Let's Talk VDML Podcast Episode 3 - Baseline Validation: Focus on Maritime Surveillance

Let's Talk VDML | Practical Conversations about AI Assurance
VDML (Validation Driven Machine Learning) is a methodology developed by KJR to guide the development of robust and reliable Machine Learning (ML) models. VDML is a practical method of undertaking responsible AI.
In this podcast you will learn about the process of baseline validation for pre-existing machine learning models. Podcast host Principal Consultant Steve Casa discusses with KJR founder Dr Kelvin Ross and KJR consultant Stan Potums the importance of using real-world production data to validate models, identifying key metrics for performance evaluation, and understanding the context in which the model will be deployed.
By sharing practical examples from maritime surveillance and computer vision applications, the episode illustrates the challenges and strategies for ensuring models are fit for purpose and highlights the iterative process of fine-tuning and optimising models.
• KJR: kjr.com.au/
• VDML: kjr.com.au/ser...
• Contact Us: [e-mail]