Internet Industry Search Engine

Protecting Artificial Intelligence from Itself

Source: infoq.com

Jan 22, 2019 7:03 AM 1+ mon ago

Applications using artificial intelligence can be fooled by adversarial examples, creating confusion in the model decisions. Input sanitization can help by filtering out improbable inputs before they are given to the model, argued Katharine Jarmul at Goto Berlin 2018. We need to start thinking of the models and the training data we put into them as potential security breaches, she said.Read more.

Search Builder

(Click to add to search box)
Search Suggestions Level 1
training data  Katharine Jarmul  security breaches  model decision process  Professor Reza Shokri  place data privacy  Big Data practices  Deep Learning Blindspots  Input sanitization  model decisions  data security concerns  input sanitization  biometric data  extraction methods  scalable approach  production systems  Basic Countermeasures  data use  security risks  ML models  pseudonymization process  privacy concerns  model success  evaluation criteria  security measures  model API  data point  increase uncertainty  data science  company KIProtect  Protecting data  training dataset  dimensionality reduction  data algorithms  AI winter  data risks  use cases  MLaaS companies  Machine Learning  data privacy  AI applications  software development  use quotes  data scientist  Methods Protecting  Katharina Jarmul  
Search Suggestions Level 2
Jobs Protecting Humanity  scratch Billionaires Shouldn  Lester Earnest  intelligence makers  World Better  video games  
Search Suggestions Level 3
security teams  Electronic Frontier Foundation  Jiang designs algorithms  shouldn t laugh  probability vector p  t Sundar Pichai  UNC Eshelman School  drug candidates  web application layer  threat vulnerability management  cyber defense teams  intrusion detection systems  video games  network traffic  AI principles  safety profiles  arms race  drug molecules  ReLeaSE method  grocery store  opponents bases  gaming genre  player controls  door B  algorithm Jiang  design molecules  player moves  Mariya Popova  Olexandr Isayev  video game  Alexander Tropsha  chemical structures  algorithm analyses  computer program  World Better  chemical entities  decision strategies  Daniel Jiang  assistant professor  Swanson School  Perfecting algorithms  scientist system  test models  testing environments  simulations need  Historical data  AI systems  ethics boards  drug candidate  Orwellian levels  intelligence experts  Tomas Mikolov  research scientist  Facebook AI  HLAI conference  gameplay decisions  intelligence system  future disasters  AI entity  task humans  Structural Evolution  change course  RL algorithm  right education  security data  security environment  patch management  Training step  Reinforcement Learning  phishing attacks  network layer  use AI  cyber defense  way machine  data modification  cyber experts  security team  Russian actors  threat detection  defense teams  Generator step  intelligence capabilities  Security AI  North Carolina  Last Update  science fiction  weaponized technology  everything right  Cyber Defense  software race  AI technologies  cybersecurity landscape  model outputs  Sundar Pichai  Last year  Washington Post  cyber threats  software product  

**Content contained on this site is provided on an “as is” basis. 4Internet, LLC makes no commitments regarding the content and does not review it, so don't assume that it's been reviewed. What you see here may not be accurate and should not be relied upon. The content does not necessarily represent the views and opinions of 4Internet, LLC. You use this service and everything you see here at your own risk. Content displayed may be subject to copyright. Content is removed on a case by case basis. To request that content be removed, contact us using the following form: Contact Us.