"single neuron to achieve single classification problem", "multiple neurons to achieve multi-classification problem"
these two sentences how to understand, a single neuron can not achieve multi-classification problem?
"single neuron to achieve single classification problem", "multiple neurons to achieve multi-classification problem"
these two sentences how to understand, a single neuron can not achieve multi-classification problem?
to talk about my personal humble opinion, Quan should throw a brick to attract jade. The essence of the classification problem is to project the original data space to another space so as to maximize the distance between classes in the new space. If you use a single neuron to output one-dimensional data, one-dimensional data can only express an one-dimensional space. Obviously, the higher the output dimension, the more you can separate the data. A single neuron can achieve two classification, the two types of data are distributed at both ends of one-dimensional space, and can also be distinguished. If you do multi-classification, it can also be achieved through segmentation, but the effect is often not the best.
Previous: How does OHIF set the default date filter?
Next: What happens when you can't scroll with vue-seamless-scroll?
Code: import tensorflow as tf import numpy as np tf.enable_eager_execution () class DataLoader (): def __init__(self): mnist = tf.keras.datasets.mnist.load_data(path = mnist.npz ) self.train_data = mnist[0][0] self.train_data = np....