20 min listen
AI Dialogues: exploring the significance of training AI models
AI Dialogues: exploring the significance of training AI models
ratings:
Length:
20 minutes
Released:
Feb 15, 2024
Format:
Podcast episode
Description
It's AI bingo at the moment! Everyone's talking about it - but did you know that the quality and quantity of training datasets are crucial to the accuracy and effectiveness of machine learning models? The more diverse and representative the data, the better the AI model can perform in terms of nuance. Did you ever wonder how to ensure that AI discerns what Americans would wear to a wedding? And what people would wear in Asia, Europe or Africa, taking into account the specificities of each country?
In this enlightening AI dialogue episode of our Globally Speaking podcast, Melanie Peterson - Program Director for Train AI at RWS - joins Vasagi Kothandapani to discuss how she tackled one of her biggest AI challenges: a rating and labelling data project for western-style womenswear using resources in the APAC region only, where the raters were 80% male.
Melanie unpacks the intricacies of her mission, shedding light on how she explains what she does to her family and friends, why a rocket scientist or a chef might have been involved in one of Train AI's latest data training projects, and what exactly we mean by data collection, data creation, data annotation and sentiment analysis.
Get ready for a captivating exploration into the heart of AI.
In this enlightening AI dialogue episode of our Globally Speaking podcast, Melanie Peterson - Program Director for Train AI at RWS - joins Vasagi Kothandapani to discuss how she tackled one of her biggest AI challenges: a rating and labelling data project for western-style womenswear using resources in the APAC region only, where the raters were 80% male.
Melanie unpacks the intricacies of her mission, shedding light on how she explains what she does to her family and friends, why a rocket scientist or a chef might have been involved in one of Train AI's latest data training projects, and what exactly we mean by data collection, data creation, data annotation and sentiment analysis.
Get ready for a captivating exploration into the heart of AI.
Released:
Feb 15, 2024
Format:
Podcast episode
Titles in the series (100)
Why is Global Branding So Hard? And So Worth It?: Branding is one of the most common concepts in marketing, but it's also one of the hardest to define. And when you consider branding from a global... by Globally Speaking