Wyniki wyszukiwania - BazTech

1

Unsupervised dynamic topic model for extracting adverse drug reaction from health forums

Eslami Behnaz, Motlagh Mehdi Habibzadeh, Rezaei Zahra, Eslami Mohammad, Amini Mohammad Amin

Applied Computer Science

|

2020

|

Vol. 16, no 1

41--59

EN

The relationship between drug and its side effects has been outlined in two websites: Sider and WebMD. The aim of this study was to find the association between drug and its side effects. We compared the reports of typical users of a web site called: “Ask a patient” website with reported drug side effects in reference sites such as Sider and WebMD. In addition, the typical users’ comments on highly-commented drugs (Neurotic drugs, Anti-Pregnancy drugs and Gastrointestinal drugs) were analyzed, using deep learning method. To this end, typical users’ comments on drugs' side effects, during last decades, were collected from the website “Ask a patient”. Then, the data on drugs were classified based on deep learning model (HAN) and the drugs’ side effect. And the main topics of side effects for each group of drugs were identified and reported, through Sider and WebMD websites. Our model demonstrates its ability to accurately describe and label side effects in a temporal text corpus by a deep learning classifier which is shown to be an effective method to precisely discover the association between drugs and their side effects. Moreover, this model has the capability to immediately locate information in reference sites to recognize the side effect of new drugs, applicable for drug companies. This study suggests that the sensitivity of internet users and the diverse scientific findings are for the benefit of distinct detection of adverse effects of drugs, and deep learning would facilitate it.

2

Projective nonnegative matrix factorization based on α-divergence

Yang Z., Oja E.

Journal of Artificial Intelligence and Soft Computing Research

|

2011

|

Vol. 1, No. 1

7--16

EN

The well-known Nonnegative Matrix Factorization (NMF) method can be provided with more flexibility by generalizing the non-normalized Kullback-Leibler divergence to α- divergences. However, the resulting α-NMF method can only achieve mediocre sparsity for the factorizing matrices. We have earlier proposed a variant of NMF, called Projective NMF (PNMF) that has been shown to have superior sparsity over standard NMF. Here we propose to incorporate both merits of α-NMF and PNMF. Our α-PNMF method can produce a much sparser factorizing matrix, which is desired in many scenarios. Theoretically, we provide a rigorous convergence proof that the iterative updates of α-PNMF monotonically decrease the α-divergence between the input matrix and its approximate. Empirically, the advantages of α-PNMF are verified in two application scenarios: (1) it is able to learn highly sparse and localized part-based representations of facial images; (2) it outperforms α-NMF and PNMF for clustering in terms of higher purity and smaller entropy.

3

Testing dimension reduction methods for image retrieval

Moravec P., Snášel V.

Image Processing & Communications

|

2006

|

Vol. 11, no 2

51-61

EN

In this paper, we compare performance of several dimension reduction techniques, namely LSI, NMF, SDD and FastMap. The qualitative comparison is based on rank lists and evaluated on a collection of faces from the Olivetti Research Lab. We compare the quality of these methods from several standpoints: the visual impact, quality of generated "eigenfaces", size of reduced matrices and retrieval performance.