ISSN: 1230-2813
Advances in Psychiatry and Neurology/Postępy Psychiatrii i Neurologii
Current issue Archive Manuscripts accepted About the journal Editorial board Abstracting and indexing Subscription Contact Instructions for authors Ethical standards and procedures
SCImago Journal & Country Rank
2/2020
vol. 29
 
Share:
Share:
more
 
 
abstract:
Original article

Binary classification of pornographic and non-pornographic materials using the sAI 0.4 model and the modified sexACT database

Wojciech Oronowicz-Jaśkowiak
1
,
Edyta Bzikowska
2
,
Klaudia Jabłońska
3
,
Anita Kłok
3

1.
3rd Department of Psychiatry, Institute of Psychiatry and Neurology, Warsaw, Poland
2.
Faculty of Educational Sciences, University of Lodz, Poland
3.
Faculty of Social Sciences, Catholic University of Lublin, Poland
Adv Psychiatry Neurol 2020; 29 (2): 108-119
Online publish date: 2020/07/22
View full text
Get citation
ENW
EndNote
BIB
JabRef, Mendeley
RIS
Papers, Reference Manager, RefWorks, Zotero
AMA
APA
Chicago
Harvard
MLA
Vancouver
 
Introduction
Neural networks may be used to solve problems in the field of psychology and sexology. In particular, it seems that neural networks may be important to limit the unintentional contact of minors with pornographic material. The aim of the study was to create the neural networks model for the classification of pornographic (also fetishist) materials from non-pornographic materials.

Material and methods
In order to create a new model, the sAI 0.3 model was used as the basic model. The fast.ai library version 1.0.55 was used. A modified version of ResNet152 was adopted as the neural network architecture. The sexACT database was modified to include new training material – 1630 non-pornographic photos of women. A total of 1304 photos (80% of the set) were used to train the network and the remaining 326 photos (20% of the set) were used for its later validation.

Results
As a result of the research, the sAI 0.4 model was created, enabling binary classification of pornographic and non-pornographic materials with 96% accuracy. The model tends to make more the first type of error than the second type of errors. The model has a high precision (0.94) and high sensitivity (0.88). The final validation loss was 0.1314.

Conclusions
The potential benefits of using the discussed model from a clinical perspective were discussed. The application of the discussed model could prevent the negative effects of contact of minors with pornographic material, which could consequently limit the prevalence of risky sexual behavior or negative psychosocial effects.

keywords:

neural networks, clinical sexology, pornographic materials

Quick links
© 2020 Termedia Sp. z o.o. All rights reserved.
Developed by Bentus.
PayU - płatności internetowe