Titulo Estágio
Behavioural Biometric Scam Detection
Local do Estágio
Coimbra
Enquadramento
Use behavioural biometric data to materially understand the usage of the data for scam risk. What signals can be quantified (hesitation, active call via gyroscopic sensors, dictation patterns, light sensors etc.)
Objetivo
Use behavioural biometric data to materially understand the usage of the data for scam risk. What signals can be quantified (hesitation, active call via gyroscopic sensors, dictation patterns, light sensors etc.)
Plano de Trabalhos - Semestre 1
"Stages:
1. Review of Feedzai data collection practices in Digital Activity
2. Investigation of behavior patterns associated with Scams
3. Define methodology to pipeline from data collection to scam evaluation
Expected results:
1. Documentation of scam-related behavior
2. Execution plan with data collection approach well defined"
Plano de Trabalhos - Semestre 2
"Stages:
1. Raw data collection with Feedzai's demo app
2. Simulation of scam behavior
3. Data understanding and normalization
4. Comparison with real-world data (pending data access)
5. Modeling of at least 1 scam-related behavior (e.g. hesitation)
6. Documentation of results
Expected results:
1. Data collection methodology guidelines for scam simulation
2. Final report with results and presentation
3. Model binary for at least 1 scam-related behavior"
Condições
Remunerated
Observações
Data is to be collected from Feedzai Digital Activity monitoring SDK which is deployed on Banking mobile and web apps. Data attributes include but are not limited to: (A) device info such as gyroscope, accelerometer sensors; (B) user behavior such as tap screen position, timestamp; (C) application info such as screen ID; (D) network data such as IP address and (E) operational feedback, ie if a session is deemed risky or fraudulent.
Orientador
David Garcia Munoz
david.garcia@feedzai.com 📩