neural   3502

« earlier    

GAN Dissection / GAN Paint
"The #GANpaint app works by directly activating and deactivating sets of neurons in a deep network trained to generate images. Each button on the left ("door", "brick", etc) corresponds to a set of 20 neurons. The app demonstrates that, by learning to draw, the network also learns about objects such as trees and doors and rooftops. By switching neurons directly, you can observe the structure of the visual world that the network has learned to model. "
neural  AI  art 
9 days ago by magnusc
Neural network organises world into concepts like we do - Technology Review
"researchers began probing a GAN’s learning mechanics by feeding it various photos of scenery—trees, grass, buildings, and sky. They wanted to see whether it would learn to organize the pixels into sensible groups without being explicitly told how.

Stunningly, over time, it did. By turning “on” and “off” various “neurons” and asking the GAN to paint what it thought, the researchers found distinct neuron clusters that had learned to represent a tree, for example. Other clusters represented grass, while still others represented walls or doors. In other words, it had managed to group tree pixels with tree pixels and door pixels with door pixels regardless of how these objects changed color from photo to photo in the training set."

Includes a link to the GANpaint app which demonstrates the process
AI  neural  art 
9 days ago by magnusc
ONNX is a open format to represent deep learning models. With ONNX, AI developers can more easily move models between state-of-the-art tools and choose the combination that is best for them.
onnx  neural  network  ai  model  format  machine-learning  machine  learning  programming 
13 days ago by vicchow
Auto-Keras and AutoML: A Getting Started Guide - PyImageSearch
In this tutorial, you will learn how to use Auto-Keras, an open source alternative to Google’s AutoML, for automated machine learning and deep learning. The end goal of both Auto-Keras and AutoML is to reduce the barrier to entry to performing machine learning and deep learning through the use of automated Neural Architecture Search (NAS) algorithms. Auto-Keras and AutoML enable non-deep learning experts to train their own models with minimal domain knowledge of either deep learning or their actual data.
neural  neuralnetworks  learning  machinelearning  deeplearning  python 
14 days ago by cyberchucktx
BrainJS/brain.js: 🤖 Neural networks in JavaScript
🤖 Neural networks in JavaScript. Contribute to BrainJS/brain.js development by creating an account on GitHub.
ai  artificial  intelligence  javascript  neural  networks 
19 days ago by sistrall
Neural networks in JavaScript |
This course gives you a practical introduction to Brain.js, a popular JavaScript library for building neural networks in the browser. And since this is Scrimba, you'll be able to interact with the neural networks whenever you want.
via:plouf  Archive  course  neural  network  scrimba 
21 days ago by sudocurse
On the Origin of Circuits • Damn Interesting
finally found a source for that ghost story someone at media lab told me
narrative  history  ghosts  ai  algorithm  electronics  ee  fpga  neural 
22 days ago by akamediasystem

« earlier    

related tags

3d  acoustic  ai  algorithm  algorithms  analysis  archive  art  artificial  augmentation  autocomplete  autoland  backpropagation  bayesian  big  biological  book  bookreview  books  boosting  brain  browser  build  calculus  camera  cdn  classes  classification  cnn  code  coding  comic  compression  conference  conferences  consulting  control  convolution  convolutional  correlction  course  courses  creative  creativity  culture  darkglass  data  datascience  deep.dreram  deep  deeplearning  derivative  detection  differential  dl  dnn  drone  dynamics  education  ee  electronics  equation  erlang  error  europe  evaluation  exchange  flight  flow  formal  format  fortin  fpga  gan  generative  ghosts  go  golang  google  grammar  graph  guided  guitar  haptics  hardware  hentai  history  howto  humpback  ibm  ideas  ifttt  image  images  improvement  induction  insights  instruction  intelligence  ir  javascript  js  knitting  learn  learning  library  light  lightelligence  linux  machine-learning  machine  machinelearning  math  matrix  mit  ml  model  narrative  net  nets  network  networking  networks  neural-networks  neuraldsp  neuralnet  neuralnetworks  neurips  neuro  newspapers  nn  noise  onnx  opencv  opensource  paper  plugin  pocket  processor  productivity  program  programming  python  pytorch  quantization  quantize  reasoning  recognition  reddit  regularization  reinforcement  relief  research  rl  runtime  rust  science  scrimba  semantic  shell  simphony  software  spectral  squeeze  style  subroutine  swiss  symbolic  synaptic  synthesis  tacs  tensor  tensorflow  terrain  tinnitus  topology  toread  tpu  transfer  turing  tutorial  uav  untether  verification  video  videos  vision  visual  visualisation  visualization  vqa  whale 

Copy this bookmark: