Semantic Paths Research Articles

TODEAR: Promoting explainable TKG reasoning through temporal offset enhanced dynamic embedding and adaptive reinforcement learning

Recently, Reinforcement Learning (RL) is utilized in Temporal Knowledge Graph (TKG) reasoning to generate analyzable reasoning paths, which achieves explainable reasoning on queries. The process of generating high-quality reasoning paths is facing two major challenges. The first is to construct an efficient entity embedding model from the complex temporal dependencies among entities. The second is to design a fine-grained reward mechanism for the reasoner based on the deep semantics of exploration paths. Motivated by the two challenges, a TKG reasoning framework based on Temporal Offset Enhanced Dynamic Embedding and Adaptive Reinforcement Learning (TODEAR) is proposed in this paper. Firstly, a temporal offset enhanced dynamic embedding model with a distance scoring function is designed to fully exploit the complex temporal dependencies among entities. In order to capture the evolutionary patterns of historical facts, it encodes both the relational structures of the TKG and the temporal offsets between events and queries. Then, a fine-grained adaptive reward mechanism is designed to optimize the reasoner. It generates real-time rewards by analyzing the logic and semantics of exploration paths to mitigate the adverse effects of sparse rewards. Extensive experiments on four benchmark datasets show that TODEAR significantly outperforms the state-of-the-art methods.

Dual-TranSpeckle: Dual-pathway transformer based encoder-decoder network for medical ultrasound image despeckling

The majority of existing deep learning-based image denoising algorithms mainly focus on processing the overall image features, ignoring the fine differences between the semantic and pixel features. Hence, we propose Dual-TranSpeckle (DTS), a medical ultrasound image despeckling network built on a dual-path Transformer. The DTS introduces two different paths, named “semantic path” and “pixel path,” to facilitate the parallel transfer of feature information within the image. The semantic path passes a global view of the input semantic features, and the image features are passed through a Semantic Block to extract global semantic information from pixel-level features. The pixel path is employed to transmit finer-grained pixel features. Within the dual-path network framework, two essential modules, namely Dual Block and Merge Block, are designed. These leverage the Transformer architecture during the encoding and decoding stages. The Dual Block module facilitates information interaction between the semantic and pixel features by considering the interdependencies across both paths. Meanwhile, the Merge Block module enables parallel transfer of feature information by merging the dual path features, thereby facilitating the self-attention calculations for the overall feature representation. Our DTS is extensively evaluated on two public datasets and one private dataset. The DTS network demonstrates significant enhancements in quantitative evaluation results in terms of peak signal-to-noise ratio (PSNR), structural similarity (SSIM), feature similarity (FSIM), and naturalness image quality evaluator (NIQE). Furthermore, our qualitative analysis confirms that the DTS has significant improvements in despeckling performance, effectively suppressing speckle noise while preserving essential image structures.

Dual-path segmentation network for automatic fabric defect detection

Fabric defect detection plays a crucial role in the production process of the textile industry. Vision-based inspection methods have emerged as an inevitable trend due to their lower labor costs and high detection efficiency. As the accuracy requirements for fabric defect detection, methods must not only identify and locate defects accurately but also describe the morphological features of the defects. This poses a challenge for the algorithm’s design, as it must consider both the semantic and texture information of the fabric. In this paper, we propose an end-to-end dual-path segmentation network called DPNet for fabric defect detection, which can extract and fuse both semantic and texture information to achieve high accuracy. The proposed framework consists of two paths: the semantic path, which has narrow but deep layers to obtain high-dimensional features, and the texture path, which has wide and dense layers to extract low-level details. To enhance the interaction between semantic and texture features, a crossed attention fusion module has been developed. Evaluations show that the proposed method outperforms other methods on different datasets in terms of mIoU, with results of 75.84% for Ngan and 70.69% for AITEX. In addition, we developed an inspection platform and tested the proposed method online. We found that it can achieve online detection at a speed of 40 m/min, making it well-suited for practical production environments.

KR4SL: knowledge graph reasoning for explainable prediction of synthetic lethality.

Synthetic lethality (SL) is a promising strategy for anticancer therapy, as inhibiting SL partners of genes with cancer-specific mutations can selectively kill the cancer cells without harming the normal cells. Wet-lab techniques for SL screening have issues like high cost and off-target effects. Computational methods can help address these issues. Previous machine learning methods leverage known SL pairs, and the use of knowledge graphs (KGs) can significantly enhance the prediction performance. However, the subgraph structures of KG have not been fully explored. Besides, most machine learning methods lack interpretability, which is an obstacle for wide applications of machine learning to SL identification. We present a model named KR4SL to predict SL partners for a given primary gene. It captures the structural semantics of a KG by efficiently constructing and learning from relational digraphs in the KG. To encode the semantic information of the relational digraphs, we fuse textual semantics of entities into propagated messages and enhance the sequential semantics of paths using a recurrent neural network. Moreover, we design an attentive aggregator to identify critical subgraph structures that contribute the most to the SL prediction as explanations. Extensive experiments under different settings show that KR4SL significantly outperforms all the baselines. The explanatory subgraphs for the predicted gene pairs can unveil prediction process and mechanisms underlying synthetic lethality. The improved predictive power and interpretability indicate that deep learning is practically useful for SL-based cancer drug target discovery. The source code is freely available at https://github.com/JieZheng-ShanghaiTech/KR4SL.

Assessing mechanisms behind crossmodal associations between visual textures and temperature concepts.

In the last decades, there has been a growing interest in crossmodal correspondences, including those involving temperature. However, only a few studies have explicitly examined the underlying mechanisms behind temperature-related correspondences. Here, we investigated the relative roles of an underlying affective mechanism and a semantic path (i.e., regarding the semantic knowledge related to a single common source identity or meaning) in crossmodal associations between visual textures and temperature concepts using an associative learning paradigm. Two online experiments using visual textures previously shown to be associated with low and high thermal effusivity (Experiment 1) and visual textures with no consensual associations with thermal effusivity (Experiment 2) were conducted. Participants completed a speeded categorization task before and after an associative learning task, in which they learned mappings between the visual textures and specific affective or semantic stimuli related to low and high temperatures. Across the two experiments, both the affective and semantic mappings influenced the categorization of visual textures with the hypothesized temperatures, but there was no influence on the reaction times. The effect of learning semantic mappings was larger than that of affective ones in both experiments, suggesting that a semantic path has more weight than an affective mechanism in the formation of the associations studied here. The associations studied here could be modified through associative learning establishing correlations between visual textures and either affective or semantic stimuli. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Diagnosis of atrial fibrillation based on lightweight detail-semantic network

Atrial fibrillation has become an important disease that threatens people's health. Wearable long-range dynamic ECG signals are effective means to screen for AF, but the complexity of existing algorithms limits the integration of signal acquisition and analysis in wearable ECG monitors. Developing an algorithm that satisfies the computational power of low-end chips is the key to the problem. In this paper, we propose a lightweight detail-semantic network (LDS Net) consisting of detail path, semantic path, and an improved cross-guidance mechanism. The detail path uses multiple layers of depthwise separable convolution to extract deep information about AF, the semantic path uses hourglass residual modules to compensate for the disadvantage of obstructed information flow of depthwise separable convolution, and an improved cross-guidance mechanism uses the idea of mutual guidance to achieve information fusion between the detail path and the semantic path. The network has only 0.2 M parametric count and 327.36 M computational volume. It achieves 99.57% accuracy on the MIT-BIH Atrial Fibrillation Database and 90.89% accuracy on the clinical dataset. The results show that LDS Net is significantly better than the classical convolutional neural network in terms of computational cost and accuracy, and has good clinical application prospects and generalizability.

Semantic Path Planning for Indoor Navigation Tasks Using Multi-View Context and Prior Knowledge

Semantic Path Planning for Indoor Navigation Tasks Using Multi-View Context and Prior Knowledge

Meta-path fusion based neural recommendation in heterogeneous information networks

As a powerful data modeling tool, Heterogeneous Information Network (HIN) has been successfully used in auxiliary information exploitation to boost recommendation performance. For HIN based recommendation, it is challenging to extract and fuse useful features of user preferences and item attributes under different semantic paths in HINs. Existing methods leverage a pre-defined fusion function to integrate different semantics for recommendation, which cannot characterize the complex nonlinear interactions between users and items. In this paper, we present a general framework named MNRec, short for Meta-path fusion based Neural Recommendation, to extract and fuse user and item embeddings under different meta-paths for recommendation. Under the framework, we propose an instantiation of MNRec with Multi-Layer Perceptron (MLP) structure. It consists of two major steps, i.e., meta-path based heterogeneous network embedding and deep learning based rating prediction. Concretely, appropriate meta-paths are first designed according to domain knowledge. Then the embeddings of users and items are obtained through a meta-path and commuting matrix based heterogeneous network embedding method. Finally, in light of the powerful nonlinear modeling capabilities of deep neural networks, the learned embeddings under different meta-paths are integrated into a two-pathway MLP structure for rating prediction. Experimental results on three real-world datasets demonstrate the superiority and effectiveness of MNRec compared with state-of-the-art baselines in rating prediction.

