Diese Präsentation wurde erfolgreich gemeldet.
Wir verwenden Ihre LinkedIn Profilangaben und Informationen zu Ihren Aktivitäten, um Anzeigen zu personalisieren und Ihnen relevantere Inhalte anzuzeigen. Sie können Ihre Anzeigeneinstellungen jederzeit ändern.

Machine intelligence

This talk is divided in 3 parts: inspiration to study; the business aspects of Artificial Intelligence; and its technical aspects, with a practical demo and technical explanations about Overfitting, NN, CNN, PCA and feature extraction. The code used during the demo will soon be on Github.

  • Loggen Sie sich ein, um Kommentare anzuzeigen.

  • Gehören Sie zu den Ersten, denen das gefällt!

Machine intelligence

  1. 1. Machine Intelligence and how it evolved
  2. 2. Wilder Rodrigues Software Engineer Crazy about A.I. And pretty much lots of other things As geek as it can get, chiefly about the X-Men Proud to be a family man and father of three @wilderrodrigues
  3. 3. One doesn’t need to be a subject matter expert to share knowledge. Whilst the one sharing must be open to criticism, the one listening is responsible to fact check, give feedback and spread what has been learned. The sharing process must be organic. Wilder Rodrigues
  4. 4. From 2015 to 2017 Machine Learning
  5. 5. What is it about?
  6. 6. Computing Machinery & Intelligence Original question: Can machines think? Alternative question: Are there imaginable digital computers which would do well in the imitation game? Alan Turing
  7. 7. A look at today’s A.I. Supervised Learning Spam classifiers Everything recognition Diseases diagnosis Speech-to-text Driving economics 38% of enterprises are already using A.I. $8 bilion market in 2016
  8. 8. A look at today’s A.I. Job Displacement Blue River Agriculture CV, ML and Robotics Goldman Sachs Trading 45 % of trading is done electronically
  9. 9. Machine Learning Arthur Samuel (1959) Machine Learning: Field of study that gives computers the ability to learn without being explicitly programmed. Tom Mitchell (1998) Well-posed Learning Problem: A computer program is said to learn from experience E with respect to some task T and some performance measure P, if its performance on T, as measured by P, improves with experience E. Classifying emails as spam or not spam. (T) Watching you label emails as spam or not spam. (E) The number (or fraction) of emails correctly classified as spam/not spam. (P)
  10. 10. A look into tomorrow’s A.I. [Hyped] Unsupervised Learning Mostly done on research still Economics 68% of enterprises using A.I. by 2018 300% increase in investment in 2017 More than $47 billion market by 2020
  11. 11. Defensive Barriers Data is the only barrier one can build! Community is open and algorithms pop up every day There is an open data movement. The idea is to speed up the A.I. evolution. “It’s not who has the best algorithm that wins. It’s who has the most data.” Banko en Brill, 2001
  12. 12. A.I. as a Product Anything a typical human can do within a second of thought, can probably be automated, now or soon, with A.I. Andrew Ng. Desire Feasibility
  13. 13. Demo
  14. 14. Feature Extraction Extract Convert Normalise Feed propagation Back propagation Initialise weights Implement forward propagation Implement cost functions Implement back propagation Overfitting
  15. 15. References and Sources Scaling to Very Large Corpora MIT Technology Forbes WikiPedia
  16. 16. Thank You!