Leaving no one behind: Make equal access to social good possible with deep learning

Description

Nowadays, people seem to be more worried about AI replacing jobs than excited about the new opportunities AI creates. However, many of these debates are centered around the “normal” workforce. Over 400M people worldwide have some sort of speech or hearing disorder that precludes them from participating in the job market. Disability should not be a disadvantage, and everyone who wants to work with others should have equal access to do so.

Goodman Gu offers an overview of Stride4All, an initiative created at Atlassian using AI and machine learning technologies to open work up for those with speech and hearing impairments and empower them for teamwork, and showcases a prototype that uses deep learning and computer vision technologies for gesture recognition of American Sign Language. He will then explore other use cases where AI can be a major force for social good. In particular, Goodman explains how to use a Keras frontend, a TensorFlow backend, and convolutional neural networks to build a multiclass classifier for sign language alphabets and convert signs into text and lifelike speech. An early adopter of Amazon Sagemaker, Goodman also shares his experience using this fully managed end-to-end machine learning pipeline, revealing how feature engineering, hyperparameter tuning, and model deployment can be simplified and streamlined to make data scientists’ and machine learning engineers’ lives easier and more productive.