More Like this

1. Visual explanations prioritize functional properties at the expense of visual fidelity

   https://doi.org/10.1016/j.cognition.2023.105414  

   Huey, Holly; Lu, Xuanchen; Walker, Caren M.; Fan, Judith E. (July 2023, Cognition)
2. Non-Visual Interfaces for Visual Learners: Multisensory Learning of Graphic Primitives

   https://doi.org/10.1109/ACCESS.2024.3513712  

   Doore, Stacy A; Brown, Justin R; Imai, Saki; Dimmel, Justin K; Giudice, Nicholas A (December 2024, IEEE Access)
3. Visual Instruction Inversion: Image Editing via Visual Prompting

   Nguyen, T; Li, Y; Ojha, U; Lee, Yong Jae (December 2023, Neural Information Processing Systems (NeurIPS))
4. Development of visual object recognition

   Ayzenberg, V; Behrmann, M (February 2024, Anthropologica Slovaca)

   Object recognition is the process by which humans organize the visual world into meaningful perceptual units. In this Review, we examine the developmental origins and maturation of object recognition by synthesizing research from developmental psychology, cognitive neuroscience and computational modelling. We describe the extent to which infants demonstrate early traces of adult visual competencies within their first year. The rapid development of these competencies is supported by infant-specific biological and experiential constraints, including blurry vision and ‘self-curation’ of object viewpoints that best support learning. We also discuss how the neural mechanisms that support object-recognition abilities in infancy seem to differ from those in adulthood, with less engagement of the ventral visual pathway. We conclude that children’s specific developmental niche shapes early object-recognition abilities 

   more » « less
5. Generative decoding of visual stimuli

   Eleni Miliotou, Panagiotis Kyriakis (August 2023, Proceedings of the 40th International Conference on Machine Learning)

   Andreas Krause, Barbara Engelhardt (Ed.)

   Reconstructing natural images from fMRI recordings is a challenging task of great importance in neuroscience. The current architectures are bottlenecked because they fail to effectively capture the hierarchical processing of visual stimuli that takes place in the human brain. Motivated by that fact, we introduce a novel neural network architecture for the problem of neural decoding. Our architecture uses Hierarchical Variational Autoencoders (HVAEs) to learn meaningful representations of natural images and leverages their latent space hierarchy to learn voxel-to-image mappings. By mapping the early stages of the visual pathway to the first set of latent variables and the higher visual cortex areas to the deeper layers in the latent hierarchy, we are able to construct a latent variable neural decoding model that replicates the hierarchical visual information processing. Our model achieves better reconstructions compared to the state of the art and our ablation study indicates that the hierarchical structure of the latent space is responsible for that performance. 

   more » « less