The importance of getting the backstory for weird findings in your data is worth the hassle of the phone call -- as described below, that's saying a lot for the importance of understanding your data! 🐻
— Caitlin Hudon👩🏼💻 (@beeonaposy) June 10, 2019
Great interview of @hadleywickham for @propublica: https://t.co/rnrNMdS2OP pic.twitter.com/J9GeFCFX1A
Ethics is being used by big tech as a smokescreen to delay responsibility and avoid liability. It’s not pegged to anything. It’s self-regulation by another name. - @mer__edith
— Matt Cagle (@Matt_Cagle) June 8, 2019
🙏🏼Preeeeach.🙏🏼#CityArtsLectures pic.twitter.com/lrKJuLx8aR
I think “even senior developers have to Google things!” is true and important but misses the point
— David Robinson (@drob) June 7, 2019
It’s like saying “even concert pianists have to practice!”
The lesson isn’t “We’re all faking it,” it’s “Searching/practicing is necessary to get better and that never stops”
If you write the same code 3 times you should write a function, but you should probably work with the same type of object about 30 times before you make it a class https://t.co/HgnMlWfdTi
— David Robinson (@drob) June 6, 2019
There were 46 papers submitted to the “dataset / software” category for #NeurIPS2019. 469 papers submitted to RL. Re-prioritization sorely needed in the field 🤦♀️🙅🏽♂️
— Eric Jang (@ericjang11) June 6, 2019
You don't need to know everything. You don't really need a formal background in this or that -- though it helps. You don't even need a PhD.
— François Chollet (@fchollet) June 5, 2019
You do, however, need to be constantly learning. Be curious. Read books. Don't be "too busy" to learn, or otherwise proud of your ignorance
When I first started analyzing data (in Python) I relied too much on object-oriented programming. If I was analyzing census data, I'd create State + County classes, with methods like get_population() / plot_income()
— David Robinson (@drob) June 4, 2019
You can burn 1000s of lines of code and feel totally productive
About 10,000 deep learning papers have been written about "hard-coding priors about a specific task into a NN architecture works better than a lack of prior" -- but they're typically being passed as "architecture XYZ offers superior performance for [overly generic task category]"
— François Chollet (@fchollet) June 1, 2019
@amcasari To job-seekers: Specifically ask the hiring manager, "are you hiring for experience or potential?" and save your time. ⏳
— Julia Silge (@juliasilge) June 1, 2019
Also, find a mentor/advocate to speak with about salary negotiation. 💰@AmeliaMN Hiring diverse teams is impactful/important but takes work.
One of the most annoying criticisms of @kaggle competitions that I come across is that "Kaggle is not like the real world." You know what is even less like the real world? Your college education. The only thing that is like the real world is real world, and only sometimes.
— Bojan Tunguz (@tunguz) May 31, 2019
How can we be good stewards of collaborative trust? Some thoughts on collaborations and credit issues:https://t.co/Or35qcCqPZ
— Chris Olah (@ch402) May 30, 2019
One of the best pieces of advice I ever got from a mentor was: “if you’ve never missed a flight, you’re spending too much time in airports.”
— Angela Bassa (@AngeBassa) May 29, 2019
Figure out what your own parameters are, and then allow yourself to risk missing a goal by cutting it too close sometimes.
