. In addition, they capture higher-order temporal relationships among video frames through averaging, which neglects the long-range dependencies information of the video. To address these issues, we designed a novel cross-modal guided spatio-temporal enrichment network (X-STEN) for few-shot action recognition. The model includes a cross-modal spatial enrichment module (X-SEM), a temporal enrichment module (TEM), and a non-parametric metrics module (NMM). Firstly, we extract and fuse multi-modal feature representations of videos. Then, we enhance the spatial context information of the video using the X-SEM and model the temporal context information of the video using the TEM. Finally, we generate the query and support prototypes and measure the similarity between them. Extensive experiments demonstrate that our X-STEN achieve excellent results on few-shot splits of Kinetics, HMDB51 and UCF101. Importantly, our method outperforms prior work on Kinetics by a wide margin (13.9%). [ABSTRACT FROM AUTHOR]

Temporal knowledge graphs (TKGs) are a form of knowledge representation constructed based on the evolution of events at different time points. It provides an additional perspective by extending the temporal dimension for a range of downstream tasks. Given the evolving nature of events, it is essential for TKGs to reason about non-existent or future events. Most of the existing models divide the graph into multiple time snapshots and predict future events by modeling information within and between snapshots. However, since the knowledge graph inherently suffers from missing data and uneven data distribution, this time-based division leads to a drastic reduction in available data within each snapshot, which makes it difficult to learn high-quality representations of entities and relationships. In addition, the contribution of historical information changes over time, distinguishing its importance to the final results when capturing information that evolves over time. In this paper, we introduce CH-TKG (Contrastive Learning and Historical Information Learning for TKG Reasoning) to addresses issues related to data sparseness and the ambiguity of historical information weights. Firstly, we obtain embedding representations of entities and relationships with evolutionary dependencies by R-GCN and GRU. On this foundation, we introduce a novel contrastive learning method to optimize the representation of entities and relationships within individual snapshots of sparse data. Then we utilize self-attention and copy mechanisms to learn the effects of different historical data on the final inference results. We conduct extensive experiments on four datasets, and the experimental results demonstrate the effectiveness of our proposed model with sparse data. [ABSTRACT FROM AUTHOR]

Temporal Knowledge Graph Completion (TKGC) aims to infer missing facts in Temporal Knowledge Graphs (TKGs), where facts are stored along with significant temporal information. However, existing TKGC methods only consider message passing on pairwise relations and fail to capture the complex temporal structural dependencies at the levels of time, predicate and entity in TKGs. To fill this gap, we collect high-frequency patterns in TKGs using mathematical statistics and propose a Temporal-Structural Adaptation Network that is equipped with three specific components, time-component, pred-component, and ent-component, as well as one general component, res-component. Concretely, specific components utilize the time consistency pattern to capture facts with significant regularity in time, and complex structural dependencies in TKGs are handled through predicate concurrency and entity collaboration. Moreover, considering low-frequency and nonoccurrence facts, an additional general component is introduced to make predictions on all entities. The outputs of different components are adaptively fused to vote for the final result. Extensive experiments on six benchmarks demonstrate that our method outperforms state-of-the-art baselines. [ABSTRACT FROM AUTHOR]

Machine learning(ML) based techniques for Land Use Land Cover(LULC) classification is crucial for extracting valuable insights from satellite imagery. The impact of severe class imbalance problems on LULC datasets, added to the limited temporal coverage, incomplete spectral information, spatial resolution constraints, and limitations due to sensor characteristics of a single satellite, hinder the efficient capturing of complex image features. The object-based classification using Gray-level co-occurrence matrix(GLCM) and Simple Non-Iterative Clustering(SNIC) captures the textural and spectral information, respectively, to enhance the accuracy in heterogeneous landscapes and overcome the limitations of pixel-based classification, such as the sensitivity towards noise and spectral confusion in mixed pixels at the cost of increased additional computational steps. In this context, the proposed study leverages high-quality fused images with diverse temporal, spectral, and spatial information obtained by fusing Landsat-8 and Sentinel-2 satellite imageries using Spatial-and-Temporal-Adaptive-Reflectance-Fusion-Model(STARFM). Further, to solve the class imbalance problem, a Granular Computing(GrC) based Segmentation and Textural Analysis(GrCSTA) framework is proposed for reducing the number of image primitives and computations required for subsequent image analysis processes in the object-based classification. The GrCSTA framework focuses on extracting the Spatial Granules(Gs) from the fused imagery using Spatial neighborhood Granulation(SNGr), textural indices of the Gs using GLCM, and reduced textural indices using Principal Component Analysis(PCA). Gs and its reduced textural indices are input features to train the Random Forest(RF) classifier. Experimental results demonstrate that the proposed GrCSTA framework achieves comparably higher accuracy than the state-of-the-art models. [ABSTRACT FROM AUTHOR]

