Bayesian Inference Insights
Jonas explains the elegance of closed-form solutions in Bayesian inference, particularly with Gaussian distributions. He emphasizes the importance of linking inference time computation to confidence estimates, suggesting a method to optimize resource usage by stopping computation when uncertainty reduction becomes minimal. This approach allows for more efficient model improvements while managing computational costs effectively.In this clip
From this podcast
Machine Learning Street Talk (MLST)
Jonas Hübotter (ETH) - Test Time Inference
Related Questions