| hwConcept | Description | Emoji | number game example | bayes net | | ---------------------- | ------------------------------------------------------------------------------------------------------------ | ----- | ------------------- | --------- | | Hypothesis space (H) | Space of possible concepts | ⬛️ | | | | Examples (X) | X = {x1, . . . , xn}: n examples of a concept C | ⚫️ | | | | Prior | p(h): domain knowledge, pre-existing biases | 🟩 | | | | Likelihood | p(X\|h): statistical information in examples | 🔴 | | | | Posterior | p(h\|X): degree of belief that h is the true extension of C | 🟦 | | | | Size principle | Smaller hypotheses receive greater likelihood, exponentially more so as n increases | 📌 | | | | Choice principle | Choice of hypothesis space embodies a strong prior, favoring natural over logical possibilities | 🌲>🎄 | | | | Hypothesis averaging | p(y ∈ C \| X): Probability that C applies to new object y, averaged over all h weighted by p(h\|X) | ⭐️ | | | | Conservation of belief | Total probability must sum to 1; increased belief in some hypotheses necessitates decreased belief in others | ⚖️ | | |