Bayesian Program Learning outperforms human capabilities in vision-related tasks

A journal Science-published research paper has unveiled of a new type of ‘one shot’ machine learning whose computer vision program has beaten a group of humans in identifying handwritten characters. The advancements in machine-vision systems are vital as they are becoming common in many aspects of life including car-safety systems, video game controls and factory robots. Researchers at the Massachusetts Institute of Technology, New York University and the University of Toronto said that the computer vision program is known as Bayesian Program Learning is quick in learning new characters in many languages and simplifying things that it has learnt. There is a difference between the current machine learning technologies known as deep neural networks and Bayesian Program Learning. In the former ones, neural networks are trained to recognize human speech…


Link to Full Article: Bayesian Program Learning outperforms human capabilities in vision-related tasks