RTB House, a global company that provides state-of-the-art retargeting technology, introduces an innovative approach to building neural network architecture. RTB House created a method to improve the way machines predict conversion value. This results in advertising campaigns bringing in ROIs. The novel concept is presented during the 2017 International Joint Conference on Neural Networks in the USA, Alaska.
Neural Networks are a biologically-inspired programming model which enable a computer to learn from observational data – similar to the way a human brain learns from perception. It is one of the most powerful tools when it comes to solving digital classification problems, used widely in industries for recognition in videos, image, speech, language, DNA or even stock markets and the weather.
Unfortunately for the marketing industry, a neural network is limited in its ability to predict continuous values, for example, figures like order value or daily revenues. But RTB House developed an innovative method that makes it possible to derive more precise and reliable estimations of such value. It can be used to enhance any neural network trained to solve value estimation tasks.
Bartek Romański, Chief Technology Officer RTB House remarks, “Neural networks (especially deep learning architectures) have become the new standard in making sense of the digital world and leveraging the enormous opportunities within data. AI has forever changed the way we do digital advertising. Google and Facebook have been training brain-inspired neural networks to better represent the real world, and classify, cluster and predict outcomes in data. Today, deep learning is finding its way into uses across every industry, from healthcare, to e-commerce, self-driving cars and even art. We are extremely proud to be a contributor to the field of deep learning in advertising.”
Konrad Żołna, Research Scientist at RTB House, explains that how the model works to find optimized values in conversions: “Our method extends the training phase of the conversion value model with carefully constructed additional targets, making the final model’s predictions more robust and precise. In practice, it means that our self-learning algorithms are able to ultra-precisely identify buyers with the largest potential basket value, and then display a personalized message encourage them to finalize the transaction.”
The general idea and results of RTB House deep learning approach will be presented at the 2017 International Joint Conference on Neural Networks (IJCNN 2017). The event will be held at the William A. Egan Civic and Convention Center in Anchorage, Alaska, USA, May 14–19, 2017 and is the premiere international meeting for researchers and other professionals in neural networks and related areas.
It is the third conference, after the 33rd International Conference on Machine Learning (ICML 2016) in New York City and the 31st AAAI Conference on Artificial Intelligence (AAAI 2017) in San Francisco, where RTB House findings in the field of artificial intelligence are presented.
More about the conference: http://www.ijcnn.org/