Proactive Management Of Lost Bank Cards

In the current rapidly advancing digital era, financial institutions need to ensure the secure handling of compromised bank cards, to reduce losses from unauthorized transactions. Although modern technologies have advanced, mitigation strategies for compromised bank cards remain largely reactive and dependent on customers reporting incidents. This reactive paradigm has proven inadequate as credit card adoption increases and financial fraud becomes more sophisticated. Existing solutions, such as manual card freeze and transaction alerts, offer partial mitigation, exhibiting limited capacity to predict or intercept losses through contextual behaviour analysis in real-time. This study proposes a proactive framework using Long-Term Short-Term (LSTM) Autoencoders to address these limitations. A customer profile is built to model expected behaviour, and detect observations that deviate from the expected behaviour profile as potential indicators of card loss. Beyond the financial landscape, applications of the proposed framework are scalable to any sector with observable digital behavioural trails. For instance, student ID cards generate location-based access logs that can help analyse access pattern behaviour, enabling the detection of misplaced student cards and timely intervention of unauthorised access. Similarly, modelling the usage of equipment in industrial settings can facilitate fault detection, offering early signs of mechanical failure. In modern connected vehicle environments, deviations from well-established driver behaviour could indicate theft, misuse, or safety concerns.

Grantor dc:publisher

Stellenbosch : Stellenbosch University

Year dc:date.issued

2026

Author dc:creator

• Ngcobo, Sakhile Pine

Advisor dc:contributor.advisor

• Gwetu, Mandla

Language dc:language.iso

en

Repository record dc:identifier.uri

https://scholar.sun.ac.za/handle/10019.1/136095

OAI identifier oai:identifier

oai:scholar.sun.ac.za:10019.1/136095