How Will AI and Machine Learning Enhance Anticipatory Intelligence?

A sneak peek at #HLTCon 2017

“I see…practical anticipatory intelligence for your agency in your future…”

Artificial intelligence, machine learning, and their ilk are ready for national security applications (of which anticipatory intelligence is something of a holy grail), posits Charlie Greenbacker. As VP, Analytics at In-Q-Tel, a strategic investor of the U.S. intelligence and defense communities, Greenbacker has the visibility into this bleeding edge to say that commercial technologies are ready to be adapted to government missions.

However, when so much government intelligence is in closed assets, can a technology that needs to train on the data it will process be accurate and effective?

Greenbacker will elaborate on this position at the Human Language Technology Conference on Tuesday, October 3 in Springfield, VA, where he will be amply supported by an array of 14 other speakers representing key figures from industry, academia, and government, developing and working with anticipatory intelligence applications.

Featured HLTCon Talks

So how do you make the most of closed assets when AI- and machine learning-driven commercial tools can never see or train on your data? Founder & CEO of Luminoso Catherine Havasi introduces “transfer learning” algorithms that apply what they’ve learned in one domain to pick up a new domain (or language) more quickly. This emerging technology is so accurate, it outperforms existing Spanish deep learning models without Spanish training data.

In machine translation, where context and domain knowledge are vital to producing accurate translations, how do you keep up with the topic of the day? Omniscien Technologies CTO and Co-Founder Dion Wiggins introduces unsupervised self-learning and adaptation of machine translation engines, which enables them to rapidly adapt to the context of what is being discussed, and thus produce more relevant and accurate analyses.

Lastly, data. Your output analysis is only as good as your input data. Research Engineer James Fairbanks of the Georgia Tech Research Institute brings new meaning to “data cleansing.” Open source is a treasure trove of data for those forecasting events but it is increasingly “infected” with intentionally fake news that grows like a virus, contaminating even mainstream news outlets. Fairbanks will speak on his work “incorporating automated credibility assessment in an anticipatory analysis pipeline for geopolitical event prediction…to identify and discount problematic and adversarial articles” within the GDELT project (global database of events language and tone).

THE conference for innovations in human language technology

If you are ready for an intimate setting to hear and interact with these presenters and 11 more, as well as take home ways to make anticipatory intelligence a reality for your organization, register for HLTCon. This event is a unique opportunity to network with fellow senior officials, expert linguists, technologists, and product owners across industry and government.

Register now for the Human Language Technology Conference (Tuesday, October 3 at The Waterford at Springfield, VA). Admission is free for government employees.