Hacker News Reader

For real.

People are downloading ready-made models from repositories to try and solve minor problems.

Guess what, your problem might be a simple linear regression. Yes you can solve it with a DNN (one level, one neuron - but hey, don't keep that from putting it into your CV) but you don't need to.

View on HN

At university, I generated Markov chains of the solution space from a single neuron that was being used as a binary classifier. You take n samples, average them out and look at the decision boundary. The decision boundary itself is linear but the margin of error is not.

It was really cool. Attempting to implent Hamiltonian MCMC on a single neuron really forced you to learn what a gradient is in regards to NN.

Reply on news.ycombinator.com