The trend suggests that as you increase the width of a neural network, the accuracy increases till a certain threshold value, an
1 answer:
Hi, you've asked an incomplete question. Here's the remaining part of the questions:
What could be the possible reason for this decrease?
A. Even if number of kernels increase, only few of them are used for prediction
B. As the number of kernels increase, the predictive power of neural network decrease
C. As the number of kernels increase, they start to correlate with each other which in turn helps overfitting
D. None of these
Answer:
<u>C. As the number of kernels increase, they start to correlate with each other which in turn helps overfitting</u>
<u>Explanation:</u>
<em>Remember, </em>Kernels refer to computer programs that occupy <em>part of the operating system memory.</em>
By looking closely at the graph we could notice that the positive trend of the test accuracy couldn't be sustained as a result of a constant increase in the width of the neural network.
You might be interested in
Two transformations that can be used to show that polygon ABCDE is congruent to polygon GFJIH : <u>rotation -90° and translation (0,2)</u>
<h3>Further explanation</h3>There are several forms of transformation, including translation, dilation, rotation or reflection
There are 2 shapes of geometry:
Similar figures:
the same shape but different in size,
Congruent figures:
the same size and the same shape
Both polygons will appear congruent if, after being transformed, the two polygons will map exactly to each other
A sequence of two transformations e can do is :
1. rotation -90°
Let point H(4,1) from polygon GFJIH , we rotate -90°
2. translation (0,2)
Keywords: transformation, translation, rotation
#LearnWithBrainly
