AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Challenges in Working with Kate Models
This chapter discusses the main challenges in working with Kate models, including treatment assignment biases, covariate shift, missing labels, and difficulty in evaluating and deploying these models in practice. Two papers related to causal discovery and treatment effect estimation are discussed, highlighting the impact of synthetic data construction on model performance and the lack of authoritative guidance on realistic data-generating processes for benchmarking.