A Study to Learn Robust and Discriminative Representation to Tackle Cyber bullying Detection

Abstract

We build up another content portrayal display in view of a variation of SDA: marginalized stacked denoising autoencoders (mSDA), which receives straight rather than nonlinear projection to quicken preparing and minimizes endless clamor dispersion so as to take in more strong portrayals. We use semantic data to grow mSDA and create Semanticimproved Marginalized Stacked Denoising Autoencoders (smSDA). The semantic data comprises of harassing words. A programmed extraction of tormenting words in view of word embeddings is proposed so that the included human work can be diminished. Amid preparing of smSDA, we endeavor to reproduce bullying highlights from other typical words by finding the idle structure, i.e. connection, amongst tormenting and typical words. The instinct behind this thought is that some harassing messages don't contain bullying words

Authors and Affiliations

Diddla Mario Praneeth| Dept. of CSE, Nova college of institute and Technology,Eluru, G. P Madhuri| Dept. of CSE, Nova college of institute and Technology,Eluru

Keywords

Related Articles

Mobile Ad Hoc Networks Authentication Using Npv Method

NPV (Neighbour Position Verification) method is companionable with security architectures counting the ones that have been projected for vehicular networks which symbolize a likely deployment environment for NPV. Suc...

The Need Of Financial Statement Analysis In A Firm/An Orgnization

Financial statement analysis play a dominate role in setting the frame watt of managerial decisions through analysis and interpretation of financial statement. This paper discusses about financial „ strength and we...

Optimal Clustering Approach Used For Concurrent Probabilistic Graphs

To tackle the testing issue two calculations, in particular the PEEDR and the CPGS bunching calculation are characterized for each of the proposed calculations, and after that likewise characterize in the ballpark of...

A Secure Off-Line MICO Payment Approach Using Multiple Physical Unclonable Functions

FRoDO, a protected off-line micro-payment approach utilizing various physical unclonable capacities. FRoDO highlights an identity component to verify the client, and a coin component where coins are not locally store...

Synchronization of Grid Connected PV System Using PI Controller

Grid connected solar system consists of converters Circuits are organized in two different ways one is DC/DC boosters and another is PWM Inverter. This combination results decrement of Quality and efficiency of elect...

Download PDF file
  • EP ID EP16965
  • DOI -
  • Views 312
  • Downloads 5

How To Cite

Diddla Mario Praneeth, G. P Madhuri (2017). A Study to Learn Robust and Discriminative Representation to Tackle Cyber bullying Detection. International Journal of Science Engineering and Advance Technology, 5(6), 643-646. https://europub.co.uk/articles/-A-16965