How Neural Networks Learn | CMU Training Part I | 11-785 | L3 | Explainer

This explainer follows the first steps of network training: using sampled data to define empirical risk, choosing differentiable activations and losses, and applying the chain rule to improve parameters. Differentiability makes optimization tractable; it does not guarantee a global optimum or reliable performance beyond the training sample. The original lecture was published in January 2026. ORIGINAL SOURCE Creator: Carnegie Mellon University Deep Learning Title: CMU Introduction to Deep Learning 11785, Spring 2026: Lecture 3 Published: 2026-01-16    • CMU Introduction to Deep Learning 11785, S...   ABOUT SPLAINERS HQ Splainers HQ provides explainers for important discussions happening in the tech world. Independent educational summary; not affiliated with or endorsed by the original creator or speaker. Watch the original for full context. #TechSplainers #DeepLearning #CMU