inference   1623

« earlier    

[1901.01624] Composite optimization for robust blind deconvolution
The blind deconvolution problem seeks to recover a pair of vectors from a set of rank one bilinear measurements. We consider a natural nonsmooth formulation of the problem and show that under standard statistical assumptions, its moduli of weak convexity, sharpness, and Lipschitz continuity are all dimension independent. This phenomenon persists even when up to half of the measurements are corrupted by noise. Consequently, standard algorithms, such as the subgradient and prox-linear methods, converge at a rapid dimension-independent rate when initialized within constant relative error of the solution. We then complete the paper with a new initialization strategy, complementing the local search algorithms. The initialization procedure is both provably efficient and robust to outlying measurements. Numerical experiments, on both simulated and real data, illustrate the developed theory and methods.
approximation  inverse-problems  statistics  inference  algorithms  numerical-methods 
16 days ago by Vaguery
[1712.07381] Extreme Value Analysis Without the Largest Values: What Can Be Done?
In this paper we are concerned with the analysis of heavy-tailed data when a portion of the extreme values is unavailable. This research was motivated by an analysis of the degree distributions in a large social network. The degree distributions of such networks tend to have power law behavior in the tails. We focus on the Hill estimator, which plays a starring role in heavy-tailed modeling. The Hill estimator for this data exhibited a smooth and increasing "sample path" as a function of the number of upper order statistics used in constructing the estimator. This behavior became more apparent as we artificially removed more of the upper order statistics. Building on this observation we introduce a new version of the Hill estimator. It is a function of the number of the upper order statistics used in the estimation, but also depends on the number of unavailable extreme values. We establish functional convergence of the normalized Hill estimator to a Gaussian process. An estimation procedure is developed based on the limit theory to estimate the number of missing extremes and extreme value parameters including the tail index and the bias of Hill's estimator. We illustrate how this approach works in both simulations and real data examples.
statistics  extreme-values  rather-interesting  algorithms  estimation  inference  data-analysis  nudge-targets  consider:looking-to-see  consider:performance-measures 
16 days ago by Vaguery
Microsoft/onnxruntime: ONNX Runtime: cross-platform, high performance scoring engine for ML models
ONNX Runtime: cross-platform, high performance scoring engine for ML models - Microsoft/onnxruntime
github  inference  deep_learning 
5 weeks ago by lxp121
The Friendship Paradox and Systematic Biases in Perceptions and Social Norms | Journal of Political Economy: Vol 127, No 2
"The “friendship paradox” (first noted by Feld in 1991) refers to the fact that, on average, people have strictly fewer friends than their friends have. I show that this oversampling of more popular people can lead people to perceive more engagement than exists in the overall population. This feeds back to amplify engagement in behaviors that involve complementarities. Also, people with the greatest proclivity for a behavior choose to interact the most, leading to further feedback and amplification. These results are consistent with studies finding overestimation of peer consumption of alcohol, cigarettes, and drugs and with resulting high levels of drug and alcohol consumption."

So this is about misinference from networks then?
matthew.jackson  networks  inference 
5 weeks ago by MarcK
Introduction to Gravity Models of Migration & Trade | Programming Historian
> A gravity model's goal is to tell the user: given a number of influencing forces (distance, cost of living) affecting migration or movement of a large number of entities of the same type (people, coffee beans, widgets) between a set number of points (39 counties and London or Colombia and various countries), the model can suggest the most probable distribution of those people, coffee beans, or widgets. It operates on the principle that if you know the volume of movement, and you know the factors influencing it, you can predict with reasonable accuracy the outcome of even complex movement within a confined system.
history  modeling  statistical-modeling  inference  migration 
6 weeks ago by tarakc02
problog / DeepProbLog — Bitbucket
DeepProbLog is an extension of [ProbLog]( that integrates Probabilistic Logic Programming with Deep Learning
logic  probabilistic  programming  prolog  knowledgegraph  inference  deeplearning 
6 weeks ago by tobym
Introduction. — ProbLog: Probabilistic Programming
Probabilistic logic programs are logic programs in which some of the facts are annotated with probabilities.

ProbLog is a tool that allows you to intuitively build programs that do not only encode complex interactions between a large sets of heterogenous components but also the inherent uncertainties that are present in real-life situations
logic  probabilistic  programming  prolog  knowledgegraph  inference 
6 weeks ago by tobym
Common statistical tests are linear models (or: how to teach stats)
Most of the common statistical models (t-test, correlation, ANOVA; chi-square, etc.) are special cases of linear models or a very close approximation. This beautiful simplicity means that there is less to learn. In particular, it all comes down to y=a⋅x+b which most students know from highschool. Unfortunately, stats intro courses are usually taught as if each test is an independent tool, needlessly making life more complicated for students and teachers alike.
statistics  linear-models  inference  statistical-tests 
6 weeks ago by tarakc02
andrewheiss/diff-means-half-dozen-ways: Run standard t-tests, simulations, and Bayesian difference in means tests with R and Stan
Run standard t-tests, simulations, and Bayesian difference in means tests with R and Stan - andrewheiss/diff-means-half-dozen-ways
statistics  Data_Science  inference  github  CDS_101  Pedagogical_Resources 
6 weeks ago by jkglasbrenner

