10 October 2019: Estimating earthquake risk, and difficulties for deep-learning - podcast episode cover

10 October 2019: Estimating earthquake risk, and difficulties for deep-learning

Oct 09, 201924 min
--:--
--:--
Download Metacast podcast app
Listen to this episode in Metacast mobile app
Don't just listen to podcasts. Learn from them with transcripts, summaries, and chapters for every episode. Skim, search, and bookmark insights. Learn more

Episode description

This week, a method for predicting follow-up earthquakes, and the issues with deep learning systems in AI.


In this episode:


00:47 Which is the big quake?

A new technique could allow seismologists to better predict if a larger earthquake will follow an initial tremor. 

Research Article: Real-time discrimination of earthquake foreshocks and aftershocksNews and Views: Predicting if the worst earthquake has passed


07:46 Research Highlights

Vampire bats transmitting rabies in Costa Rica, and why are some octopuses warty? 

Research Article: Streicker et al.Research Article: Voight et al.


10:03 Problems for pattern-recognition

Deep-learning allows AIs to better understand the world, but the technique is not without its issues. 

News Feature: Why deep-learning AIs are so easy to fool


16:31 News Chat

We roundup the 2019 Nobel Prizes for science. 

News: Biologists who decoded how cells sense oxygen win medicine NobelNews: Physics Nobel goes to exoplanet and cosmology pioneersNews: Chemistry Nobel honours world-changing batteries

Hosted on Acast. See acast.com/privacy for more information.

For the best experience, listen in Metacast app for iOS or Android