science   395308

« earlier    

We’ve studied gender and STEM for 25 years. The science doesn’t support the Google memo. - Recode
"We have been researching issues of gender and STEM for more than 25 years, and we can say flatly that there is no evidence that women’s biology makes them incapable of performing at the highest levels in any STEM fields."
damore  google  diversity  feminism  work  recode  science  2017 
yesterday by handcoding
5 Tips To Survive High School | Best Images Collections HD For Gadget windows Mac Android
5 Ideas To Endure Significant Faculty We’ve bought some tips to assist you survive higher university. GMM#510! Fantastic Legendary A lot more: Ever have a #VacayGoneCrayCray? Go to to enter for a chance to get $10k towards a journey redo, courtesy of Decision Inns. SUBSCRIBE for each day episodes: http://little **** Past […]
IFTTT  WordPress  Tips  advice  booger  choice  hotels  class  high  school  jet  ski  marsh  olympiad  popularity  radar  rollerbl...  science  swamp  Teen  Teenager  Water 
yesterday by wotek
Nautilus | Science Connected
Nautilus is a different kind of science magazine. It delivers deep, undiluted, narrative storytelling to bring science into the largest and most important conversations we are having today. It challenges the reader to consider the connecting tissue that runs through the sciences and connects them to philosophy, culture and art.
culture  magazine  science 
yesterday by kdredington
That Alien Message - Less Wrong
Followup to:  Einstein's Speed
Imagine a world much like this one, in which, thanks to gene-selection technologies, the average IQ is 140 (on our scale).  Poten
science  ai  singularity  scifi  bayes  math  complexity 
yesterday by geetarista
Detecting Tanks
There's a story that's passed around to illustrate the ways machine learning can pick up on features in your dataset that you didn't expect, and probably gained the most exposure through Yudkowsky using it in "Artificial Intelligence as a Positive and Negative Factor in Global Risk" (pdf, 2008):

Once upon a time, the US Army wanted to use neural networks to automatically detect camouflaged enemy tanks. The researchers trained a neural net on 50 photos of camouflaged tanks in trees, and 50 photos of trees without tanks. Using standard techniques for supervised learning, the researchers trained the neural network to a weighting that correctly loaded the training set—output "yes" for the 50 photos of camouflaged tanks, and output "no" for the 50 photos of forest. This did not ensure, or even imply, that new examples would be classified correctly. The neural network might have "learned" 100 special cases that would not generalize to any new problem. Wisely, the researchers had originally taken 200 photos, 100 photos of tanks and 100 photos of trees. They had used only 50 of each for the training set. The researchers ran the neural network on the remaining 100 photos, and without further training the neural network classified all remaining photos correctly. Success confirmed! The researchers handed the finished work to the Pentagon, which soon handed it back, complaining that in their own tests the neural network did no better than chance at discriminating photos.
It turned out that in the researchers' dataset, photos of camouflaged tanks had been taken on cloudy days, while photos of plain forest had been taken on sunny days. The neural network had learned to distinguish cloudy days from sunny days, instead of distinguishing camouflaged tanks from empty forest.
bias  machinelearning  ai  tanks  science  datascience  data 
yesterday by timcowlishaw

« earlier    

related tags

+  /sci/  2017  4  advice  aggregator  ahhhhhh  ai  aliens  analytics  and  animals  anode  article  asteroids  astronomy  aww  battery  bayes  bbc  bbc4  bias  bibliography  bigdata  biologie  biology  bird  black  blackdeath  body  booger  books  broken  bullshit  chemistry  choice  class  climate  clour  colour  comic  complexity  computer_science  course  crick  crisis  culture  culture_of_online_life  cute  daily  damore  data  dataanalysis  datascience  datascientist  diet  dilute  dilution  dinosaurs  discover  diversity  documentary  dog  drake_equation  earth  ebooks  eclipse  eclipse2017  eclipseselfie  economics  economy  education  energy  engineering  enseignement  environment  esprit.libre  essay  europe  exercise  experiment  explore  eyes  feminism  fermi_paradox  film  fish  fishing  flavor  food  four  francis  galileo  gender  general  geo  geographic  go  golang  google  googlemanifest  googlememo  greatness  health  high  hilarious  history  hok  hotels  human.research  ifttt  image  institute  internet  irrationality  jet  kids  language  latex  learning  lecture  libre  life-science  light  london  machinelearning  magazine  manufacturing  maps  margo  markdown  marsh  materials  math  mathematica  media  meditation  metal  meti  might_write  movie  mysterious  nanosheet  nasa  nat  natgeo  national  nature  news  notebooks  numpy  of  olympiad  online  openscience  oxide  pandas  pandoc  perl  personalhistory  philosophy  photo  photography  physics  plaintext  planets  politics  poll  popularity  production  programming  python  quiz  r  radar  recode  reference  replication  research  resources  rollerbl...  s  sagan  school  science-communication  scientists  scifi  scpectrum  self-correcting  seti  sexism  share_with_kids  shedding  short  shotoniphone  showcase  simulation  singularity  ski  sleep  social  solareclipse  solarelipse2017  space  spaceflight  specialization  spectrum  startrek  stem  sun  survival  swamp  talk  tanks  taste  teaching  tech  technology  teen  teenager  telescopes  textbooks  tips  toread  trick  ucberkeley  video  videos  vimeo  vision  visualization  voyager  vr  water  whiskey  whisky  white  wiki  wildlife  wolfram  wordpress  work  worldbuilding  writing 

Copy this bookmark: