Task design for crowdsourcing platform

Abstract

Researchers predicted that AI will eventually outperform humans in many activities in the next ten years [1]. Despite its rapid growth, AI and machine learning are still facing a significant roadblock which is the specific requirement of input data for feature generation. An alternative method which is often proposed to tackle this issue is integrating human into a traditional machine learning framework to utilize the different strengths of humans and machines.

In this project, an online web application will be developed to demonstrate the implementation of the hybrid methodology to generate features for a machine learning framework. The task of generating features will be broken down into microtasks and crowdsourced to many contributors. Similar to a machine learning framework, the platform will aggregate the crowdsourced result to build and train a learning model.