<img alt="" src="https://secure.tula9mari.com/150530.png" style="display:none;">
GET DEMO

LEVERTON Blog

AI, PropTech and more...

Subscribe

History of Deep Learning Disruption

July 11, 2016 - By LEVERTON Team

In recent years, Deep Learning techniques with Neural Networks have dominated the field of Machine Learning by outperforming several state of the art methods for many tasks involving image recognition, voice recognition and natural language processing. Deep Learning techniques have managed to gather great interest and excitement in the machine learning community.

 

Brief History of Deep Learning:

The first Deep Learning like algorithms actually date back to 1965 by Ivakhnenko et. al and to 1979 by Fukushima. And, the first real application of Neural Networks (specifically, Convolutional Neural Networks) was to classify hand written digits by the work of Yann LeCun in 1989 at Bell labs. This system is also dubbed as LeNet and was used for reading handwritten checks in United States.  Despite the success of Neural Networks, it did not manage to attract much research, because of the limitations in training (i.e., computational power and data access).

 

Golden Era of Artificial Intelligence:

The leap into the golden era started with overcoming the limitations of training Deep Neural Networks. This was possible due to the dedicated efforts made by research scientists like Geoffrey Hinton, Yann LeCun, and Yoshua Bengio, just to name a few, and with better computing power. These scientists showed promising results using Deep Neural Networks, in image recognition by blowing away state of the art methods. Today, the continuous accumulation of Big Data provides another boost to Deep Learning techniques and expands the use cases.

 

Market Players and Text Application:

Companies like Deep Mind (acquired by Google) and Magic Pony as well as the software giants Google, IBM, Facebook and Twitter are exploiting the tremendous capability of the general Deep Learning architectures for solving many different machine learning problems. Nowadays, more and more companies dive into using Deep Learning technology to solve their problems.

Four years ago, the research team around Florian Kuhlmann at LEVERTON realized the power of Deep Learning to solve problems such as Optical Character Recognition (OCR) and Information Extraction. LEVERTON is a pioneer in its field with a dedicated team of Deep Learning / Machine Learning engineers who built disruptive products that are being applied to the legal and real estate industries.