China
Africa
Explore chapters and articles related to this topic
Cooperation through Communication in a Distributed Problem-Solving Network
Published in Satya Prakash Yadav, Dharmendra Prasad Mahato, Nguyen Thi Dieu Linh, Distributed Artificial Intelligence, 2020
Anisha Singh, Akarshita Jain, Bipin Kumar Rai
Receptive field is a restricted area where cortical neurons respond to stimuli. CNNs have multiple layers of receptive fields. Collection of small neurons process some part of the input image. This is then piled up so that their input regions overlap to achieve a higher-resolution portrayal of the original image. This process is repeated with every layer. Piling up of results may give the ability to CNNs to handle the conversion of the input image. CNN may have local or global pooling layers, which joins the outputs of neuron clusters. A convolution operation on small parts of input is done to minimize the number of free parameters. A software node network consists of neurons connected. All software Node Network connection has some weights with it that tells the dominance of the relationships within the neuron when multiplied with the input value. Every neuron is associated with an activation function used to present non-linearity in the network’s modeling capabilities that tell the neuron’s output. Preparing of Software Node Network is a finding out about the estimations of parameters and this learning procedure in a Software Node Network as a monotonous procedure of “proceeding to return” through all the layers of neurons. Starting from the last layer, for example, the yield layer, misfortune data goes to all the concealed layer neurons that add to the output. The hidden layer neurons only receive some parts of the loss’s total signal, based on the relative contribution that every neuron has contributed to the actual output. This process is done, again and again, layer by layer, till all the neurons in the network have received a loss signal that tells their relative contribution to the total loss.
Multi-scale filters implemented by cellular automaton for retinal layers modelling
Published in International Journal of Parallel, Emergent and Distributed Systems, 2020
The retinal visual treatment uses the concept of receptive field. The receptive field describes the region of space where a stimulation causes an increase of the activity of the neuron. Beginning with the Hubels and Wiesel studies [5–7], the studied properties of system receptive fields and more generally of the visual cortex are extremely thorough today.