Medijski sadržaj
- Tweetovi
- Tweetovi i odgovori
- Medijski sadržaj, trenutna stranica.
-
Skimming this fascinating paper on
#DeepLearning and what we can learn from it about the human mind. One stray observation that struck me: best performing facial recognition model has 28 layers w/ 140M params; for language modeling, need 48 layers w/ > 1.5B params! https://twitter.com/talyarkoni/status/1225476054518726658 …pic.twitter.com/aa0FdkGBCK
-
Many good takes on what transpired at
#SOTU2020 - here's one I like: what happened makes clear as day to us that "2020 is going to be a year of division and polarization. The question is: will this also be the year we start writing a counter-story?" https://preemptivelove.org/blog/illusion-is-gone/ …pic.twitter.com/WB0bOfrEjM
Prikaži ovu nit -
Some neat example "journals" in this vein.
#IdeaToSelf - analyze the topical distribution of the papers in these journals and compare to traditional journals. Could also see whether these do a better job of bridging/linking science's network structure than traditional journals.pic.twitter.com/Ud5it8sE8B
Prikaži ovu nit -
Margaret Boden's idea of "transformational creativity" is a nice conceptual grounding for this too. Nicely described here (although not the original paper): http://computationalcreativity.net/iccc2015/proceedings/11_3Grace.pdf … - can be formalized with Bayesian surprise methods if you have good ways of mapping the domainpic.twitter.com/mNUaGISj3q
-
One nice example of this sort of qualitative shift measure is this work by
@pierre_azoulay: rate at which new standardized keywords are added to a field: https://onlinelibrary.wiley.com/doi/full/10.1111/j.1756-2171.2011.00140.x …pic.twitter.com/ORH5JYV2Bv
-
Folks in visual analytics also grapple with this: they want to create tools that increase p(insight), but that is hard to conceptualize and measure: https://ieeexplore.ieee.org/abstract/document/1626178 … - they come to a similar conclusionpic.twitter.com/P2L4HE9MuF
-
Provocative question! This q is quite similar to what we grapple with in
#creativity research. Agree there is no simple metric, need to triangulate. NSF-sponsored creativity support tools workshop (led in part by@benbendc) had this as a major conclusion: http://www.cs.umd.edu/hcil/CST/Papers/creativitybook_final.pdf …pic.twitter.com/Qng9pt36Kf
-
Idea I think is from Shipman et al's classic "formality considered harmful" http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.418.3597&rep=rep1&type=pdf …pic.twitter.com/VFad4aUAIv
-
#TIL Florence Nightingale invented the polar area diagram ca. 1850s h/t@drunkhistory https://www.youtube.com/watch?v=bOGSRJ1_7DI …#datavizpic.twitter.com/yRXbLb4Xt8
-
Some nice examples in this article, relating to what some call "Gates' Law": https://fs.blog/2019/05/gates-law/ … (img source: https://www.azquotes.com/quote/422146 )pic.twitter.com/NX58VgxUOx
Prikaži ovu nit -
Have you read this paper? One of our lab's canonical references for thinking about the relationship between private and public annotations/notes: https://www.microsoft.com/en-us/research/wp-content/uploads/2004/06/p112-marshall.pdf …pic.twitter.com/OcIviyPUPF
-
2: This is an old (but gold) book chapter that examines types and criteria for synthesis. The authors (Strike and Posner) are well known in ed / psych circles for their work on conceptual change: https://philpapers.org/rec/STRTOS-2 pic.twitter.com/YXpqByjyJ6
-
@MarkWhiting@brimci and I (with@imjuhokim at some point) were working on something like this for crowd civic systems! Some prior art: https://arxiv.org/ftp/arxiv/papers/1302/1302.5177.pdf … https://onlinelibrary.wiley.com/doi/full/10.1002/poi3.95?casa_token=ItLz8_aStTUAAAAA%3APnywxExZedIvaenKqUBd4OEcJTVnNpBTYxXE_FLqQMX26XsfIHdHnDStd2iRVRs6ThZhRYZhIyaDinDT …pic.twitter.com/T2437N8wNU
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.