Can We Improve Machine Vision Using Insights From Neuroscience?
9 March 2023 – 13:00 GMT
By: SP Arun, Centre for Neuroscience, Indian Institute of Science, Bangalore, India
Deep neural networks have revolutionized computer vision with their impressive performance on vision tasks. Recently their object representations have been found to match well to the visual areas of the brain. Yet their performance is still worse than humans, and it has been challenging to derive insight into why deep networks work or how they can be improved. In our lab we have been comparing object representations in brains and deep networks with the aim of understanding how we see and to make machine see better. We have shown that systematic biases in deep networks can be identified by comparing with brain representations, and that fixing these biases can improve performance. We have also been testing deep networks for the presence or absence of a variety of classic perceptual phenomena. Taken together these results suggest that accumulated wisdom from vision neuroscience can help us understand and improve deep neural networks.
Bio: SP Arun started out as an electrical engineer, read too much science fiction for his own good and turned into a neuroscientist. He is fascinated by how the brain transforms sensation into perception, particularly for vision. His lab at the Centre for Neuroscience, Indian Institute of Science, studies how the brain solves vision by investigating perception and brain activity in humans, by investigating behavior and neural activity in monkeys and by comparing vision in brains and machine algorithms.
For more details visit the homepage of his research group, the Vision Lab @ IISc. https://sites.google.com/site/visionlabiisc/