Explicit Dropout: Deterministic Regularization for Transformer Architectures

arXiv:2604.20505v1 Announce Type: new Abstract: Dropout is a widely used regularization technique in deep learning, but its effects are typically realized through stochastic masking rather than explicit optimization objectives. We propose a deterministic formulation that expresses dropout as an additive regularizer directly incorporated into the training loss. The framework derives explicit regularization terms for Transformer architectures, covering attention query, key, value, and feed-forward components with independently controllable strengths. This formulation removes reliance on stochastic perturbations while providing clearer and fine-grained control over regularization strength. Experiments across image classification, temporal action detection, and audio classification show that explicit dropout matches or outperforms conventional implicit methods, with consistent gains when applied to attention and feed-forward network layers. Ablation studies demonstrate stable performance and controllable regularization through regularization coefficients and dropout rates. Overall, explicit dropout offers a practical and interpretable alternative to stochastic regularization while maintaining architectural flexibility across diverse tasks.

Related

• AdaCubic: An Adaptive Cubic Regularization Optimizer for Deep Learning
• Geometric Layer-wise Approximation Rates for Deep Networks
• Gating Enables Curvature: A Geometric Expressivity Gap in Attention
• On the Convergence of Gradient Descent on Learning Transformers with Residual Connections
• Toward Efficient Influence Function: Dropout as a Compression Tool

Source: arXiv cs.LG | 2026-04-23