TA_NNA

4. Neural network where connections move only in one direction—from input to output.
7. Type of neural network with feedback connections allowing memory of previous inputs.
9. Trade-off that determines a model’s ability to generalize beyond training data.
10. Adaptive optimization algorithm combining momentum and RMSProp for faster convergence.
12. Simplest type of artificial neuron that performs binary classification.
14. Optimization algorithm that minimizes error by updating weights in the opposite direction of the gradient.
15. Algorithm used to train multilayer neural networks by propagating errors backward.
16. Type of learning where models find hidden patterns in unlabeled data.

1. Problem where a model learns the training data too well but fails on new data.
2. A machine learning task where inputs are assigned to discrete categories.
3. Technique that accelerates gradient descent by considering previous updates.
5. Technique used to prevent overfitting by adding a penalty to model complexity.
6. Learning paradigm where the model is trained using labeled data.
8. Regularization method that halts training when validation error begins to increase.
11. Abbreviation for Multilayer Perceptron, a type of feedforward neural network.
13. The basic biological and computational unit that processes and transmits information.