Tweaking AI software to function like a human brain improves computer's learning ability

January 12, 2021

WASHINGTON - Computer-based artificial intelligence can function more like human intelligence when programmed to use a much faster technique for learning new objects, say two neuroscientists who designed such a model that was designed to mirror human visual learning.

In the journal Frontiers in Computational Neuroscience, Maximilian Riesenhuber, PhD, professor of neuroscience, at Georgetown University Medical Center, and Joshua Rule, PhD, a postdoctoral scholar at UC Berkeley, explain how the new approach vastly improves the ability of AI software to quickly learn new visual concepts.

"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 few examples by leveraging prior learning in a way that we think mirrors what the brain is doing."

Humans can quickly and accurately learn new visual concepts from sparse data ¬- sometimes just a single example. Even three- to four-month-old babies can easily learn to recognize zebras and distinguish them from cats, horses, and giraffes. But computers typically need to "see" many examples of the same object to know what it is, Riesenhuber explains.

The big change needed was in designing software to identify relationships between entire visual categories, instead of trying the more standard approach of identifying an object using only low-level and intermediate information, such as shape and color, Riesenhuber says.

"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," he says.

Riesenhuber and Rule found that artificial neural networks, which represent objects in terms of previously learned concepts, learned new visual concepts significantly faster.

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 architecture underlying human visual concept learning builds on the neural networks involved in object recognition. The anterior temporal lobe of the brain is thought to contain "abstract" concept representations that go 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.

Despite advances in AI, the human visual system is still the gold standard in terms of ability to generalize from few examples, 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, they can also lead to improved neuroscience experiments aimed at understanding how people learn so quickly, which is not yet well understood," Riesenhuber concludes.
-end-
This work was supported in part by Lawrence Livermore National Laboratory and by the National Science Foundation (1026934 and 1232530) Graduate Research Fellowship Grants.

About Georgetown University Medical Center

As a top academic health and science center, Georgetown University Medical Center provides, in a synergistic fashion, excellence in education -- training physicians, nurses and other health care professionals, as well as biomedical scientists -- and cutting-edge interdisciplinary research collaboration, enhancing our basic science and translational biomedical research capacity in order to improve human health. Patient care and clinical research is conducted with our clinical partner, MedStar Health. GUMC's mission is carried out with a strong emphasis on social justice and a dedication to the Catholic, Jesuit principle of cura personalis -- or "care of the whole person." GUMC comprises the School of Medicine, the School of Nursing & Health Studies, Biomedical Graduate Education, and Georgetown Lombardi Comprehensive Cancer Center. Designated by the Carnegie Foundation as a "very high research activity university," Georgetown is home to a Clinical and Translational Science Award from the National Institutes of Health, and a Comprehensive Cancer Center designation from the National Cancer Institute.
Connect with GUMC on Facebook (Facebook.com/GUMCUpdate) and on Twitter (@gumedcenter).

Georgetown University Medical Center

Related Neuroscience Articles from Brightsurf:

Researchers rebuild the bridge between neuroscience and artificial intelligence
In an article in the journal Scientific Reports, researchers reveal that they have successfully rebuilt the bridge between experimental neuroscience and advanced artificial intelligence learning algorithms.

The evolution of neuroscience as a research
When the first issue of the JDR was published, the field of neuroscience did not exist but over subsequent decades neuroscience has emerged as a scientific field that has particular relevance to dentistry.

Diabetes-Alzheimer's link explored at Neuroscience 2019
Surprising links exist between diabetes and Alzheimer's disease, and researchers are beginning to unpack the pathology that connects the two.

Organoid research revealed at Neuroscience 2019
Mini-brains, also called organoids, may offer breakthroughs in clinical research by allowing scientists to study human brain cells without a human subject.

The neuroscience of autism: New clues for how condition begins
UNC School of Medicine scientists found that a gene mutation linked to autism normally works to organize the scaffolding of brain cells called radial progenitors necessary for the orderly formation of the brain.

Harnessing reliability for neuroscience research
Neuroscientists are amassing the large-scale datasets needed to study individual differences and identify biomarkers.

Blue Brain solves a century-old neuroscience problem
In a front-cover paper published in Cerebral Cortex, EPFL's Blue Brain Project, a Swiss Brain Research Initiative, explains how the shapes of neurons can be classified using mathematical methods from the field of algebraic topology.

Characterizing pig hippocampus could improve translational neuroscience
Researchers have taken further steps toward developing a superior animal model of neurological conditions such as traumatic brain injury and epilepsy, according to a study of miniature pigs published in eNeuro.

The neuroscience of human vocal pitch
Among primates, humans are uniquely able to consciously control the pitch of their voices, making it possible to hit high notes in singing or stress a word in a sentence to convey meaning.

Study tackles neuroscience claims to have disproved 'free will'
For several decades, some researchers have argued that neuroscience studies prove human actions are driven by external stimuli -- that the brain is reactive and free will is an illusion.

Read More: Neuroscience News and Neuroscience Current Events
Brightsurf.com is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com.