10:33:34	 From  Marcin Puchlik : How to apply ML in the DSP area? Any ideas?
10:36:13	 From  Dave Comer : I was think about how to apply ML to measure the vibration of a furnace to predict the failure, or health of the furnace. It seems as if the main challenge is to capture failure. Can you suggest how you might approach this?
10:37:21	 From  rakhel : great talk, I don't know nothing about ML, but I got most of the things that you went through in your talk. Thank you !
10:39:58	 From  Michael Kirkhart : Yesterday's talk on "Data Centric AI for Signal Processing Applications" discussed the use of an autoencoder for prediction of failures in air compressors.
10:41:49	 From  Dave Comer : Thanks Michael
10:41:54	 From  Leonard : here is an article where I synthesized the data and used random forest to categorize symbol
10:41:57	 From  Leonard : http://koliberengineering.blogspot.com/2020/02/categorical-estimation-of-qam-symbols_27.html
10:44:44	 From  Dave Comer : Definitely Michael. I would have to wait for a furnace to fail many time. That is my quandary...
10:45:39	 From  Dave Comer : Where can we get the Juypter file?
10:47:20	 From  Michael Kirkhart : @Dave Lomer: the autoencoder example from yesterday did not need data from furnace failures; it only needs data from a "normally operating" furnace.