Activation Functions: Do They Represent A Trade-Off Between Modular Nature of Neural Networks And Task Performance
This work addresses the design of neural network architectures for researchers and practitioners, but it appears incremental as it focuses on a specific component without presenting new results.
The paper investigates whether the ReLU activation function is optimal for achieving better modularity in neural networks, exploring the trade-off between modular structure and task performance.
Current research suggests that the key factors in designing neural network architectures involve choosing number of filters for every convolution layer, number of hidden neurons for every fully connected layer, dropout and pruning. The default activation function in most cases is the ReLU, as it has empirically shown faster training convergence. We explore whether ReLU is the best choice if one is aiming to desire better modularity structure within a neural network.