Probabilistic Machine Learning - Introduction
Probabilistic machine learning - foundations and frontiers
Speaker: Zoubin Ghahramani Video: https://livestream.com/oxuni/Stracheylecture-Ghahramani/videos/151163648
My notes from the Strachey lecture on Probabilistic machine learning at Oxford University.
- Related terms
- deep learning
- data science
- neural networks
- ai
- Applications
- computer vision
- image captioning
- face detection
- speech recognition
- scientific data analysis
- recommender systems
- self-driving cars
- robotics
- dogs playing football
- financial prediction
- computer games
- libratus: machine poker player
- computer vision
- Canonical problems
- classification: predict discrete class label from input data
- regression: predict continuous quantities from input data
- clustering: group data together
- dimensionality problem
- semi-supervised learning: learn from both labelled and unlabelled data
- reinforcement learning
- Deep learning systems are like neural networks of the 80s and 90s with innovations in some areas:
- volume of data
- larger compute resources
-
Limitations od deep learning
-
Probabilistic representation of uncertainty
- Automating machine learning
References
- https://en.wikipedia.org/wiki/Bayesβ_theorem
- https://www.automaticstatistician.com/index/
- https://en.wikipedia.org/wiki/Prior_probability
- https://en.wikipedia.org/wiki/Causality