Investigating the interaction between gradient-only line searches and different activation functions
This addresses a specific optimization problem for neural network training with discontinuous loss functions, but it is incremental as it builds on existing GOLS methods.
The study investigated how gradient-only line searches (GOLS) interact with different activation functions in neural network training, finding that GOLS are robust for most functions but sensitive to ReLU due to gradient sparsity, and showing that batch normalization and skip connections can mitigate these issues.
Gradient-only line searches (GOLS) adaptively determine step sizes along search directions for discontinuous loss functions resulting from dynamic mini-batch sub-sampling in neural network training. Step sizes in GOLS are determined by localizing Stochastic Non-Negative Associated Gradient Projection Points (SNN-GPPs) along descent directions. These are identified by a sign change in the directional derivative from negative to positive along a descent direction. Activation functions are a significant component of neural network architectures as they introduce non-linearities essential for complex function approximations. The smoothness and continuity characteristics of the activation functions directly affect the gradient characteristics of the loss function to be optimized. Therefore, it is of interest to investigate the relationship between activation functions and different neural network architectures in the context of GOLS. We find that GOLS are robust for a range of activation functions, but sensitive to the Rectified Linear Unit (ReLU) activation function in standard feedforward architectures. The zero-derivative in ReLU's negative input domain can lead to the gradient-vector becoming sparse, which severely affects training. We show that implementing architectural features such as batch normalization and skip connections can alleviate these difficulties and benefit training with GOLS for all activation functions considered.