deletions | additions       diff --git a/abstract.tex b/abstract.tex  index de7b34c..251c12d 100644  --- a/abstract.tex  +++ b/abstract.tex 
...
In the last 2-3 couple of  years a new buzzword approach  has appeared: been used with great success, called  deep learning. In 2012 Microsoft presented a pretty impressive demo Deep learning usually consists of neural networks  with an application that recognized spoken English, translated the text at least 3 layers, which are used  to Chinese and then spoke the translation with the original speakers voice. In build hierarchical representations of data. Some of  the same year, Google developed a system that, from 10 million YouTube thumbnails, learned by itself deep learning methods are unsupervised and they are used  to recognize cats (and 22.000 other categories of objects). In December 2013, Facebook opened learn better features that are than used to solve  a new research lab, led by Yann LeCunn from New York University, one machine learning problem. Other deep learning methods are supervised and can be used directly to solve classification problems. Deep learning systems have been used to obtain the state  of the foremost researchers art  in deep learning. different problems, such as object recognition in images or speech recognition, tested on standard datasets such as MNIST, CIFAR-10, Pascal VOC and TIMIT.