#TWIML Fest https://twimlai.com/twimlfest/
#Show and Tell! ML/AI Project Lightning Talks https://twimlai.com/twimlfest/sessions/show-and-tell/
Abdul Samadh
GitHub: https://github.com/samadh192
LinkedIn: https://www.linkedin.com/in/azathabdulsamadh/
Madhusoodhana Chari
GitHub: https://www.linkedin.com/in/madhucharis/
Deep Learning is being actively researched for Encrypted Network Behaviour Modelling for netwrok traffic strem classififcation for continuouis real time monitoring. The applciations of which can be pseudo visibility in the type of traffic ( audio, video, browsing), change detection, traffic shaping, sd-wan and cyber security.
In this work, basedon the statistical network behaviour, first it compares, classical neural models and deep neural model using ResNet architecture and highlights the learning advantage. Secondly, it using the per class, per feature Heuristics to give a confidence score. The aim of this work is to give a scalable solution for modelling networking traffic.
Explore the whole repository here.
- Clarifies that Deep Learning Models have future scalability with more classes and also it can learn using more deepler layer's.
- Classifies using Deep Learning ( ResNet) and gives Probability Scoring.
- Highlights the value in Statistics and Heuristics from data.
NumPy, Colab, SciPy, Pandas, Tensorflow, matplotlib
- Addition and Removal of more classes and resizing of ResNet Layer's
- Agument Deep Learnign with Classic ML using same Freuency