Emma Alexander

ealexander@northwestern.edu

Google Scholar

CV

I'm interested in low-level, physics-based, bio-inspired artificial vision.

I develop bio-inspired visual sensing techniques. Combining the forward engineering approach of computer vision with the reverse engineering perspective of vision science, I extract new principles of visual sensing and apply them to produce novel devices and a deeper understanding of vision questions across a wide span of imaging subfields. This approach draws on and advances principles in optics, algorithms, and neuroscience, while developing applications in imaging modalities from photography to microscopy.

By explaining vision systems through their underlying mechanisms rather than relying on statistical regularities in datasets that are never made explicit, this approach allows analytical reasoning about when and why systems will succeed or fail. I am particularly inspired by animals with small nervous systems, which are remarkably resource-aware in contrast to mainstream, data-driven computer vision systems. These systems often far exceed the state of the art in computer vision, not only in terms of raw performance, but also in weight, size, energy consumption, computational efficiency, and training data. One way small animals manage this is through task-specific sensing: in place of general-purpose hardware, evolution often develops specialized optics and brain structures that cooperate to prioritize vital tasks. This computational sensing approach leads to an enormous diversity in sensor design across natural vision systems, each driven by the challenges its specific ecological niche, providing ample inspiration for technologists.