| Topic List |
Page List:
1 |
|---|---|
|
FLUFFYGERM 08/13/18 12:55:00 PM #1: |
https://www.cnet.com/uk/news/google-deepminds-ai-can-now-detect-over-50-sight-threatening-eye-conditions/
This is incredibly exciting. @Asherlee10 @Antifar This is another example of what I mentioned a while back about how healthcare will become much more effective and affordable. Imagine a very near future when this technology is freely available on your smartphone and you can run these scans on yourself from the comfort of your own home. And then imagine a near future where medication for a shit ton of eye diseases can then be shipped to your house via an Amazon drone in a fraction of the time and cost you'd normally have to incur. We're going to get there soon. Couple this with advancements in genomics and gene editing technologies like CRISPR and we live in some fucking exciting times. --- Do good. Eat communists. ... Copied to Clipboard!
|
|
FLUFFYGERM 08/13/18 3:03:17 PM #2: |
bump
--- Do good. Eat communists. ... Copied to Clipboard!
|
|
s0nicfan 08/13/18 3:06:32 PM #3: |
Deepmind is both fascinating and terrifying.
--- "History Is Much Like An Endless Waltz. The Three Beats Of War, Peace And Revolution Continue On Forever." - Gundam Wing: Endless Waltz ... Copied to Clipboard!
|
|
Muffinz0rz 08/13/18 3:13:29 PM #4: |
Jeez, even doctors will be replaced with machines soon
--- Not removing this until Pat Benatar is in Super Smash Bros. (Started 8/31/2010) 2018 NFLB Summersim (7-8): https://imgur.com/7cjNtgQ ... Copied to Clipboard!
|
|
COVxy 08/13/18 3:15:51 PM #5: |
My guess is that there's going to be a crash soon with throwing DCNN's at problems and using inbred superstitious training procedures until you get a result is going to stop being impressive or useful. Not my field though, so who knows.
--- =E[(x-E[x])(y-E[y])] ... Copied to Clipboard!
|
|
s0nicfan 08/13/18 3:17:41 PM #6: |
COVxy posted...
My guess is that there's going to be a crash soon with throwing DCNN's at problems and using inbred superstitious training procedures until you get a result is going to stop being impressive or useful. Oh definitely. Not only is there the issue of requiring an obscene amount of training data, but onboard processing and storage is prohibitive for most devices. Plus unless you're really, really good at sensitivity analysis, you'll get a lot of correlations that are meaningless being baked into the NN, which is one of the reason why "big data" hasn't taken off as quickly as people thought. More fundamentally, a lot of people are using them without understanding the underlying mechanics, and they're proving to be incredibly easy to spoof which is a security/egal nightmare. --- "History Is Much Like An Endless Waltz. The Three Beats Of War, Peace And Revolution Continue On Forever." - Gundam Wing: Endless Waltz ... Copied to Clipboard!
|
| Topic List |
Page List:
1 |