Co-adaptation Deep Learning
noun
Definition: In neural networks, co-adaptation is the tendency of neurons or feature detectors to become overly dependent on the activations of specific other neurons during training, which can reduce generalization and increase the risk of overfitting; dropout is commonly used to suppress such excessive dependencies [Gelesh, Sreeja 2025].
.Example in context: “Dropout works by randomly deactivating a subset of neurons during training, eliminating excessive co-adaptations and effectively training an ensemble of subnetworks …” [Çapkan et al. 2025]
Synonyms: excessive neuron dependency; feature-detector co-adaptation (near-synonyms / explanatory variants rather than fixed terminological doubles)
Related terms: dropout; overfitting; generalization; regularization; hidden units; feature detectors