Machine Learning Resource Tips: Podcasts

I’ll be starting a series of resource recommendations for anyone who’s curious in learning more about ML. A lot of these have probably been mentioned multiple times elsewhere, but hopefully my experience as a novice can help provide some additional context.

This time, I thought I’d take some time to share with you some of my favorite podcasts.

As a bit of an aside, I finally discovered (well, started using) podcasts maybe 6~9 months ago. It’s amazing! Seriously, I haven’t had listened to the radio in my car since December. I can say with confidence I’m much happier trading the Pop Top-40 tracks and commercials for lessons on collaborative filtering, latent Dirichlet allocation, and more. I’ve got a 40~70 minute daily commute, so I have a lot of bandwidth I have to fill.

With that said, here’s the list of what I recommend as useful and engaging Machine Learning content!

1. Talking Machines

More than anything else, I think Talking Machines has been solely responsible in taking my interest in Machine Learning from innate curiosity to a consuming passion of mine. The show is hosted by Katharine Gorman and Neil Lawrence (who took over the role of co-host from Ryan Adams). The format of the show is usually:

  • Introduction and lesson about some concept in Machine Learning, Applied Statistics, AI, etc. 
    One of the reasons I think so highly of this podcast is that they do a great job of achieving a graduate-level depth without being entirely unapproachable.
  • Answering / Responding to questions from listeners
    Next, the duo spend some time answer a different listener question every week. These are generally very well-thought questions; and the hosts always have thoughtful responses.
  • Interviews with ML/AI researchers in academia and industry
    The last half of the podcast is dedicated to speaking with different members of the community and asking them about their path in life, their current research, their opinions on the direction of the field, etc.

The show consistently keeps me engaged and keeps my brain firing on all cylinders, and I only appreciate it more when considering the trouble I’ve had in finding anything that ranks remotely close to this podcast. The hosts are great; Katherine has a genuine passion and great talent for interviewing, providing a lay perspective, and distilling key insights from big ideas. Ryan, the previous host, is incredibly sharp. His delivery, his insights, and the way he frames concepts and builds up to them along the way while being still being succinct and thoughtful: amazing talents. Neil, the current host, has a more free-flowing organic style to his delivery; you really get the feeling you’re chatting with the whole gang over coffee or tea or something.

In addition to the hosts and the content, I feel like this podcast has made a lot of things seem more possible to me. I never considered graduate school until I started listening; now I’m hoarding GRE study prep books like there’s no tomorrow. I’ll be trying to attend the Neural Information Processing Systems conference (NIPS) this year; I have never attended a conference before, and I hadn’t even heard of NIPS before the podcast. Talking Machines opened my eyes to all of the depth and complexity behind the field that is helping build intelligent systems, self-driving cars, and more.

So seriously, if you’re tired of my gushing, stop reading now and download all the episodes of Talking Machines. You’ll thank me.

Also, I should mention: MY LISTENER QUESTION GOT ANSWERED! Click here to listen. Hearing the email I typed being read aloud & discussed by Katherine and Neil was amazing and inspiring, as always.

2. Linear Digressions

Linear Digressions is another good podcast, co-hosted by Katie Malone (who I didn’t realize until now is also an instructor in Udacity’s Machine Learning nanodegree! Go figure!) and Ben Jaffe. The group spends each week exploring a different applied ML topic. Katie does a good job of explaining the relevant concepts in an approachable way, while Ben (whose background isn’t explicitly ML-related) provides a good paraphrasing and reinforcing of the concepts in lay-terms.

While this podcast is good, I don’t feel like this podcast alone will make you an expert in the field. Conceptually, the content and explanations are good, but sometimes the more complicated math is glossed over or summarized. I feel like this a good podcast for someone who wants to get in the head-space of a data scientist, but isn’t necessarily trying to use the podcast to become one themselves.

That said, as a supplementary resource I think the podcast works great. The hosts keep formalities light (their podcast intro music is two guitar chords that last all of 2 seconds) and they have a running gag of starting the episodes with puns relating to the topic. All in all, it’s a light and fun podcast that is probably well suited to people just getting interested in the space.

3. Learning Machines 101

Learning Machines 101 is a podcast made by Dr Richard M Golden which provides a slowly evolving ark of knowledge that builds over each episode. I started the series in early 2017, and lately the show seems to be a mix of re-runs of past episodes and new episodes.

Being taught by a professor, the podcast has more of a ‘lecture’ feel than the previous podcasts. But there is definitely a lot of content to take in, and the website (I’m just learning about this now) has transcripts of the podcast episodes with links to papers, additional concept, and forums. And Dr Golden tries to have fun and make jokes (note the evil Terminator-esque looking robot in the logo) with regards to ML and AI.

Honestly, the only problem I have with this podcast is the presentation of the material. For whatever reason, Dr Golden just comes across as a little robotic; it might be because he’s reading the episodes transcript. That has been my only issue with getting engaged in the material.

4. Data Skeptic

Data Skeptic is another dual-host podcast with a focus on Data Science and Machine Learning. The show is hosted by Kyle Polich and Linda/Lynda/Linh Da (I will go with Linh Da, but I’m not really sure. ), who I believe is Kyle’s wife. The episodes have have a standard format, where Kyle will guide Linh Da (who has no Data Science / ML background) through a conceptual understanding of a given topic. The show also features interviews with people from the field, although it seems like this might be a more recent development.

Kyle keeps things fun and light, and the banter between him and Linh Da can be amusing at times as you’re following along with the lessons. This podcast keeps things at the entry-level though, so it can be a little lacking if you’re looking for some in-depth knowledge. I also have yet to connect how the show actually connects to thinking skeptically about/with data, but that’s a minor detail.

All in all, if you’re looking for something light to half-listen to I’d recommend this podcast.

That’s all I have for this week, and I feel like that’s more than enough to get started. Seriously though, check out Talking Machines right now. If you think there’s anything I missed, feel free to shoot me a comment and let me know!