I gave a short presentation at the 2019 KNOW Conference. See https://vegas.knowconf.com/. The KNOW Conference is mostly about digital identity (think password authentication and resource authorization), but that topic has many closely related areas such as fraud detection.

The event ran from March 24-27 at the Aria Hotel in Las Vegas. I estimate there were approximately 1500 attendees, speakers, and exhibitors at the event. Most of the attendees I talked to worked on the business/management side of things (about two-thirds of attendees) rather the technical side of things (about one-third).

Left: I introduce myself. Center: I talk about deep autoencoders. Right: I explain LSTMs by using a challenge with four students from UC Berkeley.

Attendees and speakers came from a wide range of companies and various government agencies and departments. For example: Google, Facebook, Microsoft, IBM, Oracle, MasterCard, Netflix.

I spoke about two recent developments in machine learning that have a connection to digital identity. First I described some recent work that uses a deep neural autoencoder to transform mixed numeric and non-numeric data into strictly numeric data which can then be analyzed using interpretable techniques such as k-NN classification or k-means clustering. Then I discussed recent work in NLP including the shift from LSTMs to bidirectional LSTMs to Attention LSTMs to Transformer LSTMs to the BERT algorithm.

My main takeaway from the KNOW conference is that the field of digital identity systems is highly fragmented with dozens of players but no clear industry leader. Because digital identity in an enterprise is intertwined with just about every system in an enterprise, integration is the key. My guess is that there will be rapid and increasing consolidation through mergers and acquisitions.

The 2019 KNOW Expo and main lecture room.

My secondary takeaway is that the use of sophisticated machine learning techniques in digital identity is basically non-existent. There’s a tremendous opportunity for companies to use ML to gain a competitive advantage and market differentiation.

At most technical conferences, I get the most value from attending session talks and chatting with other attendees. The session talks at KNOW were good, but I got more value from exploring the conference Expo. There were about 100 or so companies represented. I particularly enjoyed talking to the representatives from a company called Okta.

So, attending the KNOW conference was definitely a good use of my time. I’m confident I represented my company well. And I gained valuable insight that are directly applicable to my work. And, as a side effect, I renewed my energy and enthusiasm for my work.