Does anyone remember an analysis from way back in the day that showed that the term “neural networks” was associated with the least likely papers to be accepted to NeurIPS? Like the term highly correlated with paper rejections, before everything became deeplearning?
@timnitGebru first, thanks for studying this history and apologies for responding to your question with a slightly different question inspired by it.
Are you aware of any movement forming to gather together the work of people like me that are trying to resist deep learning by developing less extractive alternatives to backprop networks? A journal or conference? Otherwise this work proves to be a rather solitary and disheartening slog.
@davidbeers I don't know about a movement per se but there are journals and various groups of people publishing these types of alternatives/writings etc.
Add comment