AI-based biometric sentimental analysis protocols for ATM fraud detection

Topic Description

ATM and Cash machine frauds are on the rise, and people are mugged, or their sensitive information is used to draw cash or extract banking details. This project is to look into various sentimental analysis approaches while using the cash machine’s built-in camera. The project is going to look into various theft incidents that happen and look into various sentimental analysis techniques used to detect facial gestures and emotions. An AI-based sentimental analysis framework is going to be developed to detect possible fraud during cash machine use. 

Skills Required:

• A first degree (at least a 2:1) ideally in Computer Science or Cyber Security with a good fundamental knowledge of Cryptography and Machine Learning.
• Experience of fundamental aspects of computer science, cryptography and machine learning
• Competence in mathematical computations and authentication protocols

Desirable attributes:  

• Knowledge and understanding of using tools, such as ProVerif, Tamarin Prover and Scyther
• Knowledge and understanding of using simulation tools and techniques, such as Matlab.

Background Reading:

The project will initially rely on existing, published datasets. Here are two key sources that can be used from the start:

1. The Extended Cohn-Kanade (CK+) Dataset – widely used in facial expression and emotion recognition research. It contains over 500 labeled sequences of facial expressions captured in controlled conditions. (https://paperswithcode.com/dataset/ck)
2. RAF-DB (Real-world Affective Faces Database) – includes around 30,000 facial images with diverse expressions in real-world settings, useful for sentiment/emotion classification models. (https://www.kaggle.com/datasets/shuvoalok/raf-db-dataset)

These datasets are publicly available and well-suited for developing and validating the core algorithms.