AI · Theory

Inductive vs. Deductive reasoning

Theory

43/215

Inductive vs. Deductive reasoning

Deduction: from general rule to specific case. Sound if the rule is true. Induction: from examples to a general rule. Can be wrong on the next example. Science and ML live on induction; proofs live on deduction.

Trap — calling a neural net “deductive”. It is inductive pattern fitting unless you wrap it in a proof system.

Output — idea / do / check for Inductive vs. Deductive reasoning. Fill those three in the viva.

Exam tip

ML = induction. Logic proof = deduction.

Example

# Inductive vs. Deductive reasoning
print("idea :", "Inductive vs. Deductive reasoning")
print("do   : one tiny example on paper")
print("check: one failure case")

Output: idea / do / check for Inductive vs. Deductive reasoning. Fill those three in the viva.

Short notes

DefDeduction : rule → case.
RuleInduction : cases → rule.

AI · Theory

Inductive vs. Deductive reasoning

← All stacks

Theory

43/215

Inductive vs. Deductive reasoning

Trap — calling a neural net “deductive”. It is inductive pattern fitting unless you wrap it in a proof system.

Output — idea / do / check for Inductive vs. Deductive reasoning. Fill those three in the viva.

Exam tip

ML = induction. Logic proof = deduction.

Example

# Inductive vs. Deductive reasoning
print("idea :", "Inductive vs. Deductive reasoning")
print("do   : one tiny example on paper")
print("check: one failure case")

Output: idea / do / check for Inductive vs. Deductive reasoning. Fill those three in the viva.

Short notes

DefDeduction : rule → case.
RuleInduction : cases → rule.

Inductive vs. Deductive reasoning

Example

Short notes

Questions

Inductive vs. Deductive reasoning

Example

Short notes

Questions