timcowlishaw + machinelearning   335

Electronic Pop for the Surveillance Era | The New Yorker
The popular fear of algorithms reflects the anxiety that our lives will simply become patterned according to a program—that our autonomy will evaporate as computers tell us what songs we will like, when we need to buy more toilet paper, or what move we should make in a chess game.
music  ai  machinelearning  technology  culture  hollyherndon 
11 weeks ago by timcowlishaw
Can emotion-regulating tech translate across cultures? | Aeon Essays
Gadgets and algorithms give a robotic materiality to what the ancient Greeks called doxa: ‘the common opinion, commonsense repeated over and over, a Medusa that petrifies anyone who watches it,’ as the cultural theorist Roland Barthes defined the term in 1975.
culture  society  mentalhealth  depression  algorithms  technology  emotions  CUI  VUI  ai  machinelearning  therapy  psychology 
july 2018 by timcowlishaw
Want Less-Biased Decisions? Use Algorithms.
A not-so-hidden secret behind the algorithms mentioned above is that they actually are biased. But the humans they are replacing are significantly more biased. After all, where do institutional biases come from if not the humans who have traditionally been in charge?
bias  algorithms  accountability  fatml  ml  machinelearning  psi  society  politics  fat  fairness  transparency 
july 2018 by timcowlishaw
ConceptNet Numberbatch 17.04: better, less-stereotyped word vectors – ConceptNet blog
I wonder if the excessive focus on Mikolov et al.’s analogy evaluation has exacerbated the problem. When a system is asked repeatedly to make analogies of the form male word : female word :: other male word : other female word, and it’s evaluated on this and its knowledge of geography and not much else, is it any surprise that we end up with systems that amplify stereotypes that distinguish women and men?
wordembeddings  debiasing  nlp  word2vec  machinelearning  machine  learning  computerscience  bias  racism  sexism 
october 2017 by timcowlishaw
Facebook and Google, Show Us Your Ad Data - Bloomberg
"And just as my friends and I used to socialize at the mall when we were teenagers, nowadays we socialize online in commercial spaces like Facebook and Twitter."
parp  psi  algorithms  machinelearning  fatml  transparency  politics  technology  society 
september 2017 by timcowlishaw
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 
august 2017 by timcowlishaw
Apple's kangaroo cookie robot
The browser has to be part of the answer. If the browser does its job, as Safari is doing, it can play a vital role in re-connecting users with legit advertising—just as users have come to trust legit email newsletters now that they have effective spam filters.

Safari’s Intelligent Tracking Prevention is not the final answer any more than Paul Graham’s “A plan for spam” was the final spam filter. Adtech will evade protection tools just as spammers did, and protection will have to keep getting better. But at least now we can finally say debate over, game on.
tracking  advertising  internet  politics  usability  privacy  machinelearning  society  psi  parp 
june 2017 by timcowlishaw
« earlier      
per page:    204080120160

related tags

ablative  ablativeanalysis  academia  acapella  accountability  accuracy  acountability  activation  active  activelearning  actors  adadelta  adagrad  adaptive  adversarial  advertising  advice  aesthetics  affect  agency  agglomerative  aggregation  ai  AIA  akka  alexa  algebra  algebraic  algorithm  algorithmic  algorithms  allocation  amazon  ami  analogy  analysis  analytics  anaomalydetection  and  annotation  anthroprocentricism  architecture  arendt  arpack  art  artificialintelligence  assistant  associations  assumptions  atomised  AUC  audio  austerity  authorship  authorshop  autocorrelation  autoencoder  automaticstatistician  autonomousvehicle  aws  babbage  backprop  backpropagation  bagofwords  bandit  bayes  bayesdb  bayesian  bayesiannonparametrics  bbc  bbva  beauty  belamy  bellmansequation  berger  bias  biases  biasvariancetradeoff  bigdata  blackbox  blei  blog  boilerpipe  boilerplate  boilerplatedetection  book  books  bootstrap  bot  bsp  business  c  caffe  calibration  captions  cardinality  cart  cartography  categorical  categoricalvariable  categorisation  category  cf  cheatsheet  chi2  chineserestaurantprocess  chomsky  cities  classfication  classification  classifier  cleaning  cloud  cloudera  cluster  clustering  cnns  colinearity  collaborative  collaborativefiltering  colt  command  complexity  component  components  composition  compression  computability  computation  computational  computationallinguistics  computer  computermusic  computers  computerscience  computervision  computing  concurrency  conditional  conditionalrandomfields  confidenceinterval  conjugate  consent  containers  contentbasedrecommendation  contentextraction  context2vec  control  conversational  convnets  convolutional  convolutionalnetworks  convolutionalneuralnetworks  corpus  correlation  cosine  cost  counterfactual  creativity  credibility  CRF  CRFs  criticism  cross  crossentropy  crosses  crossvalidation  crowdsourcing  cs  cs229  cui  CUIs  culture  data  data-analysis  database  datamining  datascience  dataset  datasets  datastructures  dataviz  davidblei  db  deaplearning  debiasing  debugging  decision  decisionboundary  decisiontheory  decisiontree  decomposition  deep  deep-learning  deepdream  deepfakes  deeplearning  deepnetwork  deepnetworks  deep_learning  demiotics  demo  depression  descent  design  development  dfas  diagnosis  dicesorenson  difference  digitalhumanities  dimensionality  dimensionalityreduction  dirichlet  dirichletprocess  discovery  discriminant  discriminantanalysis  discrimination  distributed  distributedsystems  distribution  distributionalsemantics  diversity  dnns  docker  dp-means  dpmm  drawing  drum  drumming  dsp  dummy  dummyvariable  dummyvariabletrap  dunningkrugerish  duplication  dynamics  dzubia  ec2  edgedetection  education  edwarddebelamy  effects  efficiency  eigenfaces  elmo  em  embedding  embeddings  emotions  encoding  engineering  enlightenment  ensemble  ensemblelearning  ensemblemethods  ensembles  entities  entityextraction  entitylinking  epsilongreedy  erlang  error  erroranalysis  essentia  estimation  ethics  evaluastion  evaluation  example  expectation  expectationmaximisation  expectationmaximization  expectedloss  expectedutility  experiment  explanation  explanations  f#  facerecognition  fact-checking  factorgraph  factorie  factorization  factorizationmodels  fairness  fake  fakenews  fascism  fat  fatml  feature  featuredetection  featureengineering  featureextraction  featurelearning  featurepreparation  featurerepresentation  features  featureselection  federatedlearning  federeated  feedforward  ffi  fields  film  filter  filterbubble  filtering  finetuning  fingerprinting  flappybird  flickr  football  forecasting  frequentist  fullyconnected  fullyconnectedlayer  function  functional  future  fuzzycmeans  games  gamma  gan  gans  gaussian  gaussianmixture  gender  generallinearmodels  generation  generative  generativeart  generativity  genre  gensim  geodata  geography  german  ghahramani  gibbs  gibbssampling  glms  glove  go  google  gpt2  gpu  gpufromhost  gradient  gradientcheck  gradientdescent  graph  graphdispersion  graphicalmodels  graphs  groups  habits  hack  hadoop  hardness  hash  hashing  hashkernel  haskell  hci  hdp  heirarchical  heuristic  hidden  hierarchical  hierarchicaldirichletprocess  history  hlearn  hmms  hollyherndon  html  humanities  hype  hyperas  hyperopt  hyperparameters  hyperparemeter  ia  iat  identity  image  imageprocessing  imaging  imbalancedclasses  iml  improvisation  inceptionism  indianbuffetprocess  inference  information  informationretrieval  informationrtetrieval  informationscience  intelligence  intelligenceaugmentation  interaction  interactive  interactivemachinelearning  interface  internet  interpretability  interpretation  interview  introduction  invention  ir  isotonicregression  itembased  jaccard  jackknife  java  Johnson-Lindenstrauss  