Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Course Outline
Introduction to AI in the Financial Sector
- Key applications: fraud detection, credit scoring, and compliance monitoring
- Regulatory landscapes and risk management frameworks
- An overview of fine-tuning strategies in high-risk contexts
Preparing Financial Data for Model Adaptation
- Data sources: transaction records, customer profiles, and behavioural metrics
- Ensuring data privacy, anonymisation, and secure handling
- Feature engineering for tabular and time-series information
Techniques for Model Fine-Tuning
- Applying transfer learning and adapting models to financial contexts
- Selecting domain-specific loss functions and performance metrics
- Utilising LoRA and adapter tuning for cost-effective updates
Modelling for Risk Prediction
- Predictive approaches for loan default and credit assessment
- Navigating the trade-off between model interpretability and accuracy
- Managing imbalanced datasets within risk scenarios
Applications in Fraud Detection
- Constructing anomaly detection pipelines using fine-tuned models
- Strategies for real-time versus batch processing in fraud prediction
- Hybrid approaches: combining rule-based systems with AI-driven detection
Evaluation and Explainability
- Assessing model performance: precision, recall, F1 scores, and AUC-ROC
- Employing SHAP, LIME, and other explainability tools
- Conducting audits and generating compliance reports for fine-tuned models
Production Deployment and Monitoring
- Integrating fine-tuned models into existing financial platforms
- Establishing CI/CD pipelines for AI systems in banking
- Monitoring data drift, retraining schedules, and overall lifecycle management
Conclusion and Path Forward
Requirements
- A solid grasp of supervised learning methods
- Practical experience with Python-based machine learning frameworks
- Familiarity with financial data sources, including transaction logs, credit scores, and KYC information
Target Audience
- Data scientists working in financial services
- AI engineers collaborating with fintech or banking entities
- Machine learning specialists developing risk or fraud models
14 Hours
