Outlier detection is a crucial part of robust evaluation for crowd-sourceable assessment of Quality of Experience (QoE) and has attracted much attention in recent years. In this paper, we propose some simple and fast ...
详细信息
In web topic detection, detecting “hot” topics from enormous User-Generated Content (UGC) on web data poses two main difficulties that conventional approaches can barely handle: 1) poor feature representations from ...
详细信息
ISBN:
(纸本)9781467372596
In web topic detection, detecting “hot” topics from enormous User-Generated Content (UGC) on web data poses two main difficulties that conventional approaches can barely handle: 1) poor feature representations from noisy images and short texts; and 2) uncertain roles of modalities where visual content is either highly or weakly relevant to textual cues due to less-constrained data. In this paper, following the detection by ranking approach, we address the problem by learning a robust shared representation from multiple, noisy and complementary features, and integrating both textual and visual graphs into a k-Nearest Neighbor Similarity Graph (k-N 2 SG). Then Non-negative Matrix Factorization using Random walk (NMFR) is introduced to generate topic candidates. An efficient fusion of multiple graphs is then done by a Latent Poisson Deconvolution (LPD) which consists of a poisson deconvolution with sparse basis similarities for each edge. Experiments show significantly improved accuracy of the proposed approach in comparison with the state-of-the-art methods on two public data sets.
To deal with the challenge of information overload, in this paper, we propose a financial news recommendation algorithm which help users find the articles that are interesting to read. To settle the ambiguity problem,...
详细信息
To deal with the challenge of information overload, in this paper, we propose a financial news recommendation algorithm which help users find the articles that are interesting to read. To settle the ambiguity problem, a new presented OF-IDF method is employed to represent the unstructured text data in the form of key concepts, synonyms and synsets which are all stored in the domain ontology. For users, the recommendation algorithm build the profiles based on their behaviors to detect the genuine interests and predict current interests automatically and in real time by applying the thinking of relevance feedback. Finally, the experiment conducted on a financial news dataset demonstrates that the proposed algorithm significantly outperforms the performance of a traditional recommender.
In order to further improve the velocity and the utilization of information contained in samples,an improved version of the Factor Analysis Algorithm(FAA) in factor spaces is presented in this *** primary algorithm is...
详细信息
In order to further improve the velocity and the utilization of information contained in samples,an improved version of the Factor Analysis Algorithm(FAA) in factor spaces is presented in this *** primary algorithm is considered from the whole classes during the selection of the next classified factor,by which a smaller decision domain is generated than that generated by considering from each class,and the deletion of the decision domain is critical in decreasing calculation and increasing the velocity of the ***,based on inheriting the merits of the initial algorithm,the pushing way by each column is changed into that by each class during the selection of the next classified *** change not only decreases the calculation,but also improves the utilization of the sample *** testing results also indicate that the improvement is significant and the testing accuracy rate and velocity are both better than the primary algorithm.
暂无评论