Knowledge reasoning with multiple relational paths

Knowledge reasoning technology can infer new knowledge based on the existing entity relationship information in the knowledge graph to complement the knowledge graph. In the existing knowledge reasoning methods based on multi-step relationship paths, the contributions of entities and relationships in the relationship paths are not distinguished, nor are multiple relationship paths comprehensively considered. To solve these problems, we propose a knowledge reasoning model based on Bi-directional Short-Term Memory and Attention Mechanism. Firstly, the neighborhood, entity semantic category and relationship path information are integrated in the knowledge graph, and the Neighborhood Semantic Path Network model (NSPN) is constructed to obtain a hybrid representation of the multi-step relationship path. Secondly, the loss function and training process of NSPN are introduced. Finally, comparative experiments are carried out on the public dataset, and the results show that our model is superior to other models in relation prediction tasks and link prediction tasks, improves the computational efficiency and data sparsity, and provides a new idea for knowledge reasoning methods.

Autonomous Navigation of a Forestry Robot Equipped with a Scanning Laser

This abstract is an overview of our research project entitled “Innovative Forest Plantation”, currently in progress. The aim of this project is to automate traditionally manual tasks for poplar plantations in the first years after planting, in particular mechanical weeding without the use of herbicides. The poplar forest is considered as a semi-structured environment where the dense canopy prevents the use of GPS signals and laser sensors are often preferred to localize the vehicle. In this paper, we focus on one of the main functionalities: autonomous navigation, which consists in detecting and locating trees to move safely in such complex environment. Autonomous navigation requires both a precise and robust mapping and localization solution. In this context, Simultaneous Localization and Mapping (SLAM) is very well-suited solution. The constructed map can be reliably used to plan semantic paths of the mobile robot in order treat specifically each tree. Simulations conducted on Gazebo and Robot Operation System (ROS) have proven that the robot could navigate autonomously in a poplar forest.

RecepNet: Network with Large Receptive Field for Real-Time Semantic Segmentation and Application for Blue-Green Algae

Most high-performance semantic segmentation networks are based on complicated deep convolutional neural networks, leading to severe latency in real-time detection. However, the state-of-the-art semantic segmentation networks with low complexity are still far from detecting objects accurately. In this paper, we propose a real-time semantic segmentation network, RecepNet, which balances accuracy and inference speed well. Our network adopts a bilateral architecture (including a detail path, a semantic path and a bilateral aggregation module). We devise a lightweight baseline network for the semantic path to gather rich semantic and spatial information. We also propose a detail stage pattern to store optimized high-resolution information after removing redundancy. Meanwhile, the effective feature-extraction structures are designed to reduce computational complexity. RecepNet achieves an accuracy of 78.65% mIoU (mean intersection over union) on the Cityscapes dataset in the multi-scale crop and flip evaluation. Its algorithm complexity is 52.12 GMACs (giga multiply–accumulate operations) and its inference speed on an RTX 3090 GPU is 50.12 fps. Moreover, we successfully applied RecepNet for blue-green algae real-time detection. We made and published a dataset consisting of aerial images of water surface with blue-green algae, on which RecepNet achieved 82.12% mIoU. To the best of our knowledge, our dataset is the world’s first public dataset of blue-green algae for semantic segmentation.

Semantic segmentation of buildings in remote sensing images based on dual-path network with rich-scale features

To solve the problems of low utilization of spatial features and incomplete contour segmentation in building semantic segmentation of remote sensing images, a building semantic segmentation method in remote sensing images based on dual-path with rich-scale features is proposed. In the shallow spatial path of proposed method, Res2Net module and inception module are used to extract shallow rich scale features to avoid improper use of shallow features affecting segmentation results. In the deep semantic path, ResNet50 combined with hybrid dilated convolution is used as the backbone network, and the obtained high-level semantic features are pooled by a spatial pyramid to capture the deeper multi-scale features. Finally, a new feature fusion module is designed to assign weights to feature maps of different levels extracted from two paths. Experimental results on WHU and Massachusetts building datasets show that the proposed method has higher building extraction accuracy and better generalization ability compared with other semantic segmentation networks.

