MNIST: single layer NN with 784 neuron; is %90 error rate normal?

Question

I am relatively new to Neurol Networks, and to understand the basics, I converted MNIST database into a format that I liked, and wrote a single layer NN with 784 neurons from scratch without using any library related to NN.

I trained NN over 600samples, and tested it on 10000samples (I accept that the reverse would be much better). I can see that as the NN trains with more and more samples, the error decreases almost exponentially with the training sample size. However, in the end, the test error was %90.

Is this normal for such a simple NN? Where can I find performance of different types of NNs with different parameters to compare with my own NN?

Of course, this is a very simple NN, but before making it more complex, I would like to the results I am getting are in agreement with others, and I would like to get an idea about what kind of different methods & schemes could be used in other situations.

You should set up the same model with an established framework and see what the performance is. 10% correct is very near to random guess rate. — Jon Nordby, Mar 14 '20 at 12:19

Tim · Accepted Answer · 2020-03-14T13:41:48.643

2

MNIST is a pretty simple, "toy" dataset, and you can get high performance ($>90\%$ accuracy) even with simple classifiers like logistic regression, single layer neural network, or $k$-NN. You can find summary of the results on the page maintained by Yann LeCun.

edited Mar 14 '20 at 13:41

answered Mar 14 '20 at 12:38

Tim

108,699
20
212
390

MNIST: single layer NN with 784 neuron; is %90 error rate normal?

1 Answers1