Bipolar Morphological Neural Networks: Convolution Without Multiplication
This work addresses efficiency for mobile and embedded systems, but it is incremental as it builds on existing morphological neural networks with modifications for better approximation and training.
The paper tackles the problem of making neural networks more efficient for mobile and embedded systems by introducing bipolar morphological neural networks that use only addition, subtraction, and maximum operations instead of multiplication. The result shows that converting pre-trained convolutional layers to these networks on MNIST and MRZ symbol recognition tasks leads to only a moderate decrease in accuracy while enabling faster inference.
In the paper we introduce a novel bipolar morphological neuron and bipolar morphological layer models. The models use only such operations as addition, subtraction and maximum inside the neuron and exponent and logarithm as activation functions for the layer. The proposed models unlike previously introduced morphological neural networks approximate the classical computations and show better recognition results. We also propose layer-by-layer approach to train the bipolar morphological networks, which can be further developed to an incremental approach for separate neurons to get higher accuracy. Both these approaches do not require special training algorithms and can use a variety of gradient descent methods. To demonstrate efficiency of the proposed model we consider classical convolutional neural networks and convert the pre-trained convolutional layers to the bipolar morphological layers. Seeing that the experiments on recognition of MNIST and MRZ symbols show only moderate decrease of accuracy after conversion and training, bipolar neuron model can provide faster inference and be very useful in mobile and embedded systems.