Causality and the PA/SN distinction

Abstract We analyze two constructions expressing the negation of a causal relation in contemporary European Portuguese, nem por isso (lit. ‘not for this [reason]’) and não porque…mas porque (‘not because…but rather because’). We propose that the PA/SN distinction, established for adversative connectives (e.g., Spanish pero and German aber as PA, Spanish sino and German sondern as SN), can be extended to the domain of causality and provides insight on the properties of these two causal constructions. We show that their sensitivity to different types of conversational implicature as well as their interaction with negation are reminiscent of the distinction between restrictive and exclusive adversatives. Our investigation, which is the first semantic-pragmatic analysis of negated causality in European Portuguese, allows us to formulate historical predictions concerning the relation between types of negated causality and the possible development of adversative markers. Our findings provide synchronic evidence for a diachronic semantic path from negated causality to adversative meaning that has been posited for several Romance languages (e.g., Italian però, Spanish pero). Specifically, using contemporary corpus data, we show that the properties of nem por isso provide insight into the bridging contexts for semantic change from causal to adversative meanings. In addition, we identify a type of negated causality that fits within strategies of corrective contrast not exclusive to the domain of causal relations.

ADPNet: Attention based dual path network for lane detection

Recently, the task of lane detection has been greatly improved with the rapid development of deep learning and autonomous driving. However, there exist limitations like the challenging complex scenarios and real-time efficiency. In this paper, we present a novel Attention Based Dual Path Network (ADPNet) to handle the task of lane detection. The ADPNet treat the process of lane detection as a task of binary semantic segmentation, where the Detail Path is designed to capture detailed low-level information and the Semantic Path with dual attention module is designed to capture contextual high-level information. We use the Feature Aggregation Module to fuse the information of the two paths, followed by the process of lane fitting to get a parametric description of lanes. The proposed ADPNet achieves good trade-off between the accuracy and real-time efficiency on TuSimple and CULane, which are two popular lane detection benchmark datasets. The results demonstrate that our architecture outperforms the current state-of-the-art methods.

Learning attention-based representations from multiple patterns for relation prediction in knowledge graphs

Knowledge bases, and their representations in the form of knowledge graphs (KGs), are naturally incomplete. Since scientific and industrial applications have extensively adopted them, there is a high demand for solutions that complete their information. Several recent works tackle this challenge by learning embeddings for entities and relations, then employing them to predict new relations among the entities. Despite their aggrandizement, most of those methods focus only on the local neighbors of a relation to learn the embeddings. As a result, they may fail to capture the KGs’ context information by neglecting long-term dependencies and the propagation of entities’ semantics. In this manuscript, we propose ÆMP (Attention-based Embeddings from Multiple Patterns), a novel model for learning contextualized representations by: (i) acquiring entities context information through an attention-enhanced message-passing scheme, which captures the entities local semantics while focusing on different aspects of their neighborhood; and (ii) capturing the semantic context, by leveraging the paths and their relationships between entities. Our empirical findings draw insights into how attention mechanisms can improve entities’ context representation and how combining entities and semantic path contexts improves the general representation of entities and the relation predictions. Experimental results on several large and small knowledge graph benchmarks show that ÆMP either outperforms or competes with state-of-the-art relation prediction methods.

Enriching the learner’s model through the semantic analysis of learning traces

Our aim in this paper is to improve the efficiency of a learning process by using learners’ traces to detect particular needs. The analysis of the semantic path of a learner or group of learners during the learning process can allow detecting those students who are in needs of help as well as identify the insufficiently mastered concepts. We examine the possibility of using a student’s browsing path during a learning session, based on his navigation traces, to update the learner model. We assume that the domain concepts examined outside the learning platform but that are related to the course concepts are problematic to the learner. Knowing about these concepts may allow the course’s author to adapt the course to the learner’s needs regarding these concepts, as well as allow the tutor to help and assist the learner on these problematic concepts. We rely on Web data mining methods to filter, organize, and analyze the student’s browsing path. More precisely, we use a domain ontology of the course and the similarities that exist between external documents (visited pages) and the domain concepts (the course keywords). This analysis process makes it possible to detect students’ learning difficulties and to adapt the course based on the learner’s model.