We develop the scheme of indefinite constraint databases using first-order logic as our representation language. When this scheme is instantiated with temporal constraints, the resulting formalism is more expressive than standard temporal constraint networks. The extra representational power allows us to express temporal knowledge and queries that have been impossible to express before. To make our claim more persuasive, we survey previous works on querying temporal constraint networks and show that they can be viewed as an instance of the scheme of indefinite constraint databases. [ABSTRACT FROM AUTHOR]

This paper describes an AI system for planning the first attack on a forest fire. This planning system is based on two major techniques, case-based reasoning, and constraint reasoning, and is part of a decision support system called CHARADE. CHARADE is aimed at supporting the user in the whole process of forest fire management. The novelty of the proposed approach is mainly due to the use of a local similarity metric for case-based reasoning and the integration with a constraint solver in charge of temporal reasoning. [ABSTRACT FROM AUTHOR]

Temporal knowledge graphs completion (TKGC) is a critical task that aims to forecast facts that will occur in future timestamps. It has attracted increasing research interest in recent years. Among the many approaches, reinforcement learning-based methods have gained attention due to their efficient performance and interpretability. However, these methods still face two challenges in the prediction task. First, a single policy network lacks the capability to capture the dynamic and static features of entities and relationships separately. Consequently, it fails to evaluate candidate actions comprehensively from multiple perspectives. Secondly, the composition of the action space is incomplete, often guiding the agent towards distant historical events and missing the answers in recent history. To address these challenges, this paper proposes a Temporal Knowledge Graph Completion Based on a Multi-Policy Network(MPNet). It constructs three policies from the aspects of static entity-relation, dynamic relationships, and dynamic entities, respectively, to evaluate candidate actions comprehensively. In addition, this paper creates a more diverse action space that guides the agent in investigating answers within historical subgraphs more effectively. The effectiveness of MPNet is validated through an extrapolation setting, and extensive experiments conducted on three benchmark datasets demonstrate the superior performance of MPNet compared to existing state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Few-shot action recognition is a challenging problem aimed at learning a model capable of adapting to recognize new categories using only a few labeled videos. Recently, some works use attention mechanisms to focus on relevant regions to obtain discriminative representations. Despite the significant progress, these methods still cannot attain outstanding performance due to insufficient examples and a scarcity of additional supplementary information. In this paper, we propose a novel Semantic-guided Spatio-temporal Attention (SGSTA) approach for few-shot action recognition. The main idea of SGSTA is to exploit the semantic information contained in the text embedding of labels to guide attention to more accurately capture the rich spatio-temporal context in videos when visual content is insufficient. Specifically, SGSTA comprises two essential components: a visual-text alignment module and a semantic-guided spatio-temporal attention module. The former is used to align visual features and text embeddings to eliminate semantic gaps between them. The latter is further divided into spatial attention and temporal attention. Firstly, a semantic-guided spatial attention is applied on the frame feature map to focus on semantically relevant spatial regions. Then, a semantic-guided temporal attention is used to encode the semantically enhanced temporal context with a temporal Transformer. Finally, use the spatio-temporally contextual representation obtained to learn relationship matching between support and query sequences. In this way, SGSTA can fully utilize rich semantic priors in label embeddings to improve class-specific discriminability and achieve accurate few-shot recognition. Comprehensive experiments on four challenging benchmarks demonstrate that the proposed SGSTA is effective and achieves competitive performance over existing state-of-the-art methods under various settings. [ABSTRACT FROM AUTHOR]

