It’s always energizing to chat with members of the TDS community, but it’s especially so in one specific way: realizing just how wide-ranging people’s learning habits and styles are. It makes sense, of course. People come to data science from diverse professional, academic, and cultural backgrounds, and what works for some might not do the job for others.
In this week’s Author Spotlight, for example, prolific contributor Khuyen Tran shared her tried-and-true method for digesting complex topics: teaching them. It’s only when Khuyen translates concepts into something others can understand and use themselves that they click for her, too. Read our Q&A with Khuyen, which covers a lot of ground—including the benefits of public writing for attracting potential employers and her thoughts on AI ethics.
For every learn-by-teaching data scientist there must be at least one (probably more?) who loves to dive into the deep end of theory before they go back into the world and find practical uses for it. If that sound like you, we have quite a few treats for you this week. Start with the latest research from Michael Bronstein and his co-authors, which introduces geometric deep learning as a framework with a double goal:
It serves two purposes: first, to provide a common mathematical framework to derive the most successful neural network architectures, and second, give a constructive procedure to build future architectures in a principled way.
Still in the mood for a lofty, speculative read? Gadi Singer wrote an engaging analysis of the promise of cognitive AI, and how “the integration of DL with symbolic reasoning and deep knowledge” will bring about this next phase in artificial intelligence. If even that isn’t enough, you can always turn to Robert Lange, who’s back with another round of his much-anticipated, monthly selection of recommended research papers.
If you’re more of a hybrid learner—someone who likes a gentle, comprehensive introduction to a topic, but enjoys an element of hands-on action, too—we hear you. There are many of you out there! Choose your adventure:
- Follow Carolina Bento as she explains the Poisson distribution (and Poisson point process) with the best example possible: an ice cream shop.
- Lulu Ricketts covers the differences between the two most common parameter estimation methods: maximum likelihood estimation (MLE) and Bayesian estimation. Just as important, she talks about different situations that typically call for one method or the other.
- Prepare for your next round of callbacks with Emma Ding’s ultimate guide to data science business-case interviews.
- If you’d like to find new ways to leverage your data expertise in service of global issues, Thomas Olavson shows how data teams can take concrete steps to reduce carbon emissions in their data centers.
- Tommy Blanchard tackles one of the most crucial—and thorny!—questions a data scientist working in industry might ask themselves: what’s the business value of the model you’re building?
If you have a fresh take on studying data science or have accumulated some memorable experiences (good or bad) along your learning journey, consider sharing them with our community. You’d be surprised how many people might find your story helpful. (If you’re just taking your first steps, give our very own guide a try. It’s free and email-based.)
Thank you, as always, to all of you who read, contribute, and engage with the posts we publish—and special thanks go to all of you who’ve recently shared a TDS story with friends and colleagues. That means so much to us.
Until the next Variable,
Recent additions to our curated topics:
- 4 Reasons Why Correlation Does Not Imply Causation by Ines Lee
- Analyzing Music Taste by Justin Spitzer
- PR Reviews for SQL Code by Marc-Olivier Arsenault
- How to Build a Cloud-Based MLOps Framework in Two Weeks by Lars Kjeldgaard
- Faster, Smoother, Smaller, More Accurate and More Robust Face Alignment Models on CPU by Tino Álvarez
- Clean Code for Data Scientists by Ella Bor
- The Power of Democracy in Feature Selection by Ouaguenouni Mohamed
- Data Analysis: Everything You’ve Ever Wanted to Know about UFO Sightings by Travis Greene
- Artificial Consciousness Is Impossible by David Hsing
Thoughts and Theory
- What’s Lost in JPEG? by Katie He
- How to Make Topic Models Interpretable: 3 New Ideas by Ramya Balakrishnan
- Deduplication of Customer Data Using Fuzzy Scoring by Shreepada Shivananda