News for hackers. Stuff that matters.

A comment on “Less than one”-shot learning.

Re:“Less than one”-shot learning by codelord on Sunday October 18, 2020 @04:17PM (#24820790)
The title is click-bait. This has been known for several years[1], the technique has little practical value, and the assertion that you can learn from no data is completely false and misleading. The training data was compressed to a few examples. To the journalist: it's OK not to maximize for click-bait when you write an article. [1]: https://www.ttic.edu/dl/dark14.pdf
[ Parent | on HN ]
Re:“Less than one”-shot learning by bonoboTP on Sunday October 18, 2020 @04:41PM (#24820969)
This is not a magazine article written by a journalist, but a press release written by MIT marketing people to hype up research from their university.
[ Parent | on HN ]
Re:“Less than one”-shot learning by jerry80 on Sunday October 18, 2020 @05:21PM (#24821226)
Why not just make one image where the pixel values are the trained network weights? Then you can say you can distill any training set into just one image.
[ Parent | on HN ]
Re:“Less than one”-shot learning by vonwoodson on Sunday October 18, 2020 @06:11PM (#24821522)
You just nailed it. The whole thing feels like reinventing the wheel of ML.
[ Parent | on HN ]