Poor health is one of the fundamental causes behind the suffering and deprivation of human beings. One of the United Nations (UN) Sustainable Development Goals is to enhance the quality of healthcare for everyone, which includes economic coverage, availability of high-quality fundamental health-care services, and access to proper, efficient, high-quality, and affordable important vaccinations and medications. Question-Answering (QA) in the medical domain has recently piqued the interest among the researchers and other stakeholders. Medical QA systems have the potential to enhance access to healthcare services, improve patient interactions with doctors, and reduce medical costs through e-medicine. In this paper, we describe a knowledge enabled QA model, which demonstrates how large-scale medical information in the form of knowledge graphs can aid in extracting more relevant answers. The proposed model employs two scoring methods, viz.Medical Entity Scoring (MES) and Context Relevance Scoring (CRS). MES ranks the medical entities from graphs according to their relevance, while CRS is used to reason over the supporting paragraph using the query vector. The system's knowledge is obtained through the use of two distinct resources, viz. PharmKG is used for pharmaceutical terminology management, whereas Unified Medical Language System UMLS is used for general medical terminology management. Empirical results on the MASH-QA and COVID-QA datasets demonstrate that our proposed approach outperforms existing State-of-the-art in both machine evaluation and human judgment. [ABSTRACT FROM AUTHOR]

Video action segmentation have been widely applied in many fields. Most previous studies employed video-based vision models for this purpose. However, they often rely on a large receptive field, LSTM or Transformer methods to capture long-term dependencies within videos, leading to significant computational resource requirements. To address this challenge, graph-based model was proposed. However, previous graph-based models are less accurate. Hence, this study introduces a graph-structured approach named Semantic2Graph, to model long-term dependencies in videos, thereby reducing computational costs and raise the accuracy. We construct a graph structure of video at the frame-level. Temporal edges are utilized to model the temporal relations and action order within videos. Additionally, we have designed positive and negative semantic edges, accompanied by corresponding edge weights, to capture both long-term and short-term semantic relationships in video actions. Node attributes encompass a rich set of multi-modal features extracted from video content, graph structures, and label text, encompassing visual, structural, and semantic cues. To synthesize this multi-modal information effectively, we employ a graph neural network (GNN) model to fuse multi-modal features for node action label classification. Experimental results demonstrate that Semantic2Graph outperforms state-of-the-art methods in terms of performance, particularly on benchmark datasets such as GTEA and 50Salads. Multiple ablation experiments further validate the effectiveness of semantic features in enhancing model performance. Notably, the inclusion of semantic edges in Semantic2Graph allows for the cost-effective capture of long-term dependencies, affirming its utility in addressing the challenges posed by computational resource constraints in video-based vision models. [ABSTRACT FROM AUTHOR]

The task of inferring missing facts using the temporal knowledge graph (TKG) is important and has been widely studied. Extrapolation in TKG inference is more challenging because no direct historical facts are available for predicting future events. Previous methods apply recursive embedding learning to solve the extrapolation problem. However, these methods primarily focus on recent historical facts and overlook the potential knowledge hidden in earlier facts. Moreover, the recursive embedding learning process can lead to imprecise knowledge propagation. To address these issues, we propose a memory-triggered decision-making (MTDM) network. It takes advantage of earlier historical facts to establish initial node representations and then updates node representations with recent historical facts. In particular, to enhance the prediction performance and efficiency, we update node representations of each timestamp in parallel using the proposed Res-GCN and then establish temporal correlations among these representations based on the designed control gating unit. To mitigate the issue of imprecise knowledge propagation, we leverage the facts most relevant to the missing knowledge to directly make predictions. This strategy improves the predictions by focusing on the most informative and contextual-free information. Additionally, we introduce the dissolution constraint, which enables us to analyse the process of event dissolution. Extensive experiments demonstrate that MTDM outperforms existing methods in terms of prediction accuracy and computational efficiency. [ABSTRACT FROM AUTHOR]

Most of the cutting-edge 3D convolutional-based video saliency prediction models adopt a fully convolutional encoder-decoder architecture, which provides a good spatiotemporal representation of salient regions. However, the spatiotemporal cues of the saliency models will be continuously diluted during the decoding process, leading to a decreased ability to locate salient regions. To address this limitation, we propose a simple and effective gated fusion network (GFNet) to conduct video saliency prediction. Specifically, GFNet is built on a fully 3D convolutional encoder-decoder architecture and consists of a key component named the gated fusion (GF) module, which acts as a message screening unit between each level encoder and decoder features. In the GF module, the gate can be obtained via the combination of features from the previous decoder block and the current encoder block, which is employed to weight the encoder features. In this way, GFNet can control the message passing between each level encoder and decoder block. Extensive experimental results on four video saliency datasets show that our method achieves comparable performance against state-of-the-art models. The code is available at https://github.com/wusonghe/GFNet. [ABSTRACT FROM AUTHOR]

