Major depressive disorder (MDD) is one of the most common and severe mental illnesses, posing a huge burden on society and families. Recently, some multimodal methods have been proposed to learn a multimodal embedding...
详细信息
Major depressive disorder (MDD) is one of the most common and severe mental illnesses, posing a huge burden on society and families. Recently, some multimodal methods have been proposed to learn a multimodal embedding for MDD detection and achieved promising performance. However, these methods ignore the heterogeneity/homogeneity among various modalities. Besides, earlier attempts ignore interclass separability and intraclass compactness. Inspired by the above observations, we propose a graph neural network (GNN)-based multimodal fusion strategy named modal-shared modal-specific GNN, which investigates the heterogeneity/homogeneity among various psychophysiological modalities as well as explores the potential relationship between subjects. Specifically, we develop a modal-shared and modal-specific GNN architecture to extract the inter/intramodal characteristics. Furthermore, a reconstruction network is employed to ensure fidelity within the individual modality. Moreover, we impose an attention mechanism on various embeddings to obtain a multimodal compact representation for the subsequent MDD detection task. We conduct extensive experiments on two public depression datasets and the favorable results demonstrate the effectiveness of the proposed algorithm.
With the development of the architectures and the growth of AIoT application requirements, data processing on edge has become popular. Neural network inference is widely employed for data analytics on edge devices. Th...
With the development of the architectures and the growth of AIoT application requirements, data processing on edge has become popular. Neural network inference is widely employed for data analytics on edge devices. This paper extensively explores neural network inference on integrated edge devices and proposes EdgeNN, the first neural network inference solution on CPU-GPU integrated edge devices. EdgeNN has three novel characteristics. First, EdgeNN can adaptively utilize the unified physical memory and conduct the zero-copy optimization. Second, EdgeNN involves a novel inference-targeted inter- and intra-kernel CPU-GPU hybrid execution approach, which co-runs the CPU with the GPU to fully utilize the edge device’s computing resources. Third, EdgeNN adopts a fine-grained adaptive inference tuning approach, which can divide the complicated inference structure into sub-tasks mapped to the CPU and the GPU. Experiments show that on six popular neural network inference tasks, EdgeNN brings an average of 3.97×, 3.12×, and 8.80× speedups to inference on the CPU of the integrated device, inference on a mobile phone CPU, and inference on an edge CPU device. Additionally, it achieves 22.02% time benefits to the direct execution of the original programs. Specifically, 9.93% comes from better utilization of unified memory, and 10.76% comes from CPU-GPU hybrid execution. Besides, EdgeNN can deliver 29.14× and 5.70× higher energy efficiency than the edge CPU and the discrete GPU, respectively. We have made EdgeNN available at https://***/ChenyangZhang-cs/EdgeNN.
To optimize the term hierarchy in the manual e-government thesaurus, we combine the mainstream knowledge organization technology to form a complete set of ontology automation construction scheme. We build an e-governm...
详细信息
To optimize the term hierarchy in the manual e-government thesaurus, we combine the mainstream knowledge organization technology to form a complete set of ontology automation construction scheme. We build an e-government knowledge base by using subject words in the Comprehensive E-government Thesaurus as the term set and encyclopedia text as the corpus. The specific work includes the extraction of semantic features from the bag-of-words model, determination of the number of clusters by linear and nonlinear dimensionality reduction, division of terms by spectral clustering, social network analysis to determine the class label, and storing knowledge ontology via OWL. The recall rate of term hierarchy in the ontology is excellent, indicating the ontology has good knowledge extensibility, and also proving the efficiency of the scheme proposed in this work. Besides, the application model of a term hierarchy in information retrieval can show a richer semantic relation than the original thesaurus to guide the retrieval extension of government information resources. 83rd Annual Meeting of the Association for Information Science & Technology October 25-29, 2020. Author(s) retain copyright, but ASIS&T receives an exclusive publication license.
The Research on Balancing Theories and Mechanisms to Pervasive Information Systems (B2P) is one of the component studies of the Records-centered Digital Information Management Theory and Mechanisms (DI{R}Mtm) Project,...
详细信息
The field of Information Systems (ISs) has long been recognized, so has Enterprise Information Systems (EISs), a field close to it. Long existing also in organizations or enterprises is the field of records management...
详细信息
The number of word embedding models is growing every year. Most of them are based on the co-occurrence information of words and their contexts. However, it is still an open question what is the best definition of cont...
详细信息
作者:
Li YuSchool of Information
Key Laboratory of Data Engineering and Knowledge Engineerin Renmin University of China Beijing China
Collaborative filtering is an important personalized recommendation technique applied widely in E-commerce. It is not adapted to multi-interest or title recommendation for the 'general neighbourhood' problem w...
详细信息
ISBN:
(纸本)9781424432004;9780769531854
Collaborative filtering is an important personalized recommendation technique applied widely in E-commerce. It is not adapted to multi-interest or title recommendation for the 'general neighbourhood' problem which is analyzed in this paper. Based on it, collaborative filtering recommendation based on community is presented by introducing the concept 'community neighbourhood' in the paper. Unfortunately, it results into severer sparsity problem which makes heavy effect on its performance. In order to overcome it, an ontological A-priori score is used to infer user preference and to pre-fill null rating first. After pre-filling using the ontology method, then collaborative filtering based on community is executed based on a dense rating matrix. The experiment shows that collaborative filtering based on community makes generally better performance than traditional method when data is not very sparse, and ontology method can truly enhance collaborative filtering based on community since the sparsity is overcame.
This position paper, via rationalizing the potentials of carrying out a carefully crafted research project, argues that the three fields of enterprise information system, business process management, and digital recor...
详细信息
In this paper, we present an effective method to craft text adversarial samples, revealing one important yet underestimated fact that DNN-based text classifiers are also prone to adversarial sample attack. Specificall...
详细信息
暂无评论