Towards Data Science 12:52 pm on May 23, 2024
This excerpt delves into self-confirming models in machine learning, where algorithms are designed to align predictions with user actions. It references Turing's thoughts on intelligent design limitations and the evolution of computational intelligence, mentioning bots religious beliefs as an example of adaptive theories. The text also touches upon historical advancements like perceptrons and neural approximations while exploring topics such as group invariance proofs, self-fulfilling prophecies, and interpretative frameworks for deep learning models.
- Self-confirming Models: Algorithms are structured to predict user actions rather than outcomes.
- Historical Context: The evolution of computational intelligence is traced back to Turing's theories and further developments in neural networks.
- Adaptive Theories and Self-Fulfilling Prophecies: Examining how beliefs evolve, such as bots religious views shaping their perceived worldview.
- Advancements in Machine Learning: Highlighting innovations like perceptrons and neural approximations alongside group invariants in learning algorithms.
- Interpretation of Deep Learning Models: Discussion on how deep learning models, such as those based on color tableaus, are explained or influenced by user beliefs.
https://towardsdatascience.com/language-as-a-universal-learning-machine-d2c67cb15e5f
< Previous Story - Next Story >