Temporal knowledge graphs (TKGs) extend traditional knowledge graphs by incorporating temporal information to represent valid time. TKG reasoning involves inferring and predicting missing or future facts within various time intervals, and applications in critical domains include political event prediction and financial analysis. Current TKG reasoning methods rely on path-based or subgraph-based strategies to achieve enhanced interpretability. However, they mostly focus on entity interaction information within paths or subgraphs, and they fail to exploit the semantic and temporal information of relations during reasoning. To comprehensively utilize such information for refining the reasoning process, we propose a subgraph reasoning model based on time-aware relation representation (TiAR) for TKGs. TiAR learns the subgraph structures around query nodes and reasons on facts happening in future timestamps. It encodes the relation features by combining the temporal displacement values between events and extracts the semantic correlations between query and adjacency relations via an attention-based neighborhood feature aggregation module. TiAR extracts a query subgraph of interest by iteratively sampling and diffusing attention through the whole graph. In particular, TiAR includes a path constraint module to further capture the semantic correlations between the structural features of queries and subgraphs. Our evaluation conducted on four benchmark datasets demonstrates that TiAR outperforms the current state-of-the-art models, indicating its effectiveness in achieving improved TKG reasoning. In particular, for the two large ICEWS05-15/18 datasets, our model yields better performance than the latest model. Moreover, The development of TKG reasoning methods is essential for improving the reliability and interpretability of models in such domains, and our work contributes to this area of research. [ABSTRACT FROM AUTHOR]

Knowledge graph completion (KGC) has been widely explored, but the task of temporal knowledge graph completion (TKGC) for predicting future events is far from perfection. Some embedding-based approaches have achieved significant results on the TKGC task by modeling the structural information of each temporal snapshot and the evolution between temporal snapshots. However, due to the uneven distribution of data in knowledge graphs (KGs), models that only utilize local structure and time series information suffer from information sparsity, resulting in some entities failing to obtain a better embedding representation due to less available information. Moreover, existing methods usually do not distinguish between the time span and frequency of historical information, which reduces the performance of link prediction. For this reason, we propose the G lobal and L ocal Information-A ware Net work (GL-ANet) to capture both global and local information. In particular, to model global information, we capture global structural information of entities across time using a global neighborhood aggregator to enrich the representation of entities; global historical information is obtained based on the frequency and time span of historical facts, focusing on recent and frequent events rather than all historical events to suggest the performance of link prediction; to model local information, we propose a two-layer attention network to capture local structural information at each timestamp, using a gating mechanism and GRU to capture local evolution information. Extensive experiments demonstrate the effectiveness of our model, achieving significant improvements and outperforming state-of-the-art models on five benchmark datasets. [ABSTRACT FROM AUTHOR]

Temporal knowledge graphs (TKGs) have become an effective tool for numerous intelligent applications. Due to their incompleteness, TKG embedding methods have been proposed to infer the missing temporal facts, and work by learning latent representations for entities, relations and timestamps. However, these methods primarily focus on measuring the plausibility of the whole temporal fact, and ignore the semantic property that there exists a bias between any relation and its involved entities at various time steps. In this paper, we present a novel temporal knowledge graph completion framework, which imposes relational constraints to preserve the semantic property implied in TKGs. Specifically, we borrow ideas from two well-known transformation functions, i.e., tensor decomposition and hyperplane projection, and design relational constraints associated with timestamps. We then adopt suitable regularization schemes to accommodate specific relational constraints, which combat overfitting and enforce temporal smoothness. Experimental studies indicate the superiority of our proposal compared to existing baselines on the task of temporal knowledge graph completion. [ABSTRACT FROM AUTHOR]

Knowledge discovery combined with network structure is an emerging field of network data analysis and mining. Three-way concept analysis is a method that can fit the human mind in uncertain decisions and analysis. In reality, when three-way concept analysis is placed in the background of a network, not only the three-way rules need to be obtained, but also the network characteristic values of these rules should be obtained, which is of great significance for concept cognition in the network. This paper mainly combines complex network analysis with the formal context of three-way decision. Firstly, the network formal context of three-way decision (NFC3WD) is proposed to unify the two studies mentioned above into one data framework. Then, the network weaken-concepts of three-way decision (NWC3WD) and their corresponding sub-networks are studied. Therefore, we can not only find out the network weaken-concepts but also know the average influence of the sub-network, as well as the influence difference within the sub-network. Furthermore, the concept logic of network and the properties of its operators are put forward, which lays a foundation for designing the algorithm of rule extraction. Subsequently, the bidirectional rule extraction algorithm and reduction algorithm based on confidence degree are also explored. Meanwhile, these algorithms are applied to the diagnosis examples of COVID-19 from which we can not only get diagnostic rules, but also know the importance of the population corresponding to these diagnostic rules in the network through network eigenvalues. Finally, experimental analysis is made to show the superiority of the proposed method. [ABSTRACT FROM AUTHOR]

Knowledge graph completion (KGC) can be interpreted as the task of missing inferences to real-world facts. Despite the importance and abundance of temporal knowledge graphs, most of the current research has been focused on reasoning on static knowledge graphs. The data they are applied to usually evolves with time, such as friend graphs in social networks. Therefore, developing temporal knowledge graph completion (temporal KGC) models is an increasingly important topic, although it is difficult due to data non-stationarity, and its complex temporal dependencies. In this paper, we propose block decomposition based on relational interaction for temporal knowledge graph completion (TBDRI), a novel model based on block term decomposition (which can be seen as a special variant of CP decomposition and Tucker decomposition) of the binary tensor representation of knowledge graph quadruples. TBDRI considers that inverse relations, as one of the most important types of relations, occupy an important share in the real world. Although some existing models introduce inverse relation into the model, it is not enough to only learn the inverse relation independently. TBDRI learns inverse relation in an enhanced way to strengthen the binding of forward and inverse relation. Furthermore, TBDRI first uses the core tensor as temporal information to bind timestamps more adequately. We prove TBDRI is full expressiveness and derive the bound on its entity, relation, and timestamp embedding dimensionality. We show that TBDRI is able to outperform most previous state-of-the-art models on the four benchmark datasets for temporal knowledge graph completion. [ABSTRACT FROM AUTHOR]

Dynamic network link prediction is becoming a hot topic in network science, due to its wide applications in biology, sociology, economy and industry. However, it is a challenge since network structure evolves with time, making long-term prediction of adding/deleting links especially difficult. Inspired by the great success of deep learning frameworks, especially the convolution neural network (CNN) and long short-term memory (LSTM) network, we propose a novel end-to-end model with a Graph Convolution Network(GCN) embedded LSTM, named GC-LSTM, for dynamic network link prediction. Thereinto, LSTM is adopted as the main framework to learn the temporal features of all snapshots of a dynamic network. While for each snapshot, GCN is applied to capture the local structural properties of nodes as well as the relationship between them. One benefit is that our GC-LSTM can predict both added and removed links, making it more practical in reality, while most existing dynamic link prediction methods can only handle removed links. Extensive experiments demonstrated that GC-LSTM achieves outstanding performance and outperforms existing state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Upcoming space missions are requiring a higher degree of on-board autonomy operations to increase quality science return, to minimize closed-loop space-ground decision making, and to enable new scenarios. Artificial Intelligence technologies like Machine Learning and Automated Planning are becoming more and more popular as they can support data analytics conducted directly on-board as input for the on-board decision making system that generates plans or updates them while being executed. This paper describes the planning and execution architecture under development at the European Space Agency to target this need of autonomy for the ops-sat mission. [ABSTRACT FROM AUTHOR]

Automatically recognizing social relationships from videos provides intelligent systems with great potential to better understand the behaviors or emotions of human beings. Most existing methods mainly focus on inferring social characters by detecting their interactions or independently predicting each social relationship. However, they cannot directly learn all social relationships and characters. In this paper, we propose a character and relationship joint learning (CRJL) framework to simultaneously infer all social relationships and character pairs involved in videos. First, the video context and the logical associations among relationships provide important cues for social scene understanding. To incorporate these cues into social relationships and character reasoning, we design a novel character and relationship reasoning graph (CRRG). Specifically, we model the relationship passing process on the graph to learn the logical constraints among relationships. We also introduce a graph attention mechanism to capture discriminative video semantic information. Second, localizing a social character pair via supervised learning is time-consuming, as it requires the annotation of video tracks. Instead, we propose a weak label-based training strategy using clip-level relationships. Experimental results on a public benchmark demonstrate the superiority of our method. [ABSTRACT FROM AUTHOR]

Spatial puzzles composed of rigid objects, flexible strings and holes offer interesting challenges for reasoning about spatial entities that are common in the human daily-life's activities. This motivates the use of spatial puzzles as domains of study in this work. The goal of this paper is to investigate the automated solution of this kind of problems by extending an algorithm that combines Answer Set Programming (ASP) with Markov Decision Process (MDP) and Reinforcement Learning (RL), called oASP(MDP). This method is capable of constructing the set of domain states online, i.e., while the agent interacts with a changing environment. The aim of the extension proposed in this work is to add heuristics as a mechanism to accelerate the learning process, resulting in the main contribution of this paper: the Heuristic oASP(MDP) (HoASP(MDP)) algorithm. Experiments were performed on deterministic, non-deterministic and non-stationary versions of the puzzles. Results show that the proposed approach can considerably accelerate the learning process, outperforming other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Session-based recommendation is a task to recommend the next clicked item when the user's current interaction sequence is given. Accurately modeling the session representation is critical for session-based recommendation. However, we find that most current methods for session-based recommendation just use conscious behavior and information in the current session, ignoring the information of unconscious behavior in the current session and preference interaction with neighborhood sessions. In this paper, we propose a Mixed Behaviors and Preference Interaction model (MBPI), which utilizes conscious and unconscious behaviors and parallel co-attention mechanism, for session-based recommendation. In MBPI, we apply a Gated Recurrent Unit (GRU) to generate the session global preference, and employ another GRU with an item-level attention mechanism to explore the session local preference, with the multi-feature behaviors. Then, we introduce a parallel co-attention mechanism to capture the preference interaction with the help of the current session and neighborhood sessions and to update the two preferences of the current session. Finally, we combine the session global preference and session local preference as session representation and make recommendation. Experimental results on three real-world datasets show our method outperforms the state-of-the-art methods and validate the effectiveness of our approach. [ABSTRACT FROM AUTHOR]

Applied Intelligence is one of the most important international scientific journals in the field of artificial intelligence. From 1991, Applied Intelligence has been oriented to support research advances in new and innovative intelligent systems, methodologies, and their applications in solving real-life complex problems. In this way, Applied Intelligence hosts more than 2,400 publications and achieves around 31,800 citations. Moreover, Applied Intelligence is recognized by the industrial, academic, and scientific communities as a source of the latest innovative and advanced solutions in intelligent manufacturing, privacy-preserving systems, risk analysis, knowledge-based management, modern techniques to improve healthcare systems, methods to assist government, and solving industrial problems that are too complex to be solved through conventional approaches. Bearing in mind that Applied Intelligence celebrates its 30th anniversary in 2021, it is appropriate to analyze its bibliometric performance, conceptual structure, and thematic evolution. To do that, this paper conducts a bibliometric performance and conceptual structure analysis of Applied Intelligence from 1991 to 2020 using SciMAT. Firstly, the performance of the journal is analyzed according to the data retrieved from Scopus, putting the focus on the productivity of the authors, citations, countries, organizations, funding agencies, and most relevant publications. Finally, the conceptual structure of the journal is analyzed with the bibliometric software tool SciMAT, identifying the main thematic areas that have been the object of research and their composition, relationship, and evolution during the period analyzed. [ABSTRACT FROM AUTHOR]

Digitization of agricultural processes is advancing fast as telemetry data from the involved machines becomes more and more available. Current approaches commonly have a machine-centric view that does not account for machine-machine or machine-environment relations. In this paper we demonstrate how to model such relations in the generic semantic mapping framework SEMAP. We describe how SEMAP's core ontology is extended to represent knowledge about the involved machines and facilities in a typical agricultural domain. In the framework we combine different information layers – semantically annotated spatial data, semantic background knowledge and incoming sensor data – to derive qualitative spatial facts and continuously track them to generate process states and events about the ongoing logistic process of a harvesting campaign, which adds to an increased process understanding. [ABSTRACT FROM AUTHOR]

