Revolutionizing deep learning: powering models with limited data

Revolutionizing deep learning: getting more out of models with limited data

On May 30, 2024 we gathered at the Catholic University of Pereira for Revolutionizing deep learning: powering models with limited data.

Talks:

Talk 1:

Speaker: Leiver Campeon

Role: Machine Learning Engineer and Tech Lead at Expert Intelligence

Continual Learning and Catastrophic Forgetting in Modern AI Systems

Once a deep learning model is deployed, the challenge becomes learning new tasks continuously without losing what it already knows. This talk explores strategies for getting past catastrophic forgetting so models keep evolving effectively.

Talk 2:

Speaker: Sebastián Franco Gómez

Role: ML Engineer at Expert Intelligence

Optimizing deep learning models when data is not abundant

Deep learning has transformed machine learning and AI over the past decade, and part of why it worked so well is the explosion of available data — the more data a model can ingest, the better it behaves. In the real world, though, we run into problems with specific and scarce data. How do we use deep learning’s superpowers while dodging its appetite for enormous datasets?

Original post on Meetup.com

---

Sources

• Original event page: Meetup.com
• Content migrated from the production archive (main branch) for date/content parity.