journalism  jruby  json  justice  jvm  k  k-means  kaggle  kalmanfilter  kdnuggets  keras  kernel  kerneldensityfunctions  kitsch  kmeans  kohonen  kohonennetworks  labelprop  labour  language  languagemodel  languages  latent  latentdirichlectallocation  latentdirichletallocation  latentfeaturespace  latentsemanticanalysis  law  layer  lda  ldp  learn  learning  learningtheory  learningtorank  leaveoneout  lebesque  legal  lgbtq  libby  libraries  library  libsvm  liftedinference  linear  linearalgebra  linearregression  linguistics  linux  list  literature  locallyconnected  locallyconnectedlayer  location  logic  logistic  logisticregression  logistiregression  loo  lopq  loss  lsa  lsi  lstm  lyric  lyrics  machine  machine-learning  machinelearning  machine_learning  mahout  mallet  managerialism  manifolds  map  mapreduce  maps  markov  markovchainmontecarlo  markovchains  markovdecisionprocess  math  mathematics  maths  matrix  matrixfactorization  maximization  maximumlikelihood  mcmc  mdp  measurement  media  medical  memory  mentalhealth  metal  metaphor  methods  metrics  mfcc  microscopes  midi  mining  mir  mistakes  mixturemodels  ml  mlclass  mle  mllib  moathematics  model  modelling  models  modelselection  modules  momentum  monoid  monoids  montecarlo  mood  moonstone  multicolinearity  music  musicology  naive  naivebayes  narrative  natural  naturallanguageprocessing  nca  ndcg  nearestnabour  neighbourhood  netflix  network  networks  neural  neuralnet  neuralnets  neuralnetwork  neuralnetworks  neurology  news  nlp  nltk  noise  nonparametric  nonparametrics  normalisation  notes  noughts  numerical  numpy  objective  objectivefunction  obvious  ols  onehot  online  onlinelearning  ontology  operations  optimisation  optimization  organisations  outofcore  overfitting  paas  pac  pandas  parallel  parameter  parameterestimation  parameters  parp  pathregression  pattern  pca  perception  perceptron  performance  persistence  persiststence  phd  philosophy  physics  piano  pig  platttscaling  play  playfulness  pocket-read  poincare  politics  pooling  poolinglayer  pornography  power  powermethods  precision  predictin  prediction  predictive  preemption  pregel  principal  prior  privacy  probabilistic  probability  process  processing  product  productdesign  productmanagement  programming  projections  proof  prototype  psi  psychology  pybossa  pybrain  pylearn  pymc  pystruct  pything  python  qlearning  quantisation  r  race  racism  radialbasisfunction  ramnking  random  randomforest  randomsplit  rank  ranking  rant  rbf  realscienceofpop  real_libby  recall  receiveroperatingcharacteristic  recognition  recomendatersystems  recommandations  recommendation  recommendations  recommender  recsys  recurrent  recurrentneuralnetworks  redis  reduction  reference  regression  regular  regularisation  regularization  reinforcement  reinforcementlearning  report  research  resources  retrieval  reuters  rnn  rnns  robots  ROC  royalsociety  rsquared  rub  rubumanortalk  ruby  ruleofthumb  sampling  scala  scale  science  scienceofpop  scientific  scikit-learn  scikits  scikitslearn  sciruby  scotch  scraping  search  seldon  selection  selfdrivingcar  semantics  semiotics  semisupervised  sentiment  sentimentanalysis  sequence  sequences  serendip  serendipity  sexism  sexuality  sgd  shallowtextfeatures  shannon  signal  signalprocessing  similarity  simulation  singularvaluedecomposition  situationism  situationist  slowdown  smartspeaker  smoothing  socialism  socialmedia  socialscience  society  sociology  software  soms  sound  spark  sparse  sparsecoding  sparsenn  sparserandomprojections  speech  sports  spotify  stanford  statistical  statisticalrelationallearning  statistics  stats  statsmodels  stickbreakingprocess  stories  storylinedetection  stream  structuralism  structure  structured  structuredlearning  study  sunspring  support  supportvectormachines  surgery  surveilllance  svd  svm  synthesis  synthesiser  systemic  systems  tanimoto  tanks  taxonomy  technocracy  technocratic  technology  tensor  tensorflow  test  testing  text  textmining  tfidf  theano  theory  therapy  thoughtlessness  timeseries  time_series  tips  tokenization  tool  topic  topicmodel  topicmodelling  topicmodels  topological  topologicaldataanalysis  topology  torch  tracking  train  training  transcription  transferlearning  transparence  transparency  trap  tricks  trust  truth  truthiness  tsne  ttw  ttw19  tutorial  tv  tversky  twitter  uber  UCB  UI  uncanny  underfitting  universality  unsupervised  update  urbanism  usa  usability  userexperience  usersatisfaction  utility  ux  vae  validation  variable  variance  Variational  vctheory  vector  veracity  video  videoannotation  vision  visualisation  visualization  vui  weather  web  weights  weka  wela  whisky  windowing  wolfe  word  word2vec  wordembedding  wordembeddings  work  writing  yandex  yhat  yves  zlib  zoubin 

Copy this bookmark: