Few-shot knowledge graph completion (FKGC), which aims to infer missing facts about a relation from only a few reference triples, has recently attracted great attention. The core of solving the FKGC task is to learn a...
Few-shot knowledge graph completion (FKGC), which aims to infer missing facts about a relation from only a few reference triples, has recently attracted great attention. The core of solving the FKGC task is to learn a vector representation for each few-shot relation using the corresponding entity represen-tations. To this end, existing models generally enhance entity representations with their direct neighbors. However, a large number of entities have few direct neighbors. Hence, encoding only direct neighborhood is insufficient to obtain satisfactory en-tity representations. In addition, current models typically utilize static embeddings to represent entities, ignoring their diverse semantics, i.e., an entity may show distinct semantics within different few-shot relations. To address these issues, we propose a new FKGC framework, namely TransD-based Multi-hop Meta Learning (TDML). TDML consists of three main components: a multi-hop neighbor encoder to enhance entity representations by aggregating heterogeneous multi-hop neighbors, a transformer encoder to generate the relation meta representations, and a TransD-based relation representation updater that allows each entity to exhibit relation-specific semantics and tune the relation meta representations. Extensive experiments on two public datasets demonstrate that our model outperforms state-of-the-art FKGC methods.
When acquiring labels from crowdsourcing platforms, a task may be designed to include multiple labels and the values of each label may belong to a set of various distinct options, which is the so-called multi-class mu...
详细信息
The remaining useful life (RUL) of bearings is critical to the proper operation of mechanical equipment, maintenance of equipment costs and availability. The existing domain adaptation methods have had great success i...
详细信息
The remaining useful life (RUL) of bearings is critical to the proper operation of mechanical equipment, maintenance of equipment costs and availability. The existing domain adaptation methods have had great success in RUL prediction. However, when the target bearing data are unavailable or unknown to be involved in model training, the domain adaptation approaches also incapable. To solve the problem, we propose a parallel reversible instance normalization method based on adaptive threshold stage division for remaining useful life prediction of unknown bearings. First, we design an adaptive threshold method to find degradation points to divide the healthy and degradation stages. Then according to time series, we merge the original vibration data and its instance normalized data to increase the data distribution diversity. Finally, we combine instance normalization and parallel reversible normalization of the source bearing data into unified RUL learning framework to solve the uncertainty of counterfactual data and improve RUL prediction performance. The results show that the method is superior to the state-of-the-art methods for RUL prediction of unknown bearings.
Developers’ API needs should be more pragmatic, such as seeking suggestive, explainable, and extensible APIs rather than the so-called best result. Existing API search research cannot meet these pragmatic needs becau...
详细信息
In multi-label learning, each instance is associated with a set of labels simultaneously. Most existing studies assume that the set of labels for each instance is complete. However, it is generally difficult to obtain...
In multi-label learning, each instance is associated with a set of labels simultaneously. Most existing studies assume that the set of labels for each instance is complete. However, it is generally difficult to obtain all the relevant labels of each instance, and only a partial or even empty set of relevant labels is available, which is called semi-supervised multi-label learning with missing labels. To tackle this problem, we propose a novel framework that considers label correlations and instance correlations to recover the missing labels and utilizes a large amount of unlabeled data simultaneously to improve the classification performance. Specifically, a new supplementary label matrix is firstly obtained by learning the label correlation. Secondly, considering each class label may be decided by some specific characteristics of its own, a label-specific data representation is hence learned for each class label. Thirdly, instance correlations are utilized not only to recover the missing labels, but also to propagate the supervision information from labeled instances to unlabeled ones. In addition, a united objective function is designed to facilitate the above processing and an accelerated proximal gradient method is adopted to solve the optimization problem. Finally, extensive experimental results conducted on several benchmark datasets demonstrate the effectiveness of the proposed method compared to competing ones.
Domain adaptation (DA) -based RUL prediction methods have achieved great success for the adaptation ability of the distribution discrepancy between the source and target domains. However, DA methods are powerless when...
详细信息
Domain adaptation (DA) -based RUL prediction methods have achieved great success for the adaptation ability of the distribution discrepancy between the source and target domains. However, DA methods are powerless when the target domain data are not available for training. To solve this problem, we propose an inter-domain intra-domain normalized generalization (IDIDNG) network, which consists of three modules, respectively, the pre-processing module, the feature transformation module, and the RUL prediction module. First, we design the pre-processing module to process the bearing vibration data with peak-to-peak and Z-score. Finally, it is connected into a four-dimensional array. In the feature transformation module, via intra-domain and inter-domain normalization as well as mean-variance cross-swap, we transform the data distribution expressions of invariant features of bearings from the perspective of different bearings and different degradation stage of one bearing, such that the model enables to learn the intra-domain and inter-domain discrepancy. Further we design four adaptable weighting parameters into the intra-domain normalization to learn the appropriate normalized mean and variance via the model training. Finally, we design the GRU-based RUL prediction module to predict the unknown bearings. We conducted experiments under the PHM2012 dataset, experimental results show that our method achieves satisfactory prediction accuracy in the unknown bearings.
Temporal information is pervasive and crucial in medical records and other clinical text,as it formulates the development process of medical conditions and is vital for clinical decision ***,providing a holistic knowl...
详细信息
Temporal information is pervasive and crucial in medical records and other clinical text,as it formulates the development process of medical conditions and is vital for clinical decision ***,providing a holistic knowledge representation and reasoning framework for various time expressions in the clinical text is *** order to capture complex temporal semantics in clinical text,we propose a novel Clinical Time Ontology(CTO)as an extension from OWL *** specifically,we identified eight timerelated problems in clinical text and created 11 core temporal classes to conceptualize the fuzzy time,cyclic time,irregular time,negations and other complex aspects of clinical ***,we extended Allen’s and TEO’s temporal relations and defined the relation concept description between complex and simple ***,we provided a formulaic and graphical presentation of complex time and complex time *** carried out empirical study on the expressiveness and usability of CTO using real-world healthcare ***,experiment results demonstrate that CTO could faithfully represent and reason over 93%of the temporal expressions,and it can cover a wider range of time-related classes in clinical domain.
Multi-label classification deals with the problem where an instance is associated with multiple labels simultaneously. Most existing multi-label classification algorithms assume that the labels of the training data ar...
Multi-label classification deals with the problem where an instance is associated with multiple labels simultaneously. Most existing multi-label classification algorithms assume that the labels of the training data are complete. However, we can obtain only a partial label set of each instance in some real applications since labelling data is difficult or costly. Some existing works on multi-label classification with missing labels focus on exploiting label correlations to complete the original label space and simultaneously build a multi-label learning model using label specific features. However, these methods may be suboptimal since they do not preserve feature-label space consistency. In this paper, we propose a Space Consistency-based Multi-Label classification algorithm named SCML to address this issue. First, label correlation in label space is learned to augment the incomplete original label matrix to a new supplementary label matrix, and the multi-label classifier is constructed simultaneously based on the new supplementary label matrix. Then, correlation information in feature space is learned based on the probabilistic neighborhood similarities to preserve feature-label space consistency. Moreover, the proposed algorithm has an effective mechanism for learning label-specific features to improve the multi-label classification with missing labels. Extensive experiments on twelve benchmark data sets validate the effectiveness of the proposed approach for improving the generalization performance of state-of-the-art algorithms of multi-label learning with missing labels.
Deep learning methods have shown significant performance in medical image analysis tasks. However, they generally act like 'black box' without explanations in both feature extraction and decision processes, le...
详细信息
One of the recent best attempts at Text-to-SQL is the pre-trained language model. Due to the structural property of the SQL queries, the seq2seq model takes the responsibility of parsing both the schema items (i.e., t...
详细信息
暂无评论