2 ms·
This article is mis-titled. The ORNL project seems to be entirely about optimizing DNN hyperparameters (like Google's AutoML, I think), and not about inventing
by randcraw 9y ago
This article is mis-titled. The ORNL project seems to be entirely about optimizing DNN hyperparameters (like Google's AutoML, I think), and not about inventing novel DNN architectures.
Worse, the work has absolutely nothing to do with Scientific Discovery, which is a difficult and hugely ambitious area of ML study which involves generating novel induction of necessary mechanisms, not just the mostly random generation and testing of plausible patterns against outcomes.
Unless a ML method proposes hypotheses that are based on the induction of novel mechanistic principles, it ain't discovery. It's mere trial and error.
[Edited for clarity.]
- chriswarbo 9y agoYes, when I saw 'fueling scientific discovery' at first I thought of Adam and Eve: http://www.cam.ac.uk/research/news/robot-scientist-becomes-first-machine-to-discover-new-scientific-knowledge http://www.cam.ac.uk/research/news/robot-scientist-becomes-f... http://www.cam.ac.uk/research/news/artificially-intelligent-robot-scientist-eve-could-boost-search-for-new-drugs http://www.cam.ac.uk/research/news/artificially-intelligent-... Or work by the likes of Hod Lipson on discovering mechanisms/explanations from observations: https://www.creativemachineslab.com/eureqa.html https://www.creativemachineslab.com/eureqa.html I thought AI-applied-to-AI might be related to the work of Schmidhuber et al on self-improving search algorithms: ftp://ftp.idsia.ch/pub/techrep/IDSIA-16-00.ps.gz http://people.idsia.ch/~juergen/goedelmachine.html http://people.idsia.ch/~juergen/goedelmachine.html https://arxiv.org/abs/1210.8385 https://arxiv.org/abs/1210.8385