Uploaded August 2018 | Updated September 2026, 1 week ago
I continue on the topic of neural networks and in this video I use Microsoft Excel to demonstrate how the convolution operation works (so no R!). The first few minutes just explains some Excel basics, so skip a bit if you are familiar with it.
The idea (at least in the first few convolutions) is to learn basic patterns such as edges. As the filter moves across the image, element-wise multiplication and summation creates a new tensor (with the depth based on the number of filters (kernels) used).
The actual values in the filter are parameters which the networks learns. In this video, though, I use special values to demonstrate how a filter can show where an edge in an image is.
I continue on the topic of neural networks and in this video I use Microsoft Excel to demonstrate how the convolution operation works (so no R!). The first few minutes just explains some Excel basics, so skip a bit if you are familiar with it.
The idea (at least in the first few convolutions) is to learn basic patterns such as edges. As the filter moves across the image, element-wise multiplication and summation creates a new tensor (with the depth based on the number of filters (kernels) used).
The actual values in the filter are parameters which the networks learns. In this video, though, I use special values to demonstrate how a filter can show where an edge in an image is.
