A new computer system developed by Parham Aarabi of the University of Toronto can store and retrieve information strategically, just like our brains.
In addition, the associate professor at the Edward S. Rogers Sr. department of electrical and computer engineering in the Faculty of Applied Science & Engineering has developed an experimental tool that uses the novel algorithm to assist persons with memory loss.
“Most people think of AI as more robot than human,” says Aarabi, whose framework is explored in a paper being presented this week at the IEEE Engineering in Medicine and Biology Society Conference in Glasgow. “I think that needs to change.”
In the past, computers have relied on their users to tell them exactly what information to store. But with the rise of artificial intelligence (AI) techniques such as deep learning and neural nets, there has been a move toward “fuzzier” approaches.
“Ten years ago, computing was all about absolutes,” says Aarabi. “CPUs processed and stored memory data in an exact way to make binary decisions. There was no ambiguity.
“Now we want our computers to make approximate conclusions and guess percentages. We want an image processor to tell us, for example, that there’s a 10 percent chance a picture contains a car and a 40 percent chance that it contains a pedestrian.”
Aarabi has expanded this similar imprecise approach to knowledge storage and retrieval by mimicking many qualities that aid people in determining what to remember—and, as important, what to forget.
According to studies, humans value more recent events above less recent ones. We also emphasise more important memories, and we condense extensive narratives to their core.
“For example, today I remember that I saw my daughter off to school, I made a promise that I’d pay someone back and I promised that I’d read a research paper,” says Aarabi. “But I don’t remember every single second of what I experienced.”
The ability to ignore some information could boost existing machine learning algorithms.
Machine learning algorithms now sift through millions of database entries in search of patterns that will assist them in appropriately associating a given input with a given output. Only after many rounds can the algorithm become accurate enough to deal with new problems that it hasn’t seen before.
If bio-inspired artificial memory enables these algorithms to prioritise the most relevant data, they may be able to produce meaningful results much faster.
The method could also support technologies that process natural language to assist those suffering from memory loss in keeping track of important information.
“Ultimately, it’s geared to people with memory loss,” Aarabi says. “It helps them remember things in a way that’s very human, very soft, without overwhelming them. Most task management aids are too complicated and not useful in these circumstances.”
“I’ve been using it myself,” says Aarabi. “The goal is to put the demo in people’s hands—whether they’re dealing with significant memory degradation or just everyday pressures—and see what feedback we get. The next steps would be to build partnerships in health care to test in a more comprehensive way.”
“These days, AI applications are increasingly found in many human-centered fields,” says Professor Deepa Kundur, chair of the department of electrical and computer engineering. “Professor Aarabi, by researching ways to better integrate AI with these ‘softer’ areas, is looking to ensure that the potential of AI is fully realized in our society.”
Aarabi says that this algorithm is just the beginning.
“Biologically inspired memory may very well take AI a step closer to human-level capabilities.”