Fixing the problems of deep neural networks will require better training data and learning algorithms
This work tackles the problem of aligning DNNs with biological vision for researchers in computational neuroscience and AI, but it is incremental as it builds on existing critiques.
The paper addresses the issue that deep neural networks (DNNs) achieve human-level accuracy using strategies different from biological vision, and it proposes methods to build DNNs that better model biological vision as they scale up.
Bowers and colleagues argue that DNNs are poor models of biological vision because they often learn to rival human accuracy by relying on strategies that differ markedly from those of humans. We show that this problem is worsening as DNNs are becoming larger-scale and increasingly more accurate, and prescribe methods for building DNNs that can reliably model biological vision.