I was wondering if it’s possible, since all the examples I found online use convnets. I am trying to get an ok accuracy (say 90%) with just a feedforward net and no convolutions.

I know the convnet is the way to go, but I am still curious.

For example, even the pytorch tutorial uses a convnet for MNIST, but I can get around 97% accuracy in it with a feedforward without convolutions.

So far however, for CIFAR10 I’m only achieving 28 to 30% accuracy with the following setup: (linear = linear combination plus bias)

- linear(32
*32*3, 300), ReLU - linear(300, 200), ReLU
- linear(200, 100), ReLU
- linear(100, 50), ReLU
- linear(50, 10), Softmax

. Cross Entropy Loss