Saturday 17:15–17:50 in Auditorium

Computer-aided design of advertisement product compositions

Kay Hoogland

Audience level:
Novice

Description

Is it possible to apply data science to a creative task? In my talk, I will cover my research to the (semi-)automated creation of advertisement product compositions. The resulting application co-creates compositions with graphic designers by making use of an evolutionary algorithm with user input. The algorithm is built in DEAP and runs on AWS.

Abstract

The growing e-commerce market is closely associated with an increasing demand for online advertisements. Graphic designers are scarce and creative agencies are looking for solutions to keep up with the demand for advertisements by retailers. Data Science and Machine Learning have been applied successfully in the marketing domain already, but rarely on the creation part of online advertisements.

In my talk, I will describe the research steps that led to the final data science solution for this problem. The research resulted in an application that co-creates product compositions with designers for online advertisements. Opinions of designers are incorporated into the model using a man-in-the-middle evolutionary algorithm. The evolutionary algorithm of the application is written in Python’s DEAP package and runs on AWS.

Subscribe to Receive PyData Updates

Subscribe