Probabilistic Programming
• 1 min read 1 min
What I worked on
Read up on probabilistic programming based on Vikash Mansinghka’s MIT course. Compared it to DNN approaches and explored how Bayes nets fit into AI.
What I noticed
- Probabilistic programming uses structured uncertainty instead of weights
- Bayes nets visualize relationships between variables
- LLMs aren’t purely probabilistic programs but share generative structure
”Aha” Moment
n/a
What still feels messy
How probabilistic models and deep neural networks could be combined in one learning pipeline.
Next step
n/a