Overview / Usage

Knowledge has long been known for its incredible power and influence for those that have it, yet too many of the world's prospective students lack access to it in forms they can easily digest. While teachers at all levels can be very informed and knowledgeable about the subjects they are trying to teach, oftentimes there is a disconnect between ability and effectively conveying material. In less affluent neighborhoods where a quality education is hard to come by, students are left without clear direction and only resources they can find on the internet if it is accessible to them. This project aims to help teachers and students alike overcome this communication barrier that exists in education. By using artificial intelligence to rephrase what a teacher is saying with the help of a personalized model that understands the student's background, one can more effectively convey ideas between these two parties and increase the overall knowledge of the populace. This tool could help students break down internet resources they discover and give them the opportunity to better understand complex material through the help of this network, even if there is no supporting teacher nearby.

Methodology / Approach

As the education problem is in many ways an issue of language, I propose to solve it much like linguists and machine learning researchers approach the problem of machine translation. By first learning a student's own unique dialect of language through their online activity, written work, or perhaps dialogue, a corpus that contains their commonly used words and phrases can be developed. Then a standard encoder decoder model using pretrained word vectors and LSTMs could be used to take what was being said by the teacher and convey it in terms the student can more easily understand, so explanations are no longer the barrier to knowledge, only the true difficulty of the material.

Technologies Used

Intel Optimization for Tenserflow