Interesting analysis suggesting that the reason for the disappointing performance of many modern CNN architectures is that their depthwise convolutions are memory-bound. https://t.co/C55Q6tv3LN pic.twitter.com/nXafyOseH3
— Jeremy Howard (@jeremyphoward) January 24, 2020
Soros on Science vs Alchemy:
— hardmaru (@hardmaru) January 24, 2020
“Scientific method seeks to understand things as they are, while alchemy seeks to bring about a desired state of affairs. To put it another way, the primary objective of science is truth—that of alchemy, operational success.”https://t.co/KvfMYYvM0Q
Good read on engineering career growth and effective management. https://t.co/d7mjuA4ZVU
— Xavier Amatriain (@xamat) January 24, 2020
Sumo wrestlers and basketball players ⛹️♀️ may be statistically indistinguishable (p=0.1) if you compare their weights or height, but multivariable approach (height&weight) has >90% classification accuracy. This is an essential lesson for psychiatric neuroimaging! @INM7_ISN @mrccbu pic.twitter.com/iCUShT7NFK
— Dr Camilla Nord (@camillalnord) January 23, 2020
Interactive demo to see how IEEE floats are representedhttps://t.co/N6WhN5DTPB pic.twitter.com/XR6vAm18vj
— hardmaru (@hardmaru) January 23, 2020
A good way to be unsure about something is to ask for one more opinion. More opinions often lead to indecision, not clarity. If necessary, ask for a few, add your own, make a call, and move on. Nearly all decisions are temporary, but stalling is permanent time lost.
— Jason Fried (@jasonfried) January 22, 2020
Less Software:
— Jason Fried (@jasonfried) January 22, 2020
You’d think that twice as much code would make your software only twice as complex. But actually, each time you increase the amount of code, your software grows exponentially more complicated.https://t.co/fOZn3S6gAZ
Actionable items for ML researchers re carbon emissions:
— Rachel Thomas (@math_rachel) January 21, 2020
- Quantify emissions
- Choose cloud providers & data center location wisely
- Reduce wasted resources (efficient algos, good tests, early debugging)
- Choose more efficient hardwarehttps://t.co/smYbaYVd57 @alex_lacoste_
Unpopular opinion: The biggest weakness of books is that too few of them are actually read. Therefore whatever people do to make them more likely to read them (cut apart, write in, convert to audio etc) is good. https://t.co/NGf4ik7LuS
— Chris Albon (@chrisalbon) January 21, 2020
4 Principles for Responsible Government Use of Technology
— Rachel Thomas (@math_rachel) January 21, 2020
- Listen to local communities
- Beware how NDAs obscure public process & law
- Security is not the same as safety
- Policy decisions should not be outsourced as design decisionshttps://t.co/lU6JsFxtAb #TechPolicyCADE
Sundar Pichai, CEO of Google, calls for sensible regulation of AI.
— Yann LeCun (@ylecun) January 21, 2020
I agree.
"Companies such as ours cannot just build promising new technology and let market forces decide how it will be used. It is equally... https://t.co/DPRGRVZ4ox
I'm going to create a twitter bot that recognizes whenever someone writes the hacky argument that "the government can't innovate" and then copies a link to the seminal '98 deep learning paper on recognizing handwritten digits for the US Postal Service https://t.co/nOaFyWF0sQ
— Jonathan Hersh (@DogmaticPrior) January 20, 2020
