Networking, recruiting, and how DNN demands will drive our future growth
This past week, I had the tremendous opportunity to travel to Melbourne, Australia for the 56th annual meeting of the Association of Computational Linguistics (ACL). This is the single largest gathering for the NLP and ML communities in the world, which meant a busy week for Daniela and I. We made great new contacts, attended a fascinating keynote address on the development of Deep Neural Networks, and met some brilliant talent spread across enterprise and the academy that will drive AI into the future.
With three years of steep growth under our belt, this year’s ACL was a chance for us to define our leadership role in driving this industry toward that future. Our booth had a steady stream of visitors, many representing some of the biggest names in AI on the planet. We weren’t surprised. It’s been a bit of pattern for us this year as we’ve formed a wide array partnerships and rapidly expanded our client roster in 2018.
More businesses are realizing the value of sourcing high-quality, scalable training data (and the high costs of settling for anything less). We’re uniquely positioned to provide exactly that, which means we’ve been having some fascinating conversations with some incredible companies lately. As always, stay tuned for more on this soon!
For now, as we return to our bases in Seattle and Lisbon we’ll certainly be discussing Anton van den Hegel’s invited talk, “Deep Neural Networks and what they’re not very good at.” As Daniela’s been saying for years, Deep Neural Networks (DNN’s) have long been the “holy grail” of machine-learning development, as they offer the clearest path to self-learning AI that can truly improve on-the-fly.
DNN’s are already responsible for major breakthroughs in fraud detection and manufacturing optimization. But, for all they’ve accomplished, DNNs still fall short in tasks that require contextual interpretation.
Take the image below:
If I asked you, “Who’s wearing the sunglasses?” You’d say “the pineapple” without second thought. DNN models? Not so much. They’re not capable of integrating the linguistic, visual, and contextual understandings necessary to come up with the correct response. At least… not yet.
As always, the barriers to these capabilities are falling. The brilliant scientists and engineers we met at ACL are…well… brilliant after all.
But, for DNN’s to obtain these contextual reasoning and interpretation capabilities, they’ll need incredibly precise, accurate and complexly structured data. DefinedCrowd is the only firm that can deliver the kind of job-tailored high-quality data necessary to train and test these kinds of models. It’s an exciting time to be here!
Which brings me to my final point. As more demand for our data grows, our team will too. Actually, it already is. Right now, we have 23 open positions across our offices in Seattle, Lisbon, Porto, and Tokyo. We met a lot of great talent at ACL. To all of you who stopped by our booth, it was a pleasure. If you passed on a resume, you’ll hear from us soon.
And if we missed you? There’s still time! Go to our careers page to see how to come build amazing things with us.