Two neurologists have discovered that a computer-based artificial intelligence (AI) can have intelligence at par with humans if they're trained to apply a faster technique of learning. 

Dr Maximilian Riesenhuber, PhD, professor of neuroscience at Georgetown University Medical Center and Dr Joshua Rule, PhD, a postdoctoral scholar at the University of California, Berkeley, have released a paper in the journal Frontiers in Computational Neuroscience that explains how AI can learn new visual concepts quickly. 

“Our model provides a biologically plausible way for artificial neural networks to learn new visual concepts from a small number of examples,” says Riesenhuber. “We can get computers to learn much better from a few examples by leveraging prior learning in a way that we think mirrors what the brain is doing.”

The human ability to learn about images with sparse images still remains unrivalled when compared to AI algorithms. For example, a 3-4-month-old baby can tell a zebra apart from a cat or even a horse, with little training. However, for an AI to be that accurate, it needs to "see" hundreds of examples of the same example to remember it. 

“The computational power of the brain’s hierarchy lies in the potential to simplify learning by leveraging previously learned representations from a databank, as it were, full of concepts about objects,” Riesenhuber says. Therefore, he explains, that they tweaked the designing software so that it identifies an object by computing low-level and intermediate information rather than building relations between complex visual categories. 

Rule explains, “Rather than learn high-level concepts in terms of low-level visual features, our approach explains them in terms of other high-level concepts. It is like saying that a platypus looks a bit like a duck, a beaver and a sea otter.”

The brain's object recognition depends on neural networks related to visual concept learning. It is widely understood that the anterior temporal lobe of the brain contains “abstract” concept representations, beyond shape. These complex neural hierarchies for visual recognition allow humans to learn new tasks and, crucially, leverage prior learning. “By reusing these concepts, you can more easily learn new concepts, new meaning, such as the fact that a zebra is simply a horse of a different stripe,” Riesenhuber says.

As AI continues to emulate human abilities, as we still retain the skill to learn from limited inputs, and can robustly deal with image variations, and comprehend scenes, the scientists say.

“Our findings not only suggest techniques that could help computers learn more quickly and efficiently, but they can also lead to improved neuroscience experiments aimed at understanding how people learn so quickly, which is not yet well understood,” Riesenhuber concludes.