Develop evolution algorithms to search for images that maximally activate visual neurons

What is it about?

What do the neurons in your visual system encode? To answer that question, visual neuroscientists need to find images that make the neuron respond strongly and then summarize these images. Higher order visual neurons usually prefer complicated visual features which could evade our common intuition. Thus, to find these feature, we used an evolutionary algorithm and an image generator to synthesize these images.

Featured Image

Photo by Ion Fet on Unsplash

Photo by Ion Fet on Unsplash

Why is it important?

Using this algorithm, we could increase the maximal activation by 60% in synthetic neuron models and by 40% in real visual neurons. This significantly accelerated the process of finding images that visual neurons are interested in.