Indoor Positioning and Navigation Model Based on Semantic Grid

Traditional GPS positioning technology cannot be used in indoor space. With the development of the new positioning technology and the Internet of things, the indoor mobile object positioning and navigation model have been the focus of the relevant research institutions at home and abroad. Based on this, indoor positioning technology was studied starting from Wi-Fi, RFID, and iBeacon technology in this paper. However, the accuracy of indoor positioning and navigation needs to be further improved. This paper presents a semantic space model based on artificial intelligence technology, through semantic pattern matching, semantic concept extension, semantic reasoning and semantic mapping, and interior semantic localization is realized. The indoor semantic network and indoor grid navigation model are constructed, and the indoor semantic path is modeled from time, location, user, and congestion. At the same time, the improved Term Frequency-Inverse Document Frequency is combined with the Hidden Markov Model to improve the accuracy of matching the stay area with the most likely location to visit and improve the accuracy of semantic annotation. It was found that the research on the indoor positioning and navigation model based on the semantic grid can realize the uniform expression of the complex spatial semantics of the theme, geometry, connectivity, and distance, which can promote the development of indoor positioning and navigation.

AN ALGORITHM FOR CONSTRUCTING ASSOCIATIVE SERIES OF HASHTAGS FOR SEMANTIC NAVIGATION IN SOCIAL NETWORKS

Nowadays hashtags are an important mechanism of semantic navigation in social media. In this study, we consider the solution of the problem of building associative series of hashtags for one of the largest social networks. These series should meet two criteria: they should be short and shouldn’t have wide semantic gaps between sequential hashtags. An algorithm that allows us to create an associative series of hashtags could be used to increase the quantity of hashtags in posts, which will facilitate semantic navigation through posts in a social network. The paper proposes a formal definition of the semantic path building problem as a multicriteria optimization problem on the co-occurrence network of hashtags in posts. First, we built a co-occurrence network for hashtags from a big dataset of messages from Instagram. Then, we develop a combined optimization function for both criteria from the semantic path building problem. For measuring semantic similarity between hashtags, we use a metric based on the word2vec embeddings of hashtags. Using empirical paths obtained with various algorithms, we tune the parameters of a generalized optimization function that can be used to construct semantic paths using Dijkstra’s pathfinding or special greedy algorithms.

Joint semantics and data-driven path representation for knowledge graph reasoning

Reasoning on a large-scale knowledge graph (KG) is of great importance for KG applications like question answering. The path-based reasoning models can leverage much information over paths other than pure triples in the KG but face several challenges. Firstly, all the existing path-based methods are data-driven, lacking explainability, namely how the path representations and the reasoning results are obtained with human-understandable explanations. Besides, some approaches either consider only relational paths or ignore the heterogeneity between entities and relations both in paths, which cannot capture the rich semantics of paths well. To address the above challenges, in this work, we propose a novel joint semantics and data-driven path representation that balances explainability and generalization in the framework of KG embedding. Specifically, we inject horn rules to obtain the condensed paths through a transparent and explainable path composition procedure. The entity converter is designed to transform entities along paths into the representations in the semantic level similar to relations for reducing the heterogeneity between entities and relations. The KGs, both with and without type information, are considered. Our proposed model is evaluated on two classes of tasks: link prediction and path query answering. The experimental results show that our model obtains significant performance gains over several state-of-the-art baselines.

TEIL-GANZES-RELATIONEN IM KONTRASTIVEN VERGLEICH (AM BEISPIEL DER KÖRPERTEILBENENNUNGEN IN DEN GERMANISCHEN UND SLAWISCHEN SPRACHEN)

This paper deals with the research of the Part-Whole-Relations in the lexical system of the language exemplified by the human body part names in the Germanic (German, English) and Slavonic (Ukrainian, Russian) languages. The contrastive approach reveals the main differences in the nomination and segmentation process of the lexical units under analysis. The system of naive segmentation of the human body is proved to be the rich source of anthropomorphic metaphor in the compared languages. The polysemy level of the body part names in the contrasted languages is distinguished with the help of the index of polysemy. The study focuses on the classification of the regular semantic change paths of the body part names, highlights cross-language similarities and differences in their semantics and outlines their semantic productivity in the field of terminology.