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.
Duration 14 hours
Course Outline
Core Principles of Safe and Fair AI
- Essential concepts: safety, bias, fairness, and transparency
- Categories of bias: dataset bias, representation bias, and algorithmic bias
- Introduction to key regulatory frameworks (such as the EU AI Act and GDPR)
Bias Dynamics in Fine-Tuned Models
- Mechanisms by which fine-tuning may introduce or exacerbate bias
- Analysis of case studies and real-world operational failures
- Methods for detecting bias within datasets and model predictions
Strategies for Mitigating Bias
- Data-centric approaches (rebalancing, data augmentation)
- In-training methods (regularization, adversarial debiasing)
- Post-processing solutions (output filtering, model calibration)
Enhancing Model Safety and Robustness
- Techniques for identifying unsafe or harmful outputs
- Managing adversarial inputs
- Conducting red teaming and stress tests on fine-tuned models
AI System Auditing and Continuous Monitoring
- Evaluation metrics for bias and fairness (e.g., demographic parity)
- Leveraging explainability tools and transparency frameworks
- Implementing ongoing monitoring and governance protocols
Toolkits and Practical Application
- Utilizing open-source libraries (e.g., Fairlearn, Transformers, CheckList)
- Practical exercise: Detecting and reducing bias in a fine-tuned model
- Generating safe outputs via advanced prompt design and constraints
Enterprise Applications and Compliance Preparation
- Best practices for embedding safety measures into LLM workflows
- Creating documentation and model cards for regulatory compliance
- Readiness for audits and external reviews
Conclusion and Path Forward
Requirements
- A solid grasp of machine learning model architectures and training workflows
- Practical experience in fine-tuning processes and working with Large Language Models (LLMs)
- Proficiency in Python and a working knowledge of Natural Language Processing (NLP) principles
Target Audience
- AI compliance teams
- Machine learning engineers
Custom Corporate Training
Training solutions designed exclusively for businesses.
- Customized Content: We adapt the syllabus and practical exercises to the real goals and needs of your project.
- Flexible Schedule: Dates and times adapted to your team's agenda.
- Format: Online (live), In-company (at your offices), or Hybrid.
Price per private group, online live training, starting from 2900 € + VAT*
Contact us for an exact quote and to hear our latest promotions