use dataset above for your project. Write short description of the dataset and what your CNN will
No answer yet for this question.
Ask a Tutor
Question:
- use dataset above for your project.
- Write short description of the dataset and what your CNN will recognize.
- Import the appropriate libraries: tensorflow, keras, numpy, glob, matplotlib, MaxPooling2D, and any others that would fit your model.
- Initialize the CNN.
- Utilize the following arguments to add a convolutional layer: Filters, Kernel_size, Padding, Activation Function - Relu, and Input shape.
- Down sample the images by applying a pooling operation.
- Did you choose max pooling, average pooling, or global pooling? Explain which type of pooling you used and why. What are the advantages and disadvantages of your pooling choice?
- Repeat steps 5-7 to add 3 more convolutional layers.
- Convert the dataset into a 1-D array for input into the next layer (flattening the dataset), which is fully linked.
- Use the dense class to create a fully connected layer (relu activation) and output one (softmax activation).
- Train, then appraise the CNN you just did. Compile the CNN model using compile, with three parameters:
- Loss Function: use categorical_crossentropy
- Optimizer: your choice (Adam, Momentum, Nesterov Accelerated Gradient, or Min-Batch Gradient Descent).
- Metrics Arguments: Accuracy to evaluate performance. Fit the model on the training set with at least 85 iterations (epochs). Evaluate the result. Compare the accuracy and loss function for both the training and test dataset. Plot the loss graph. Plot the accuracy graph. - Discuss how the CNN model is utilized in recognizing the images from the dataset and which optimizer provides for the performance model (highest accuracy and how many times to get to that level) the overall performance of your model. Justify your choice of optimizer by comparing it to two other optimizers.
Related Book For
Auditing a business risk appraoch
ISBN: 978-0324375589
6th Edition
Authors: larry e. rittenberg, bradley j. schwieger, karla m. johnston
Posted Date: