Across
- 8. A type of loss function in classification tasks, measuring the distance between probability distributions (12 letters).
- 9. A neural network architecture commonly used for sequence-based tasks like time series or language modeling (19 letters).
Down
- 1. The process of improving the performance of a model by fine-tuning its hyperparameters (15 letters).
- 2. A type of learning in which the system automatically detects patterns in data without labeled responses (16 letters).
- 3. The framework that uses a combination of multiple learning models to improve accuracy (9 letters).
- 4. A state-of-the-art deep learning technique used for text generation and translation, often involving attention mechanisms (11 letters).
- 5. A technique where a model generates predictions by aggregating the output of multiple algorithms (12 letters).
- 6. A method where agents learn by interacting with their environment and receiving feedback (18 letters).
- 7. The statistical property that refers to a model’s ability to perform well on unseen data (12 letters).
- 10. The process of reducing the complexity of a model to avoid overfitting (13 letters).