Gentler Podcasts (still quite useful, but are approachable to those who don't read ML whitepapers frequently):
Talking Machines is definitely my favorite ML related podcast, the audio quality could be a bit better, but the hosts, topics, and interviews are really good, and I've used actually ended up actually using quite a few things I first heard about on the podcast.
Totally agree. I really like the format of Talking Machines, which is :
- deep dive into nuanced topic like Latent Dirichlet allocation
- answering a reader question
- in depth interview with a luminary in the field (good mixture between "celebs" like Andrew Ng and lesser known, but still brilliant, practioners)