Crossing Duck

3. The 1943 pioneers of the first artificial neuron model — Warren _______ and Walter Pitts.
4. The method of representing knowledge across many connections rather than one neuron.
7. The 2016 Google system that switched from rule-based translation to neural networks.
8. The brain-like system that learns by adjusting connections based on examples.
11. The learning process where the network adjusts after seeing examples and errors.
13. The property that enables a neural system to adjust automatically when exposed to new data.
14. The AI approach that learns patterns rather than following explicit rules.
16. The problem highlighted by Minsky and Papert that the Perceptron could not solve — the XOR or “________ OR” problem.
17. The simplified machine created by Frank Rosenblatt that could learn from examples.
19. The capability that allows neural networks to function even when some neurons fail.
20. The output of a neural network that identifies what the system has recognized (e.g., “cat”).

1. The rival approach to Connectionist AI that relies on logic and symbols.
2. The algorithm that measures how far off the prediction is from the correct answer.
5. The phrase describing Hebbian learning: “Neurons that fire together, _____ together.”
6. The 1980s algorithm that allowed training of multilayer neural networks.
9. The key mathematical process used to adjust weights and reduce prediction errors.
10. The small adjustable numerical values that determine a neural connection’s importance.
12. The type of data processing where many neurons work at once rather than one by one.
15. The ability of neural networks to apply what they’ve learned to unseen data.
18. The visual AI system example used to demonstrate distributed representation — recognizes images like “dog” or “cat.”