Paul Bogdan
University of Southern California, CA·Computer Science
The human brain runs on about 20 watts, yet it can learn from a single experience, adapt to new situations and switch between tasks. Today's AI systems need thousands of training examples, can forget what they learned and use vastly more energy.
- Fields
Computer Science
Neuroscience
Engineering
- Topics
NeuroAI
brain-inspired computing
artificial intelligence
- Grants
- 1 known grant, through Sep 2028
What they work on
- Develop computational techniques that use multimodal neuroimaging and sensing to identify living neurons, non-neuronal cells such as glia, and the connections between them.
- Measure and monitor how living neuronal networks grow, tracking mass distribution, membrane fluidity and functional activity over time with advanced imaging and sensing.
- Build mathematical frameworks to determine the learning mechanisms of living neuronal and glia networks and apply them to design distributed artificial neural networks that learn from limited data.
More about the research
Generative AI is roughly 4 to 10 orders of magnitude less energy-efficient than the brain. This project asks how living networks of neurons learn and make decisions so efficiently. The goal is to carry those principles into brain-inspired AI, known as NeuroAI.
Why it matters
AI training consumes enormous amounts of energy and data. By uncovering how living neuron and glia networks learn efficiently, this work could guide AI designs that learn from few examples and use far less power. Such designs could matter for AI, biotechnology and public safety.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Christopher Kanan
University of Rochester, NY·Computer Science·Early-career
Develop brain-inspired continual learning algorithms that learn online and then run sleep-like consolidation periods, based on complementary learning systems theory, so deep neural networks gain new knowledge without forgetting old knowledge.
Tai Sing Lee
Carnegie Mellon University, PA·Neuroscience
Develop a machine learning framework that formalizes the rapid recurrent plasticity, or fast rewiring, of local circuits in the early visual cortex.
Gheorghi Guzun
San Jose State University, CA·Computer Science·Early-career
Develop adaptable quantization and compression algorithms that take advantage of sparsity, meaning many zero values, in deep neural networks while keeping accuracy loss and overhead on non-sparse data low.
The school
University of Southern California: 47 labs on PI Pages, $1.3B a year on research, 882 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.