Projection of Orthogonal Tiling from the Retina to the Visual Cortex

Song, Min; Jang, Jaeson; Kim, Gwangsu; Paik, Se-Bum

doi:10.1016/j.celrep.2020.108581

Cited by 12 publications

(9 citation statements)

References 70 publications

(93 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Notably, the current result also suggests a possible scenario through which to understand how viewpoint invariant selectivity can arise in infant animals. From the similarities between the CNN and the biological brain models, i.e., that the fundamentals of both CNNs and sensory cortices are based on the hierarchical feedforward structure and that the process of convolution via weight sharing in CNNs can be approximated by a biological model 73 – 76 of periodic functional maps with hypercolumns in the visual cortex, our result may inspire insight into how innate invariance can arise in infant animals.…”

Section: Resultsmentioning

confidence: 90%

Face detection in untrained deep neural networks

et al. 2021

Self Cite

View full text Add to dashboard Cite

Face-selective neurons are observed in the primate visual pathway and are considered as the basis of face detection in the brain. However, it has been debated as to whether this neuronal selectivity can arise innately or whether it requires training from visual experience. Here, using a hierarchical deep neural network model of the ventral visual stream, we suggest a mechanism in which face-selectivity arises in the complete absence of training. We found that units selective to faces emerge robustly in randomly initialized networks and that these units reproduce many characteristics observed in monkeys. This innate selectivity also enables the untrained network to perform face-detection tasks. Intriguingly, we observed that units selective to various non-face objects can also arise innately in untrained networks. Our results imply that the random feedforward connections in early, untrained deep neural networks may be sufficient for initializing primitive visual selectivity.

show abstract

Section: Resultsmentioning

confidence: 90%

Face detection in untrained deep neural networks

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…The theory predicts that the organization of a visual cortical map should be determined by the retinal position of ON and OFF ganglion cells sampling visual space with the two eyes, as proposed by previous retinal models 8,12,15,41,42 . However, unlike previous retinal models, the theory does not require (and does not rely on) a Moire interference of ON and OFF retinal arrays to generate orientation maps.…”

Section: Resultsmentioning

confidence: 92%

“…Surprisingly, although visual cortical maps are very diverse across species, the mapping of stimulus orientation can be strikingly similar in different mammalian orders such as primates, carnivores and scandentia 1,3,6 . This puzzling balance between map similarity and diversity has been the topic of intensive research over the past decades and inspired a large number of computational models [7][8][9][10][11][12][13][14][15][16] . However, while previous models were very successful at simulating general map patterns for some stimulus dimensions, they were challenged by the use of limited biological constraints and the complexity of simulating inter-related topographies for a large number of stimulus dimensions that include spatial position, eye dominance, light-dark polarity, orientation, spatial resolution, stimulus selectivity, and receptive field structure.…”

Section: Introductionmentioning

confidence: 99%

A theory of cortical map formation in the visual brain

Najafian

Koch

Teh

et al. 2022

Preprint

View full text Add to dashboard Cite

The cerebral cortex receives multiple afferents from the thalamus that segregate by stimulus modality forming cortical maps for each sense. In vision, the primary visual cortex also maps the multiple dimensions of the stimulus in patterns that vary across species for reasons unknown. Here we introduce a general theory of cortical map formation, which proposes that map diversity emerges from variations in sampling density of sensory space across species. In the theory, increasing afferent sampling density enlarges the cortical domains representing the same visual point allowing the segregation of afferents and cortical targets by additional stimulus dimensions. We illustrate the theory with a computational model that accurately replicates the maps of different species through afferent segregation followed by thalamocortical convergence pruned by visual experience. Because thalamocortical pathways use similar mechanisms for axon sorting and pruning, the theory may extend to other sensory areas of the mammalian brain.

show abstract

“…Currently, deep neural network (DNN) models, which have a biologically inspired hierarchical structure, provide an effective approach for investigating functions in the brain (Paik and Ringach, 2011;DiCarlo et al, 2012;Yamins and DiCarlo, 2016;Sailamul et al, 2017;Baek et al, 2020Baek et al, , 2021Jang et al, 2020;Kim et al, 2020Kim et al, , 2021Park et al, 2021;Song et al, 2021). Several studies have reported that DNNs trained to natural images can predict the neural responses of the monkey inferior temporal cortex (IT) (Cadieu et al, 2014;Yamins et al, 2014), known as the area for object recognition.…”

Section: Untrained Alexnetmentioning

confidence: 99%

“…A model study using a biologically inspired deep neural network (DNN) (Krizhevsky et al, 2012; Simonyan and Zisserman, 2015) has been suggested as an effective approach to this problem (Paik and Ringach, 2011; DiCarlo et al, 2012; Yamins and DiCarlo, 2016; Sailamul et al, 2017; Baek et al, 2020, 2021; Jang et al, 2020; Kim et al, 2020, 2021; Park et al, 2021; Song et al, 2021). DNNs consist of a stack of feedforward projections inspired by the hierarchical structure of the visual pathway and can be used as simplified model to investigate various visual functions.…”

Section: Introductionmentioning

confidence: 99%

Invariance of object detection in untrained deep neural networks

Cheon

Baek

Paik

2022

Preprint

Self Cite

View full text Add to dashboard Cite

The ability to perceive visual objects with various types of transformations, such as rotation, translation, and scaling, is crucial for consistent object recognition. In machine learning, invariant object detection for a network is often implemented by augmentation with a massive number of training images, but the mechanism of invariant object detection in biological brains - how invariance arises initially and whether it requires visual experience - remains elusive. Here, using a model neural network of the hierarchical visual pathway of the brain, we show that invariance of object detection can emerge spontaneously in the complete absence of learning. First, we found that units selective to a particular object class arise in randomly initialized networks even before visual training. Intriguingly, these units show robust tuning to images of each object class under a wide range of image transformation types, such as viewpoint rotation. We confirmed that this "innate" invariance of object selectivity enables untrained networks to perform an object-detection task robustly, even with images that have been significantly modulated. Our computational model predicts that invariant object tuning originates from combinations of non-invariant units via random feedforward projections, and we confirmed that the predicted profile of feedforward projections is observed in untrained networks. Our results suggest that invariance of object detection is an innate characteristic that can emerge spontaneously in random feedforward networks.

show abstract

Projection of Orthogonal Tiling from the Retina to the Visual Cortex

Cited by 12 publications

References 70 publications

Face detection in untrained deep neural networks

Face detection in untrained deep neural networks

A theory of cortical map formation in the visual brain

Invariance of object detection in untrained deep neural networks

Contact Info

Product

Resources

About