Question

Project Description: In this project, you need to complete the following tasks: 1.) First, generating the double half-moon pattern following the example code shown during the computer experiment video. The radius of the moon is 10, width is 6, and distance between two half-moon is 0.5. Total sample number is 2500. 2.) Constructing your neural network: 2 inputs (coordinates), 1 extra input (bias), 1 neuron, 1 activation function, and 1 output (-1 or 1). 3.) Initializing weights vector using random number generator. 4.) Choosing training data (500) and testing data (2000). 5.) Training your neural network with Gauss-Newton method. Details could be found from slides P30 ~ P31. 6.) Comparing your results with training data use the initial weight vector and the well-trained weight vector. 7.) Comparing your results with testing data use the initial weight vector and the well-trained weight vector. 8.) Plotting the MSE respect to training iterations.

          Project Description:
In this project, you need to complete the following tasks:
1.) First, generating the double half-moon pattern following the example code shown during
the computer experiment video. The radius of the moon is 10, width is 6, and distance
between two half-moon is 0.5. Total sample number is 2500.
2.) Constructing your neural network: 2 inputs (coordinates), 1 extra input (bias), 1 neuron, 1
activation function, and 1 output (-1 or 1).
3.) Initializing weights vector using random number generator.
4.) Choosing training data (500) and testing data (2000).
5.) Training your neural network with Gauss-Newton method. Details could be found from
slides P30 ~ P31.
6.) Comparing your results with training data use the initial weight vector and the well-trained
weight vector.
7.) Comparing your results with testing data use the initial weight vector and the well-trained
weight vector.
8.) Plotting the MSE respect to training iterations.

Project Description:
In this project, you need to complete the following tasks:
1.) First, generating the double half-moon pattern following the example code shown during
the computer experiment video. The radius of the moon is 10, width is 6, and distance
between two half-moon is 0.5. Total sample number is 2500.
2.) Constructing your neural network: 2 inputs (coordinates), 1 extra input (bias), 1 neuron, 1
activation function, and 1 output (-1 or 1).
3.) Initializing weights vector using random number generator.
4.) Choosing training data (500) and testing data (2000).
5.) Training your neural network with Gauss-Newton method. Details could be found from
slides P30 P31.
6.) Comparing your results with training data use the initial weight vector and the well-trained
weight vector.
7.) Comparing your results with testing data use the initial weight vector and the well-trained
weight vector.
8.) Plotting the MSE respect to training iterations.

Added by Danny L.

Question

Please give Ace some feedback