Impact Statement:Most current deep neural networks consist of the basic framework of convolution and multilayer perceptron, which lack biological plasticity. Designing a novel network fra...Show More
Abstract:
Simulating the method of neurons in the human brain that process signals is crucial for constructing a neural network with biological interpretability. However, existing ...Show MoreMetadata
Impact Statement:
Most current deep neural networks consist of the basic framework of convolution and multilayer perceptron, which lack biological plasticity. Designing a novel network framework that more closely resembles the neurons of the human brain has been one of the top priorities for artificial intelligence. In this article, we present a multidendritic shallow network architecture called MDPN based on the biological framework of pyramidal neuron to solve the image classification problem. In comparison with current network models, MDPN has the following advantages. 1)MDPN mimics the biological structure of a multipolar neuron to model a five-layer lightweight network, including presynaptic layer, dendritic layer, membrane layer, soma layer, and axon layer. To reflect the biological plasticity of the model, we propose for the first time a repeated information input as well as multiple nonlinear dendritic computations. This information processing increases the width of the model and solves the prob...
Abstract:
Simulating the method of neurons in the human brain that process signals is crucial for constructing a neural network with biological interpretability. However, existing deep neural networks simplify the function of a single neuron without considering dendritic plasticity. In this article, we present a multidendrite pyramidal neuron model (MDPN) for image classification, which mimics the multilevel dendritic structure of a nerve cell. Unlike the traditional feedforward network model, MDPN discards premature linear summation integration and employs a nonlinear dendritic computation such that improving the neuroplasticity. To model a lightweight and effective classification system, we emphasized the importance of single neuron and redefined the function of each subcomponent. Experimental results verify the effectiveness and robustness of our proposed MDPN in classifying 16 standardized image datasets with different characteristics. Compared to other state-of-the-art and well-known networ...
Published in: IEEE Transactions on Artificial Intelligence ( Volume: 5, Issue: 9, September 2024)