« earlier    

related tags

1965  1970s  1990s  2018  abduction  abstraction  accelerator  accuracy  active  adv300  advertising  agent  ai  algorithms  alien-character  analogy  analysis  analytical-holistic  analytics  anglo  anglosphere  anomaly  anonymity  anthropomorphizing  api  apollonian-dionysian  apple  approximation  art  article  artificial+intelligence  artificial  artificialintelligence  arxiv  attention  aws  axioms  bayes  bayesian  behavior  being-becoming  bias  biases  bibliography  bio  blackboxai  blood  book  books  boolean  boundary  brain-scan  brain  c#  capta  capture  category  causal-inference  causal  causality  causation  cds_101  chart  choice  christianity  clipper  clippings  clips  coarse-fine  cocktail  code  cog-psych  commentary  commodityquantity  communication  comparison  competition  complex-systems  composition-decomposition  computation  computational-geometry  computational-statistics  computational  computer-science  computer  computers  computing  concept  conceptual-vocab  concrete  confidence  consider:algorithms  consider:gp  consider:looking-to-see  consider:performance-measures  consilience  context  contradiction  correlation  cosma-shalizi  cost-benefit  counterfactual  course  creative  cug-tests  curiosity  current  cyc  data-analysis  data-balancing  data-cleaning  data  data_science  dataframe  decision  deep-learning  deep-materialism  deep  deep_learning  deeplearning  definition  density  deploy  deployment  design  desire  detail-architecture  detection  development  dikw  dimension  direction  discussion  disease  do  docker  dsreads  echo  edge  egalitarianism-hierarchy  elastic  empirical  ends-means  energy  epgm  epistemic  error  essay  essence-existence  estimation  europe  evidence  examples  exegesis-hermeneutics  expert-experience  expert  expertsystems  explainableai  explainer  explanans  extreme-values  eyeriss  facebook  fastai  feature-extraction  fiction  filetype  finance  fmri  forgotten  framework  free  frequentist  friston  functional-languages  future  fuzzy  fuzzyconcept  fuzzysemantics  gcp  genetic-programming  georgelakoff  gigerenzer  github  given  gnon  golang  google  gpu  grammar  graphs  ground-up  hanson  hardware  heuristic  heuristics  history  hmm  homo-hetero  homo-heuristicus  hosting  howto  hsu  huge-data-the-biggest  human  ideas  identity  image  imaging  incomplete  increase-decrease  indefinite  induction  influence  information  innovation  int  intel  intelligence  interpretation  interventions  intricacy  introduction  intuition  inverse-problems  ios  iot  iteration-recursion  jan_koenderink  japan  jit  judea  jupyter  jvm  kdd  knowledge  knowledgegraph  labor  lambda  language  law  learning  liberty  library  linear-models  links  loftizadeh  logic  looking-to-see  low  lstm  machine-learning  machine  machinelearning  macos  magnet  mapping  markov  maryloujepsen  matching  math  matrices  matthew.jackson  meaning  measure  medicine  menoh  meta:math  metabuch  metal  methodology  microsoft  migration  mind  missing-data  mkl  ml  mobile  model  modeling  models  montecarlo  motivation  mps  mtg  multi  n-factor  nature  near-far  nervana  nested  net  network-analysis  network-regression  network-structure  network  networks  neural  neuralnetworks  neuro-nitgrit  neuro  neuron  neurons  neuroscience  new-religion  nibble  nlp  noise  nonexpert  notebook  novelty  now.js  nudge-targets  numerical-methods  nvidia  occam  occident  of  omission  open-closed  opengm  opensource  openvino  opsec  optimism  oracle  order-disorder  orders  org:theos  oxygen  p-value  pandas  paper  papers  parsimony  partial  pdf  pedagogical_resources  peral  perception  pessimism  petabyte  pgmpy  philosophy  pi  plane-geometry  policy  politics  power  precision  prediction  predictive-processing  presentation  principle  privacy  probabilistic  probability  problem-solving  processing  product  production  programming-language  programming  prolog  proof  protestant-catholic  psychiatry  psychology  python  pytorch  qap  quantization  questions  r-project  random-graphs  raspberry  raspberrypi  rather-interesting  ratty  raw  reaction  realness  reason  reasoning  reduction  reference  reflection  regression  regulation  reinforcement  religion  representation  req  research-article  research  resolution  responsibility  review  rhetoric  rights  rigor  scala  scan  science  scipy  scitariat  script  sdes  security  seq2seq  sequents  serverless  serving  sight  significance  similarity  skeleton  sna  software  softwre  spatial  speculation  spengler  standards  stanford  statistical-inference  statistical-modeling  statistical-tests  statistics  stats  stick  stress  stripe  structuralcausalmodels  structure  subjective-objective  summary  surveillance  syllabus  symmetry  taken  tdd  tech  technocracy  technology  telos-atelos  tensorflow  test  tests  the-great-west-whale  thenet  theory-of-mind  theory-practice  theory  theos  thick-thin  things  thinking  threat-modeling  threshold  time  to-read  to-understand  to-write-about  tpu  trace  training-data  training  tricki  truth  turing  tutorial  type  uncertainty  unfixed  unsharp  untether  usability  usb  vague  values  variable  variational  video  vino  vision  visuo  volo-avolo  voxel  vpu  whole-partial-many  wiki  wire-guided  wisdon  work  wtatic  yoga 

Copy this bookmark: