Local Critic Training of Deep Neural Networks
This method addresses computational bottlenecks in training deep neural networks, offering incremental improvements for researchers and practitioners in machine learning.
The paper tackles the problem of layer-wise dependency in backpropagation training by introducing local critic networks to compute error gradients without full feedforward and backward propagation, resulting in improved computational efficiency and performance in tasks like structural optimization and ensemble classification.
This paper proposes a novel approach to train deep neural networks by unlocking the layer-wise dependency of backpropagation training. The approach employs additional modules called local critic networks besides the main network model to be trained, which are used to obtain error gradients without complete feedforward and backward propagation processes. We propose a cascaded learning strategy for these local networks. In addition, the approach is also useful from multi-model perspectives, including structural optimization of neural networks, computationally efficient progressive inference, and ensemble classification for performance improvement. Experimental results show the effectiveness of the proposed approach and suggest guidelines for determining appropriate algorithm parameters.