Published inTowards Data Science·Jun 23, 2021“We Need to Know Our Data”For Pride 2021, a selection of stories on the intersection of data, bias, and LGBTQIA+ communities — There’s a widespread—and understandable—unease about attempts to measure or quantify traits that people have historically associated with marginalized groups. This iffy (or even icky) feeling goes back to the very origins of modern data science, and attempts by 19th-century pseudosciences like phrenology and physiognomy to detect and catalog deviance from…Tds Features4 min readTds Features4 min read
Published inTowards Data Science·May 19, 2021What Does “Safety” in “AI Safety” Actually Mean?Three takes on a complicated question — In a recent episode of the TDS Podcast, guest Rosie Campbell, Head of Safety Critical AI at Partnership on AI (PAI), chatted with Jeremie Harris about the potential risks of publishing cutting-edge AI research. …Ai Safety3 min readAi Safety3 min read
Published inTowards Data Science·Apr 29, 2021What Kind of Knowledge Does Data Science Produce?Taking a step back from the hum of day-to-day model-training and number-crunching, Phoebe Wong asks some very big questions in her 2019 post about the types of knowledge data scientists explore — and generate. Starting off with the long-simmering tension between models that predict real-world outcomes and those that seek…Deep Dives1 min readDeep Dives1 min read
Published inTowards Data Science·Mar 31, 2021Master’s of the House? On Degrees and Credentials in Data ScienceThe debate around advanced degrees is as lively as ever — Common perceptions of data science (and of tech more broadly) suggest it is an egalitarian, merit-based field—one where it doesn’t matter whether you graduated from Stanford or aced a few online courses. …Masters Degree5 min readMasters Degree5 min read
Published inTowards Data Science·Mar 24, 2021Curious about Variational Autoencoders (VAEs)? Start Here.An accessible, illustrated introduction to a complex topic — In recent years, GANs (generative adversarial networks) have been all the rage in the field of deep-learning generative models, leaving VAEs in relative obscurity. But there’s much to gain from a solid footing in variational autoencoders, which tackle similar challenges but use a different architectural foundation. If you were looking…Variational Autoencoder1 min readVariational Autoencoder1 min read
Published inTowards Data Science·Feb 16, 2021How Do Data Scientists Use Twitter? Let Us Count the WaysDiscover some of the best Twitter data analyses from the TDS archive. — Anecdotally, it would seem that you can be on Twitter, or you can attempt to find joy, health, and balance in your life, but not both. The writers, journalists, and niche food-opinion-havers in my timeline form a fairly diverse lineup. What unites us is a strong ambivalence—yes, I’m mincing words…Data Science4 min readData Science4 min read
Feb 13, 2021Editing, ChieflyOr is it “chefly” or “cheffily” or “chuffingly” — According to some North American etiquette authorities, it’s acceptable to send newlyweds a gift up to an entire year after the wedding. I remember being floored by this tidbit when I first learned about it many, many years ago, as I grew up in a culture where it would be…Towards Data Science3 min readTowards Data Science3 min read