The Applied Intelligence (APIN) published its first issue in 1991 and it is one of the important academic journals in the field of computer science. This study aims to examine the structure and citation landscape as well as the evolution of the APIN journal based on all the APIN articles included in web of science core collection database. Research results of this study including the author characteristics, manuscript characteristics, productive and influential institutions and countries. Various kinds of networks are provided to visualize the complex relationships from different perspectives, including institutions and countries. The co-occurrence network and detection of the bursting keywords are used to explore the research hotspots in the APIN journal. The main advantage of this study is that it provides a comprehensive and objective view of the APIN journal. This study is very helpful for scholars to grasp the content structure and development process of APIN journal. It is also valuable for scholars in the field of artificial intelligence to identify the research hotspots in this field. [ABSTRACT FROM AUTHOR]

Many real-life dynamical systems experience abrupt changes followed by almost stationary periods. In this paper, we consider streams of data exhibiting such abrupt behavior and investigate the problem of tracking their statistical properties in an online manner. Wedevise a tracking procedure where an estimator that is suitable for a stationary environment is combined together with an estimator suitable for a dynamic environment. The current estimate is based on the stationary estimator unless a statistically significant difference is observed between both estimators. The stationary estimate is deemed off track and a large update (jump) is given to get the stationary estimate back on track. We use the Stochastic Learning Weak Estimator (SLWE) as the dynamic estimator. The SLWE is known to be the state-of-the art solution to tracking the properties of non-stationary environments, due to its multiplicative update form. Therefore, the SLWE is a better choice to accompany a stationary estimator than the far more common sliding window based approach. A theoretically well founded statistical testing procedure is developed to detect a significant difference between the stationary and dynamical estimators. Although our procedure bears similarities to the event detection procedure suggested by Ross et al. (Pattern Recogn Lett, 33(2):191-198, 2012), it is rather well founded theoretically. First, Ross et al. ignore the uncertainty in the stationary estimator in the detection procedure. Second, the detection threshold is determined based on heuristics and therefore lacks a solid statistical foundation. Extensive simulation results, based on both synthetic and real-life data related to news topic classification, demonstrate that our estimation procedure is easy to tune and outperforms legacy works. [ABSTRACT FROM AUTHOR]

Determining topological relations has proved to be one of the most important operations on spatiotemporal data, which still merits further attention. In this paper, we propose a valid and efficient topological relationship mechanism that allows identification of topological relations of uncertain spatiotemporal data over time. Our approach adopts polygon approximation and triangulation to represent uncertain spatiotemporal data. The unique feature is that our approach not only considers the polygon approximation of a spatiotemporal region but also takes number of the salient points into account. Moreover, determining topological relations of uncertain spatiotemporal data is detailed investigated based on counter-clock-wisely directed triangle. Finally, we apply our approach to meteorological events and experiments are run to validate our approach and show its performance advantages. [ABSTRACT FROM AUTHOR]

A Constraint Satisfaction Problem (CSP) is a very powerful framework for representing and solving constraint problems. Solving a CSP often requires searching for a solution in a large search space. Very often, much of the search efforts are wasted on the part of the search space that does not lead to a solution. Therefore, many search algorithms and heuristic techniques have been proposed to solve CSPs efficiently by guiding the search and reducing its size. Variable and value ordering techniques are among the most efficient ones as past experiments have shown that these heuristics can significantly improve the search performance and lead to the solution sooner. One such heuristic works by gathering information during search to guide subsequent decisions when selecting variables. More precisely, this heuristic gathers and records information about failures in the form of constraint weight during constraint propagation. In this paper, we propose a variant of this heuristic where the weight of a constraint is also based on the conflict and support counts, of each variable attached to this constraint, gathered during constraint propagation. We also propose a dynamic value ordering heuristic based on the support and conflict count information. Experiments have been conducted on random, quasi-random, pattern and real world instances. The test results show that the proposed variable ordering heuristic performs well in the cases of hard random and quasi-random instances. The test results also show that combining the proposed variable and value ordering heuristics can improve the performance significantly for some difficult problems. [ABSTRACT FROM AUTHOR]

