We didn't t claim it is novel because the change sounds incremental, but it is indeed a more elegant view of tree boosting in optimizing a single objective function. We also wrote a tutorial on this derivation inhttps://t.co/0x7acTD48F
— Tianqi Chen (@tqchenml) October 30, 2018
TIL XGBoost and LightGBM use newton steps instead of gradient steps, and I can't find an actual discussion of that. Also the source Chen and Guestrin cite seems not to do gradient boosting at all. I'm a bit confused.
— Andreas Mueller (@amuellerml) October 30, 2018
Posting this on here is of course a magnet for highly toxic replies, so I'm not doing it for fun. I think it's important to call a spade a spade. We should use the meaning of words as intended.
— François Chollet (@fchollet) October 30, 2018
Inspiring from James Dellinger, who shows how to create a bird classification model from scratch - and then explains how he very nearly quite deep learning before he started, but carried on regardlesshttps://t.co/rWK1o07UAI pic.twitter.com/3AyPQLY78o
— Jeremy Howard (@jeremyphoward) October 29, 2018
🎨 use outside themes, or roll your own…
— Mara Averick (@dataandme) October 29, 2018
💅 "RStudio IDE: Custom Theme Support" by Maria Semplehttps://t.co/1OFauQYp35 #rstats @rstudio
/* plus 🌙🦉looking fly as ever, 💕 @sarah_edo */ pic.twitter.com/sBkKbjL2g0
In the manga Fullmetal Alchemist, bad actors sometimes fuse beings into one, yielding a "chimera". Perhaps a rogue state alchemist fused @GaryMarcus with Bob Dole, explaining both the Senator's absence from the limelight and Gary's excessive use of the 3rd person. https://t.co/jjJASxZgSJ
— Zachary Lipton (@zacharylipton) October 29, 2018
"I can't believe you're worrying about something minor like issue X, when there are more serious issues like Y and Z." --> Often said by people who are doing nothing to address Y or Z.
— Rachel Thomas (@math_rachel) October 28, 2018
--> Often the folks being criticized are actively working on X, Y, and Z, all at once.
📈 Welp, @abresler (of {nbastatR} fame) discovered @jbkunst's {highcharter}.
— Mara Averick (@dataandme) October 28, 2018
🏀 So prepare to become a better GM than Bobby Marks, kids…
💅 https://t.co/kydGzFwY8h #rstats #dataviz #NBA pic.twitter.com/QKOQ5lJ2mC
This is amazing https://t.co/xRzBgQH0oE
— Peter Skomoroch (@peteskomoroch) October 27, 2018
Visualizing Science: Illustration and Beyond.
— Rachel Thomas (@math_rachel) October 26, 2018
This article is beautiful, interesting, & comprehensive. @sciam @ChristiansenJen https://t.co/0ZugLPMptb
Humans are hardwired to be upset by negative feedback. It is valuable to work on personal strategies to address this, but also, don't let people make you feel bad about letting comments from strangers bother you (because it's a natural reaction).
— Rachel Thomas (@math_rachel) October 26, 2018
"The secret to getting into the deep learning community is high quality blogging... Don’t just write something ok, either — take 3 or 4 full days on a post and try to make it as short and simple (yet complete) as possible." - @iamtrask https://t.co/nfTonejvhi
— Rachel Thomas (@math_rachel) October 26, 2018
