Which technique helps prevent overfitting in neural networks by randomly deactivating neurons during training?

Correct answer: Dropout

Option C is correct because dropout is a regularization technique that randomly sets a fraction of neuron activations to zero during each training step, forcing the network to learn redundant representations and preventing co-adaptation of neurons, which significantly reduces overfitting. Option A is incorrect because L1 regularization adds a penalty proportional to the absolute value of weights to the loss function, encouraging sparsity, but it does not randomly deactivate neurons during training. Option B is incorrect because batch normalization normalizes layer inputs to stabilize and accelerate training, and while it has a mild regularizing effect, it does not work by randomly deactivating neurons. Option D is incorrect because early stopping halts training when validation performance begins to degrade, preventing overfitting by limiting training epochs, but it does not involve randomly deactivating neurons.