The purpose based access control model has been proposed recently to restrict the access to the sensitive data which are out of control of their owner. This model can be enforced by ensuring that the user who wants to access the private data will respect the specific plan of tasks/actions that leads to achieving the intended objective to use these data. The Organization Based Access Control (OrBAC) model is suitable to integrate this principle, but in a dynamic environment such as the cloud computing, the authorization rules should be expressed in flexible way, and they may include optional tasks which can be skipped in some cases in order to adapt temporarily to the changes in the context. To meet these requirements, we propose in this paper a new extension of the OrBAC model using the temporal nonmonotonic description logic (TL-JClassicδ휖+) that allows to represent formally the policy rules as hierarchical planning that includes a set of ordered tasks that may admit exceptions in special cases and when the access request is made, the access control system depending on the current context will infer dynamically the appropriate sequence of actions that can be performed by subject who demands access to private data that may be outsourced into the cloud. [ABSTRACT FROM AUTHOR]

Human activity recognition (HAR) deals with recognition of activities or interactions that include humans within a video. Entities occurring in a video frame can be abstracted in variety of ways, ranging from the detailed silhouette of the entity to the very basic axis-aligned minimum bounding rectangles (MBR). On one end of the spectrum, using a detailed silhouette is not only demanding in terms of storage and computational resources but is also easily susceptible to noise. On the other end of the spectrum, MBRs require less storage, computation and abstracts out noise and video specific details. However, for abstraction of human bodies in a video an MBR does not offer an adequate solution because in addition to abstracting away noise, it also abstracts out important details such as the posture of the human body. For a more precise description, which offers a reasonable tradeoff between efficiency and noise elimination, a human body can be abstracted using a set of MBRs corresponding to different body parts. However, for a representation of activities as relations between interacting objects, a simplistic approximation assuming each MBR to be an independent entity leads to computation of redundant relations. In this paper, we explore a representation schema for interaction between entities that are considered as sets of rectangles, also referred to as extended objects. We further show that, given the representation schema, a simple recursive algorithm can be used to opportunistically extract topological, directional and distance information in O( n l o g n) time. We evaluate our representation schema for HAR on the Mind's Eye dataset (), the UT-Interaction (Ryoo and Aggarwal 2010) dataset and the SBU Kinect Interaction dataset (Yun et al. 2012). [ABSTRACT FROM AUTHOR]

With spatiotemporal applications increasing, a large amount of spatiotemporal data emerges. Because temporal and spatial attributes are often vague, research on fuzzy spatiotemporal data, especially querying fuzzy spatiotemporal data, has attracted a lot of attention. However, although fuzzy logic is incorporated in querying fuzzy spatiotemporal data and querying fuzzy data in XML, relatively little work has been carried out in querying fuzzy spatiotemporal data in XML. In this paper, we propose an algorithm, called FSPTwigFast, to match fuzzy spatiotemporal XML twig pattern. We represent fuzzy spatiotemporal data by adding temporal and spatial attributes associating with fuzziness in crisp data. We extend Dewey code to mark fuzzy spatiotemporal data for special process and determine structure relationship of fuzzy spatiotemporal nodes in XML documents. Our technique uses streams to store leaf nodes in XML document corresponding to leaf query nodes, which are filtered to delete unmatched nodes. After filtering, output lists are built for every matched leaf node. Finally, the experimental results demonstrate the performance advantages of our approach. [ABSTRACT FROM AUTHOR]

In this paper, we focus on how to handle disruptions during plan execution. Disruption are common in realistic plan execution, and the replanning is needed when the original plan is failed. Aiming to reduce the number of unnecessary replanning in this situation, a novel heuristic HTN (Hierarchical Task Network) planning approach is proposed. The approach includes two components: F-HTN and Controller. F-HTN is a heuristic temporal HTN planner. F-HTN uses STNs (Simple Temporal Networks) to express the complex temporal constraints in planning, and an STN-based heuristic search is designed to guide the search direction in F-HTN. When the planner F-HTN is terminated, it will generate a flexible plan which includes a complete plan and an STN associated with it. Then, Controller checks and adjusts the flexible plan when a disruption occurs during the plan execution. Integrating the plan generation and plan execution, this approach will reduce the number of unnecessary replanning. In the experimental study, we demonstrate the effectiveness and practicability of this approach through some emergency logistics problems. [ABSTRACT FROM AUTHOR]