By Mikhail Yu. Khachay, Natalia Konstantinova, Alexander Panchenko, Dmitry Ignatov, Valeri G. Labunets
This ebook constitutes the court cases of the Fourth foreign convention on research of pictures, Social Networks and Texts, AIST 2015, held in Yekaterinburg, Russia, in April 2015.
The 24 complete and eight brief papers have been conscientiously reviewed and chosen from one hundred forty submissions. The papers are equipped in topical sections on research of pictures and video clips; trend reputation and computing device studying; social community research; textual content mining and typical language processing.
Read Online or Download Analysis of Images, Social Networks and Texts: 4th International Conference, AIST 2015, Yekaterinburg, Russia, April 9–11, 2015, Revised Selected Papers PDF
Similar analysis books
- Elementary Classical Analysis
- Harmonic analysis of functions of several complex variables in the classical domains.
- Set Theoretical Aspects of Real Analysis
- Topics in Analysis: Colloquium on Mathematical Analysis Jyväskylä 1970
- Rapid Methods for Analysis of Biological Materials in the Environment
- A posteriori error analysis via duality theory : with applications in modeling and numerical approximations
Extra resources for Analysis of Images, Social Networks and Texts: 4th International Conference, AIST 2015, Yekaterinburg, Russia, April 9–11, 2015, Revised Selected Papers
Example text
In what follows, we refer to the target class of users who left a comment as “depressed”, and their opposite as the “random” class. The class distribution 28 A. Semenov et al. Table 1. Variable descriptive statistics summary. 122 in our dataset was 32 % depressed users and 68 % random. This asymmetry may be caused by having depressed users being potentially more stealthy or shy and hiding their profile data, whereas their ego network data is open to anyone regardless of privacy policies. Descriptive statistical summary of the obtained dataset is presented in Table 1.
Paragios, N. ) ECCV 2010, Part VI. LNCS, vol. 6316, pp. 57–70. Springer, Heidelberg (2010) 8. : Representing shape with a spatial pyramid kernel. In: 6th ACM International Conference on Image and Video Retrieval CIVR 2007, pp. 401–408 (2007) 9. : Image recognition based on wavelet transform and artificial neural networks. In: IEEE International Conference on Machine Learning and Cybernetics, pp. 789–793 (2008) 10. : Multi-column deep neural network for traffic sign classification. Neural Netw. 32, 333–338 (2012) 11.
The transformed feature distributions are shown on Fig. 3. 1 Analysis Framework Following the design of [9], we used binary logistic regression to identify the features of profiles that are able to predict a user’s depression propensity. We choose the Area Under the Curve (AUC) criterion as our measure of goodness of fit [8]; in particular, this measure is robust to asymmetric class distributions. Discerning Depression Propensity Among Participants of Suicide 31 In order to compare classification performance of different models, we utilized bootstrap sampling.