Facial reconstruction

Search LJMU Research Online

Browse Repository | Browse E-Theses

Application of deep learning for detection of toxic images in social media

Mac Dermott, AM, Motylinski, M, Iqbal, F, Stamp, K, Hussain, M and Marrington, A Application of deep learning for detection of toxic images in social media. In: DFRWS APAC 2022, Adelaide, Australia. (Accepted)

[img] Text
DFRWS_APAC_2022_paper_4.pdf - Accepted Version
Restricted to Repository staff only

Download (680kB)


Detecting criminal activity online is not a new concept but how it can occur is changing. Technology and the data generated by the influx of new devices has a vital part to play in this changing landscape. For example, using social media information such as content, location data, and activity
are examples of information that can be used to detect criminal activity and, in some studies, can even be used to forecast and predict future occurrences. This paper explores the application of deep learning for the detection of embedded toxic text content in images shared on social media platforms. In this work, we propose a machine learning solution for the detection of toxic images based on embedded text content. The project utilizes GloVe word embeddings for data augmentation for improved prediction capabilities. Our methodology details the implementation of Long Short-term memory Gated recurrent unit models and their Bidirectional variants, comparing our approach to related works, and highlighting evident improvements. The best performing model, Bidirectional LSTM, achieved 0.92 testing accuracy and 0.88 inference accuracy with 0.92 and 0.88 F1-score accordingly.

Item Type: Conference or Workshop Item (Paper)
Subjects: Q Science > QA Mathematics > QA76 Computer software
Divisions: Computer Science & Mathematics
SWORD Depositor: A Symplectic
Date Deposited: 30 Jun 2022 12:36
Last Modified: 30 Jun 2022 12:36
URI: https://researchonline.ljmu.ac.uk/id/eprint/17187

Actions (login required)

View Item View Item