A deep neural network fits the training data well but performs poorly on test data. Which of the following should you consider to reduce overfitting? (Choose three.)

Overfitting occurs when a model learns the training data too well, including noise, leading to poor generalization on unseen data. To combat this, we need to introduce mechanisms that constrain the model's complexity. Increasing regularization (e.g., L1 or L2 regularization) adds a penalty to the loss function for large weights, encouraging the model to learn simpler patterns and preventing it from relying too heavily on any single feature. Increasing dropout randomly sets a fraction of neuron activations to zero during training. This forces the network to learn more robust features that are not dependent on the presence of specific neurons, effectively creating an ensemble of smaller networks and reducing co-adaptation. Reducing the number of feature interactions simplifies the model by decreasing the number of complex, non-linear relationships it can learn. This directly limits the model's capacity to memorize noise in the training data. Decreasing regularization or decreasing dropout would allow the model more freedom to fit the training data, exacerbating overfitting. Increasing the number of feature interactions would also increase model complexity, making overfitting more likely.