9 ms·
I think he should just go on with his research and don’t bother with understanding every obscure reference in papers. One can grasp the core ideas surprisingly
by tzahola 9y ago
I think he should just go on with his research and don’t bother with understanding every obscure reference in papers. One can grasp the core ideas surprisingly well even when skipping over proofs and formulas.
And when the time comes to write his own papers, he should remember to intentionally make it harder to read for outsiders. E.g. instead of writing “I calculated the total error by summing the per-neuron errors”, one should write “the loss function utilized an integral over the output lattice using a discretized method by Newton et. al.”, or some other bullshit.
- gfodor 9y agoAs an amateur who has jumped in and out of learning basic ML over the years, it has been interesting to see the web of terminology expand to the point where your post is no longer satire. Writing a dictionary or annotating papers to decipher ML-speak to basic-math speak would be a pretty worthwhile endeavor for someone (I see glimmers of this in the work being done by the folks at fast.ai) and in general would probably not remove much real information.