Building AI for Low-Resource Languages: Bezoku's Innovative Approach | Intel

Channel:

Intel

Subscribers:

566,000

Published on October 10, 2024 2:59:49 PM ● Video Link: https://www.youtube.com/watch?v=-u5sk5YweuE

Duration: 0:00

1,432 views

47

Learn about the development of low-resource language models with Ian Gilmour, founder of Euler Digital and creator of Bezoku. Ian shares how Bezoku evolved from AI Forum, addressing the challenge of creating AI models for languages with limited data. He emphasizes the need for low-resource language models and their potential use cases, like polling and surveying, through techniques like reinforcement learning. Gilmour also delves into technical strategies, such as homomorphic encryption and synthetic data, for building secure and efficient AI solutions.

00:00 Introduction – Ian Gilmour, creator of Bezoku
02:18 Dialect-based, low resource language system
03:32 Large language models don’t account for variants and dialects
06:55 Practical example – polling for elections
09:21 Process of ‘annealing’
17:52 Stripping away complexity by thinking right
20:53 Using small language model and efficiencies gained
25:51 Creating efficient workflows with Intel Liftoff for Startups progra

Subscribe now to Intel on YouTube: https://intel.ly/3IX1bN2

About Intel:
Intel, the world leader in silicon innovation, develops technologies, products and initiatives to continually advance how people work and live. Founded in 1968 to build semiconductor memory products, Intel introduced the world's first microprocessor in 1971. This decade, our mission is to create and extend computing technology to connect and enrich the lives of every person on Earth.

Connect with Intel:
Visit Intel WEBSITE: https://intel.ly/Intel
Follow Intel on X: https://intel.ly/Twitter
Follow Intel on INSTAGRAM: https://intel.ly/Instagram
Follow Intel on LINKEDIN: https://intel.ly/LinkedIn
Follow Intel on TIKTOK: https://intel.ly/TikTok

Building AI for Low-Resource Languages: Bezoku's Innovative Approach | Intel