The ‘Spam email categorization using natural language processing and attention-embedded deep learning’ project is a three-way partnership between Federation University, La Trobe University and Westpac.
Federation University is responsible for coordinating and delivering the project using its in-house capabilities in AI, natural language processing, cyber security, knowledge, and experience of banking industry cyber threats to deliver tools to triage cyberattacks.
Westpac Bank is providing datasets and the know-how of the operating environment of the developed tools and Latrobe University is lending its expertise in determining the signatures of the cyber-attacks in a simulated sandboxed environment.
This proposal aims at tackling the spam email problem from a multidimensional perspective. The objective of the proposal is to classify emails into scams, malware, spam, phishing, and other categories.
In this research, we will propose and evaluate a novel method consisting of feature extraction and innovative machine learning technique to classify spam emails.
The feature extraction and selection technique will include natural language processing techniques to reveal the distinct inherent structural characteristics that are commonly present in the emails of various categories.
Machine learning techniques will explore and enhance the capability of the newly proposed neural networks based on attention mechanism.
The outcome of this project will provide customers with increased protection from financial fraud resulting from email-based malware as well as aid the banking industry to strengthen their security further and gain enhanced customer trust by stopping such attacks.