You are here:  Home  >  Data Blogs | Information From Enterprise Leaders  >  Current Article

Biologically Inspired Intelligence Is Foundation Of AI-One's Platform

By   /  August 27, 2012  /  No Comments

Biologically inspired intelligence. That impressive catchphrase is used by ai-one – but what does it mean?

“The idea is that our technology works as the human brain does,” says Olin Hyde, vp of business development at the vendor. “And the idea of the business is to make it easy for developers to build intelligent apps. We want them to embed AI (artificial intelligence) into every device.” The hope is to empower developers to build applications to extract intelligence from content, to analyze and discover meaningful patterns, by making AI easy.

The issue with AI, he says, is that the way it’s traditionally been approached doesn’t correspond to how the human brain works as much as it is “based on some fancy math that was discovered in the 18th and 19th centuries.” But people were intelligent long before then, learning contextually from patterns, gaining knowledge dynamically through a series of associations, building understanding autonomically without models. Guided by that foundational idea, ai-one looks to what it calls the holo (whole) semantic (meaning) data space (the dynamic area into which information is fed).

The holosemantic data space (HSDS) that provides complex AI with reasoning and learning capability operates offers answers to questions you may not know you want to ask, the company says. Or, as Hyde explains, think of it as providing answers when you don’t know just what you are looking for, thanks to relating pieces of information together, dynamically, without any prior knowledge to inform it: Say a government agency wants to understand international terrorist communications over Twitter. Obviously, a terrorist group is not going to come right out and say it will plant an explosive at a particular coordinate.

“They will speak in code. So you know the problem, but not what question to ask to solve the problem. So you feed Twitter data into the HSDS and let us build all those relationships and we tell you what questions to ask,” he says. And it’s the anomalies that are outside the normal pattern – the talk about baseball coming from someone in Afghanistan, for example – that raise red flags. “Why would that person be talking about American sports on Twitter? It’s a targeting term,” Hyde says.

Topic-Mapper For Text

ai-one wants the capabilities it’s developed to serve a range of purposes, which is why its focus is on continuing to enhance and license its core technology for others to use. Its primary offering that is available now is its Topic-Mapper machine-learning API, which is AI-optimized for the unique grammar of text and unstructured data. The first version was a 32-bit API but the new Version 2 migrates to the cloud as a 64-bit API. “It has a theoretical capacity of 20,000 times greater than the total corpus of human knowledge,” says Hyde.

“Our technology generates a lightweight ontology that is machine-generated, not human-curated,” Hyde says. ‘It is self-organizing,” operating at the byte level and agnostic as to data, including the language it is in; recording every unique byte pattern once and detecting how it relates to other byte patterns; not being reliant upon rules or training sets; and able to unlearn mistakes and re-teach itself correctly with enough information.

For instance, once it has learned “caterpillar,” it might classify butterfly as a caterpillar until it “sees” enough butterflies to realize that a butterfly is what a caterpillar becomes. “Think of it as an associative network,” he says. Following current theory about how the brain learns – that its wiring changes when stimulated – “the wiring inside the holosemantic data space changes any time you stimulate it, and it becomes more and more refined.”

It can work hand-in-hand with a full-fledged ontology that has benefited from extensive human curation, he notes. “The things you put into an ontology are a canonization of many years of research and knowledge, and that’s why that can make our technology more powerful,” though it’s not a requirement. There are two situations in knowledge management, he says – one where you know what you are looking for, where an ontology can serve to inform the exploration, and, as mentioned above, one where you don’t know what to ask.

Ai-one has developed a few prototype implementations to give potential customers an idea of what they can accomplish. The TopicMapper-enabled ai-Fingerprint, for example, can find information by comparing the differences, similarities and intersections of information on multiple websites, using artificial intelligence to identify clusters and visually show how each cluster relates to another.

Topic-Mapper could have some company soon, as the company also has in store AI APIs optimized for computer vision, called UltraMatch, and signal processing, called Graphalizer. In fact, it has among its specialized applications ASTIS, which is based on its computer vision API, for the analysis and matching of shoe print tracks in forensic pattern analysis.


About the author

Jennifer Zaino is a New York-based freelance writer specializing in business and technology journalism. She has been an executive editor at leading technology publications, including InformationWeek, where she spearheaded an award-winning news section, and Network Computing, where she helped develop online content strategies including review exclusives and analyst reports. Her freelance credentials include being a regular contributor of original content to The Semantic Web Blog; acting as a contributing writer to RFID Journal; and serving as executive editor at the Smart Architect Smart Enterprise Exchange group. Her work also has appeared in publications and on web sites including EdTech (K-12 and Higher Ed), Ingram Micro Channel Advisor, The CMO Site, and Federal Computer Week.

You might also like...

Three Traditional Storytelling Techniques That Add Value to Data and Analytics

Read More →